Kling Crosses 12M, China’s Aging Tech Problem, Thinking Machine Turmoil | Rich Greenfield, Jan Sramek, George Lewin, Ara Kharazian

You're watching TVPN.

0:00

Today is Wednesday, January 21, 2025, 2026. 2026. Throw and smoke. It's a new year. We're live for the TBB and UltraDome, the temple of technology, the fortress of finance, the capital of capital. And now that the smoke has cleared, I can tell you about ramp.com, time is money save, both easy to use, corporate cards, bill pay, accounting and a whole lot more all in one place. And big news, ramp. We have AR Khorazian from ramp live in the Ultra Dome at 2pm today. Let's show everyone the linear lineup because we got a lot of great stuff for you. We got Rich Greenfield coming on at 11:30.

0:02

Many people are saying he has one of the best names of all time.

0:41

He does, he does.

0:44

If you enjoy business, if you're a business enthusiast, lots of people like Greenfield projects.

0:45

You start fresh. Greenfield projects are where it's at. Yan Srimak, founder and CEO of California Forever, is in the ultradome at 1. Then we have George from Testudo at 140 and Ara closing out 150 in person. It's going to be a fun show. Linear, of course, is the ultimate product, the system for modern software development. 70% of enterprise workspaces on Linear are using agents today and you should be too. So cling AI. Cling AI, the video generation video model has been on a tear. There's an article in the Wall Street Journal showing some pretty staggering Numbers. They hit 12 million monthly active users and they generated more than 20 million in revenue just last year. I wanted to dig in and understand where this came from. The history of Kuaishou, the Chinese company behind it. There's a new farmer filling up the trough for everyone.

0:51

Really interesting that they're so neck and neck with Higgs Field. Yeah, Higgs Field last week comes out, announces they got to 200 million.

1:51

Yeah, yeah.

1:59

This story drops. Yeah, that cling seems to be just slightly ahead of them, but really neck and neck and seemingly competing over the same opportunity.

2:00

Yeah, I mean on the enterprise model, API level, even the prosumer level, I think 30% of their revenue is coming from API and 70% is for prosumer. So basically you have someone in a marketing role or they're just a fun individual creator and they want some AI video products, some generative video and the.

2:09

Silos at Davos on Kling.

2:29

It's crazy.

2:31

It's insane.

2:32

It's actually crazy.

2:33

It's insane.

2:34

They refuse to ask any hard questions. You got Demis on stage. You got Dario on stage and you're not asking them about Kling and the absolute terror that Kuashiu's been on. Actually, I mean, Dario did address the Chinese question. He said that they don't lose. They're not losing enterprise deals to Chinese companies right now. They're. They're going up against OpenAI and DeepMind. Yeah, there is a. There is an interesting Chinese export control issue that I think we should get into at some point with, with regard to this story. But first, let's tell you about Restream 1 livestream. 30 plus destinations. If you want a multi stream, go to restream.com so cling has hit 12 million MAUs, 20 million revenue in the last month. Now they seem to be on the and to the right curve. It's a pretty massive ramp. The product launched 18 months ago, but CLING is not its own startup. It is a new project from Kuaishou Technology. This is a Chinese company and the founders. I couldn't find a single product.

2:35

John, I'm going to blow your mind right now, please. Guess who worked at Kuaishou, who's been on the show? Okay, I share a name with him.

3:36

Wait, Connor Hayes. No, no.

3:47

Jordan Schneider.

3:50

Jordan Schneider worked at Kai Show.

3:51

Yes.

3:52

No way.

3:52

Yes.

3:53

He just responded to the daily newsletter.

3:53

Does he want to hop in and.

3:56

Said lol, I worked at Kwai Show. What a comeback. Jordan. Jordan, come get on the show.

3:57

You're welcome. Come hang out if you want to.

4:02

I'll send him a note.

4:05

It'd be great to hear more of more of the story from him directly, but he might be the only person that we can get to go on the record about Kuaishou. Because the founders, I don't think they've ever done a podcast about appearance. They haven't been certainly on the victory lap podcast American circuit, like many founders that get a company to this scale. But there are some interesting links to other American companies. So Kuaishow is pretty old, especially for the AI boom. It launched in 2011 as a mobile app for creating and sharing gifts. This was before vine and before musical lyrics. They were sort of like a precursor to all those just by a year or two.

4:06

So you could almost say that Jordan Schneider made his money in GIFs.

4:48

I think so. I think he created TikTok. Basically we should put him in charge. We should get him in the CEO seat, put him in charge. We're bringing TikTok back to America and Jordan should be running it for sure. So there's some similar threads. So Dom Hoffman, the founder of vine, he worked at Yahoo before founding Vine. And Alex zhu at Musical Ly worked at eBay and Microsoft and I think SAP as well. And so both of those companies, they had like big tech experience. Then they went and founded these companies. And the founders of Kuaishou Suhua and Cheng Yi Xiao also did the same. One of them worked at Google, one of them worked at Hewlett Packard. And then they jumped into the social media boom that was going on in the early aughts, or I guess the late aughts, early 2010s. Is there a word for the period between 2010 and the teens, the 20s, the tens. The tens, I don't know. But that era, like Post, Facebook, there was Foursquare, Twitter, Pinterest, Snap, there were just like a new one every year was popping up. And so they jump in on this. And the mobile Internet was expanding really rapidly in China at that time. There was a big boom. There were a bunch of big winners that came out of that era. Kuaishou hit 100 million DAU by 2013. So basically 18 months after they launch, they're a pretty sizable social media app. And they actually pivoted in that 18 month timeline from GIFs to videos. And so once they had the new product dialed, they ramped pretty quickly. Now there was a little bit of a slowdown between 2013 and 2019 because they didn't hit 200 million DAU until 2019. So they were sort of saturating then. But they were on a fundraising tear the whole time. They raised $350 million from Tencent in 2017. They integrated with WeChat to accelerate distribution. And the company was worth around $18 billion by 2018. Just seven years to the journey. So not bad, not bad. The quiet IPO was a particularly crazy moment for the company. So they raised $5.4 billion at the IPO. They were massively oversubscribed for that. So they say, hey, we want to raise 5 billion. And 165 billion of demand shows up from the market. So retail investors just went insane. They're like, what, 165 billion of demand, only 5 billion for sale. And so the stock trades up 192% at the open, and all of a sudden the company's worth $180 billion. And both of the founders are deca billionaires. Pretty sick Phantom cash. Fund your wallet without exchanges or middlemen and spend with the phantom card. So that didn't last, though. There was a massive sell off and there were a couple like speed bumps that Kuaishou seemingly hit shortly after going public. So China had a big crackdown on in regulation on local tech companies. ByteDance became much more dominant as a competitor and was, you know, had Kuaishou in their sights. And then the Kuaishou user growth just sort of slowed down and so there were a couple misses on, you know, Dau MAU growth. The metrics weren't looking as good and so the investors said, hey, we're rotating to other things. They were pulling out. And so the Stock traded down 80% within six months. So not great. But today the market cap's around $40 billion. Like it's still a very real business. And also just financially the business is very solid, like it's not losing money. You can think about it this way. So it's a $40 billion market cap 20. This is all USD. I converted everything so 40 billion market cap, 20 billion revenue, 11 billion gross profit, net profit 2.6 billion, two and a half billion. So that's like a lot of cash flow, a lot of net income to work with, certainly enough to do some training run, start a NEO lab or a the video lab or take some new bets. And so that's exactly what they did. So in 2024 they launched the first version of cling. And they did this just three months after OpenAI demoed Sora. So it was kind of like the Chinese answer to Sora. Since then they've updated Clang 30 times and carved out a nice little place in the grid of trade offs around model performance and cost. For cinematic footage, people often recommend VO3 still, but for Clang, it has some really strong characteristics in motion control and physics simulations. There's a number of places where cling's been outperforming there. And then there's a number of other sort of niche applications where cling's really great. And at $0.10 per second, pricing has been very attractive relative to some of the other options in the market. Although there are a bunch of ways to get free credits and then if you buy bulk. None of the pricing is exactly apples to apples, but they've certainly come out as a frontier quality model at a very affordable price. And that's led to a bunch of adoption as we've seen with the 12 million monthly active users. Interestingly, Kuaishou claims that CLING is gross margin positive. And so now it's unclear. That doesn't include training and we don't really I mean it's gross profit, so it doesn't include the hundred or so R and D people on the team. But my question is like what happens if they try and scale another order of magnitude? Will they be paying more for inference? Will they be able to get those chips? Because there's still this big debate over how many Nvidia GPUs should be able to be sent over to China. So there's a little bit of debate there. And it's also just. It honestly gets me excited about MSL dropping a video model because a lot of the same sort of like precursor elements in terms of like the training data in the ecosystem. When you think about just the raw data in reels that it could wind up being a model that's a little bit more opinionated. Maybe it has some mid journey sprinkled in there. Like I think that the msl. The pressure on MSL has been intense and the initial Vibes launch was not loved. But they certainly have the compute, they have the team now, they have the data, they have all the key ingredients to a really successful video generation model launch. The only question is when will it happen and will it be one of those temporary leapfrogs where it's better than Sora and VO3 for a month until the other labs update. Will it have its own characteristics? Because VO3 when it came out it was like oh, we can do audio in there. Then Sora was really good at vertical video and a lot of these labs, it feels like if you're in video there's more room to be opinionated and come out with something where it's like we're really good at making content that's all one scene or our model by default will do cuts and cut from different shots to another. Or it's really good at front facing videos or it gets the faces really accurate or does face replacement. You can upload your own. There's all these things, features that you can win on. I think nowadays since all the different base video models are starting to look pretty photorealistic. So a lot of it's about.

4:51

Yeah, I mean I'm so. I can't wait to see Meta's next real launch here.

11:57

Right?

12:02

Yeah.

12:03

They have so many advantages. They have the talent now the pressure is immense but they should be able to come out with something that's super competitive.

12:03

Yeah, Turbo puffer search. Every byte serverless vector and full text search built from first principles and object storage. Fast 10x cheaper and extremely scalable. So last month they did 20 million and of course, Kuaishou put out a statement about it. Their shares went 3.6% higher to rank among the top gainers on the Hong Kong exchange. Year to date, Kuaishou is up 23%. So they've been on a run and there's a lot of investor optimism about Kuaishow's potential to monetize AI. The company's AI powered video and graphics generation capabilities will likely drive earnings over the next few years, S and P global ratings analysts wrote in a recent note. That could lower content creation and advertising costs and boost content creation on Kuaishou Short form video platform the company has the second largest independent short form video platform after ByteDance's Douyin, which is basically the Chinese version of TikTok. And so is this going to be super important to the AI race in America? Probably not. Like we're still living in a world where you have OpenAI, anthropic Google really battling it out and then Xai is doing interesting stuff, MSL has interesting stuff and then there's a couple NEO labs that are maybe doing interesting stuff. But the Chinese labs haven't been putting a ton of pressure. But I think it's interesting because we're actually getting firm data on how these products are monetizing. Of course, Google, they're not even breaking out Gemini revenue or metrics. They'll throw out sort of random numbers. But it's not like YouTube where they have to report out specific things, or like Amazon with AWS where they have to break out those financials. Kuaishou is kind of doing a service to the public markets. At least they're creating a comp for what, what revenue looks like relative to MAUs. And I mean right now if they're doing 20 million in revenue on 12 million monthly active users, they're getting what, $1.80 per monthly active user, $1.80 a month. Basically like a $20 a year subscription. That still feels a little low, but not bad for total arpu for such a new. For such a new service that really we don't know the mix of like enterprise versus prosumer where they.

12:11

Yeah. And they're not breaking out where their users are. Right.

14:34

So CLING is available on aws. Like you can go and provision it in America. And so yeah, we're not exactly sure and I think some of that inference might run in America in that case. So not exactly sure where their biggest customers are. But it's just interesting to see you're not gonna get this level of detail from Google because De Mine's its own thing and then Gemini's in there and then you can't even just subscribe to just VO3. You have to subscribe to Gemini Ultra and then you get VO3 credits. And so even Google, like, I mean, I'm sure Google knows how much like I pay for Gemini Ultra. And then they know how many deep research reports I file, how many normal prompts I use, how many times I just talk to it. And they know my my usage pattern relative to how many VO3 generations I'm doing per month for my, you know, $250 or whatever I pay.

14:38

Yeah.

15:36

Anyway, moving on. Vanta Automate Compliance and Security Vanta is the leading AI trust management platform. There is another interesting, just sort of like nuanced narrative that I hadn't really noticed while I was digging into Kuaishou. One is this China's aging tech workers issue the curse of 35 they're firing unks over there. That's what's happening. They're firing.

15:37

You can't even be an unk.

16:06

You truly can't. You truly can't. Discrimination against older employees particularly apparent in sector where executives openly state a preference for youth and so different approach. We've had back and forth on, you know, in VC Twitter about, oh, should your team be really young and cracked or should you get the experts in there? Obviously in America you cannot discriminate based on age. But that hasn't stopped plenty of people from opining about what the correct mix is, whether or not you should follow those rules. But there's an article in the Financial Times that sort of has some charts here about how big tech groups in China have been downsizing in the past few years. We can go through that, but let's read through a little bit of this China's aging tech workforce.

16:07

First Hint Lao Bai, 34, received that his position at short video app Kwai show might be at risk was when a 35 year old colleague was sacked. I was both shocked and anxious. I realized that our situations were very similar and the same thing could soon happen to me, said Lao Bai, using his nickname to avoid repercussions from his former employer. Just months from his 35th birthday, the developer was dismissed, another victim of the group's reorganization, known internally as Limestone.

16:56

Wow.

17:23

They're giving code names to age discrimination? That's insane.

17:24

Brutal.

17:29

Kuaishou is pushing out junior workers in their mid-30s. According to five people with direct knowledge of Limestone, including current and former employees. Lao Bai was told his termination was part of the company's overall redundancy program. The so called curse of 35 has long plagued workers across white collar professions, with older staff widely perceived as being less willing to put up with long working hours because of responsibilities at home as China's tech sector reels from Beijing's crackdown, this article is from 2024, by the way, and economic slowdown. Tens of thousands of jobs have been cut over the past several months, and older workers are seen as particularly vulnerable. Technology companies have made no secret of favoring young and unmarried workers. Ageism in the tech sector is a big problem. There's a perception that older workers don't keep up with the latest technological developments, they don't have the energy to keep up the hard work, and that they're too expensive. Thankfully, our very own John Coogan, who is you're very much keeping up with the latest technological developments. Otherwise we wouldn't be talking about this because you didn't just read about cling, you studied and you wrote about it. While China's labor law prohibits employers from discriminating on the grounds of attributes such as ethnicity, gender and religion, it does not explicitly refer to age. Whoa. Hence Project Limestone. But Yang said some had interpreted the law more broadly as prohibiting discrimination against older people, meaning employers would not explicitly cite age as a reason for dismissal. Chinese tech executives have long publicly voiced their preference for younger workers. In 2019, Tencent President Martin Lau announced a plan to reshuffle 10% of the company's managers, saying their jobs will be taken up by younger people, new colleagues who may be more passionate.

17:30

Hmm. Baidu chief Robin Lin, in an internal letter also made public in 2019, announced the company's plan to become more youthful by promoting more workers born after 1980 and 1990. This thinking is deeply embedded across most tech companies. Well, let me tell you about Label Box reinforcement, learning environments, voice robotics, evals, and expert human data. Label Box is the data factory behind the world's leading AI teams. So that's right between 20 and 30, most people are full of energy. You're more willing to march forward and sacrifice yourself for the company. But once you become a parent, your body starts aging. How are you going to keep up with the 996 schedule? Says a former sales manager at Meituan, referring to the Chinese tech sector's infamous work routine of 9am to 9pm six days a week. ByteDance, which owns the video app TikTok, and e commerce giant Pinduoduo have some of the youngest recruits among Chinese tech companies data suggests the average age of their workers is just 27. According to the latest figures, the average age of staff at Kuaishou is 28. Against the 33 years old at ride hailing app Didi, the average age of the worker in China is 38. According to the All China Federation of Trade Unions. This trend has become only more entrenched with progressive waves of layoffs. And this is the chart that I wanted to show you. So big tech groups in China have been downsizing in the past few years. So this is 2021, 2022 and 2023 for Alibaba, Baidu, Kuaishou and Tencent. And they're all downsizing across those three years. Now I was wondering if this was just an anomaly post Covid just one particular period during some economic gyrations. So I went and pulled more recent data on what's going on, and it feels like they're all continuing to shrink. The one that's, that's growing again is Tencent. But Baidu has significantly downsized. Alibaba has continued to downsize and shrink, and I think Kuaishow has as well. And so there's an interesting dynamic there that I didn't. I mean, we hear about layoffs in big tech in America, but it feels like it's like a very temporary layoff and shrink it and then it starts growing again pretty quickly as they do more acquisitions. Now, to be clear, some of these companies have shrunk because they've divested whole business units or sold off a piece of a company. And so you can't really count that because it's like the jobs still exist. They're just at a separate company anyway. Lambda is the super intelligence cloud building AI supercomputers for training and inference that scale from one GPU to hundreds of thousands. We have our first guest.

19:16

Before we jump in, it's worth noting, I pulled up Mag 7 employee counts, and every single company in the Mag 7 has increased headcount over that same period that we just showed.

21:59

Yeah, yeah, there have been like periods of layoffs, but even when Mark Zuckerberg comes out and says, hey, we're laying off 1,000 people in the Metaverse team, in the Reality Labs team, it's like, well, he's adding tons of people to the AI teams and tons of people to the Instagram teams and threads teams and Facebook core and marketing and salespeople and events people, like the whole organization is just growing so fast that one niche layoff is not really going to cause A multi year trend, at least in America. We're still growing, so just sort of interesting that I didn't realize that was going on that there was such a concerted effort across four major Chinese tech companies to really reduce headcount. But it is happening.

22:10

Yeah. It is worth noting though that that in the chat is sharing this. Well, that 2022 headcount and current headcount look very, very similar. So there hasn't been a ton of growth. Anyway anyways, with that graphite code review.

22:49

For the age of AI graphite helps teams on GitHub ship higher quality software faster. And without further ado, we'll bring in our first guest of the show, Rich. Welcome to the TPP and ultradam. How are you doing, Rich?

23:03

Great to have you back.

23:16

Busy week, Busy week in media world. Earnings kicking off?

23:18

Yes, earnings. What's top of mind? What, what, what's keeping you the busiest?

23:22

Look, I think everyone's trying to figure out what is the future of Netflix. What happens to Warner Brothers? Does Paramount comeback? I mean, we talked about this on your show a few weeks ago, right? We were talking about sort of this battle. Does Paramount come back? You know, they haven't raised their bid since we talked, you know, before the holidays. They have not raised their bid as of yet.

23:27

But they did say it wasn't there. They said it wasn't their best and final. Right, that is true.

23:50

But you know, when you look at sort of the, how much Netflix clearly wants this, you know, Netflix yesterday went to all cash on their offer. The Warner Brothers board came out and talked about the piece that Netflix isn't buying, which Paramount said was worthless. The Warner Brothers board said was worth, you know, $3, maybe even more, as much as 4 to 5, even more if it gets sold into pieces, which we think is possible. And so the bar, you know, in terms of like what does Paramount have to do to win? That bar is going up and up and up. And I think that's the challenge right now is how badly do they want this? Do they want to over lever the company to get this or is there a better use of capital? I mean, it's not like there aren't other assets out there. I mean, you could go into the video game world, right? Like there's assets like take two, two of the biggest, enter actually the two biggest entertainment franchises in the world.

23:57

Okay, but what's actually, what's actually important for Paramount, is it to build a compelling consumer subscription, right? Like, because when you look at them overpaying for UFC or maybe Maybe they're not overpaying, but if they. If they.

24:49

It launches this Saturday.

25:05

Yeah. Which I'm excited for. But they pay 700 million or. Sorry, no, it was more than that. What was it?

25:06

They spent a Billion won on UFC.

25:17

Yeah, billion.

25:19

About 2x. What? It was what ESPN was paying.

25:21

Yeah.

25:24

And I guess my question is, how effed are they if they don't get the Warner Brothers assets? Just. And I. And I. And I mean that from like a. How compelling is their. Is their subscription Without. Without Warner Brothers, those assets?

25:25

Look, you can license a lot of content. I mean, Netflix, despite trying to buy Warner Brothers, Netflix actually just went out and did a global licensing deal with Sony. So after Sony movies are in movie theaters, the next place they go is to Netflix all around the world. That's a deal that Paramount could have done. They could have outbid Netflix for that deal. Those licensing deals are out there. I mean, Universal movies are actually going to now be. After they do a short stint on Peacock, they're going to show up on Netflix. Like, if you want to license catalog content, there is lots of ways to do it. I mean, heck, Amazon, the craziest thing that's happened this week, honestly, Amazon, you probably remember a few years ago, they went out and they spent $8 billion on MGM. Everyone thought that was absolutely insane. Then they had to go back because you actually, Even for that 8 billion, you didn't control the James Bond franchise. They ended up paying the Broccoli family another billion dollars to get access to control the James Bond franchise. And then what happened today, guys? Today the entire James Bond franchise is licensed to Netflix.

25:43

Netflix.

26:48

Like, literally, you can go watch Dr. No on Netflix right now instead of watching this show. Like, it's crazy, right? So there is lots of content to license, but you have to be aggressive. And so I think, you know, again, if Paramount has, you know, Larry Ellison's willing to put $12 billion into the Warner Brothers deal. The Middle east is willing to give them 24 billion to invest in this transaction. That's a heck of a lot of capital if you want to start really both creating a lot more content internally and licensing a lot more. Paramount has the ability to do that. There's more NFL rights coming up for bids soon. Like, there are many ways to attack this beyond buying Warner Brothers. And I just think there comes a point where overpaying for something gets silly and you're better off finding an alternative. And I think that's sort of the question mark, because, remember, for.

26:49

For.

27:42

For Paramount, they need to Buy all of these linear cable networks. Because unlike Netflix, remember, Netflix generates $11 billion of free cash flow a year. Paramount has negative free cash flow. So, like, the ability to bid here is very different. These are very different companies.

27:42

So what's Netflix's rationale for buying versus versus licensing? Is that just thinking long term that they're, that they'll make more money over the long term, or is there some other other rationale or value that they would get from owning the whole thing as opposed to just licensing it?

28:01

John, it's a great question because I think if you step back and they even said this, I asked this question like, you know, what changed? Because Netflix had not been interested in making acquisitions. They were a builder, not a buyer, you know, throughout their entire history. And I don't. If Paramount hadn't put this in play and like this asset hadn't become sort of in play, I don't think we'd be sitting here talking about Netflix buying Warner Brothers. I think once it became clear this asset was going to trade and that there was a way to buy it without getting all of those sort of dying cable networks, that there was a way to just buy the streaming business and most importantly, the studio. Because what this is really about, this is like when Bob Iger at Disney, remember years ago, he bought Pixar, he bought Marvel, he bought Lucasfilms. Those three iconic acquisitions of IP have totally transformed the Walt Disney Company. I think that's what Netflix sees here, is, hey, these IP libraries are going to trade. This is a once in a lifetime opportunity. Someone's going to own them. Let it be us. Like, let's not let this get away from us because we can do so. Just like Amazon for Bond needs Netflix, the ability to take all of that Warner Brothers catalog and shine a light on it. The way Netflix can, I think, was just too big of a, you know, talk about a 5, 10 year old opportunity is too big to pass up.

28:21

How do you think the studios and the platforms are kind of trying to process how Gen AI will impact the value of ip? Because I look at this and I'm like, if you have IP and the cost to produce new content drops dramatically, which I think is my, my assumption generally is that you're going to be able to produce, like when you look at animated content and even like kind of blockbuster films, I expect the, I expect the cost to produce these things is going to drop dramatically. Which means if you actually have a monopoly on the ip, you can create a lot more different content around it. And I Think that Genai will also mean that there's just a bunch of more and more and more and more content like net new content, net new IP being created. But it's unclear if you have. It will maybe even become harder to generate real breakout hits if the whole zone is just being flooded with a ton of content.

29:41

Well, look, in a world where, first of all, let me just step back. I agree with you, Jenny is going to lead to an explosion of content. And I don't think this is just video content. I think it's an explosion of content. So whether it's gaming content, you think about a UGC platform like Roblox, you think about a short form or shorter form UGC world of video like YouTube, you think about Spotify, like all of the platforms. In a world where content creation gets cheaper the platforms when that is exactly part of the long term investment thesis on why own Netflix now? Yes, it will get easier to make content from the Warner Brothers library. It'll get cheaper for Netflix. I mean certainly animated content, which takes three to four years to make. I mean, look, you know, one of our questions for Netflix on their earnings call was like how big of a miss is it that K pop Demon Hunters 2 isn't coming out until 2029? Like that's embarrassing, right? Like it shouldn't take four years to get that sequel out. But that's because of the timeline for animation that's going to compress dramatically.

30:36

Is that everything? Because I thought that there was also sort of just an unpredictable. It was an unpredictable success. They hadn't negotiated the contract with Sony and so they kind of got caught flat footed when it became such a success. And again, it was like it was a success that only would have happened on Netflix where a movie can go viral, where if that had had some limited run in theaters, it probably would have just been forgotten.

31:41

It's true. But they struck the deal at the end of 25 and you're not getting a movie till 29. Yeah, it's just the timeline. Like Sony Pictures Animation, which is up in Canada. Sure, there's a backlog of films that are before this and there's just no way to accelerate that at the moment.

32:06

But.

32:21

But I think I want to go back to Jordy's point. Yeah, making it cheaper to make content does mean the value of existing content probably comes under pressure. Like there is no doubt that since you're going to have so much more content created, existing content will have less value. I think that's a negative for Netflix. It's a negative for Disney. But the platform side of the equation, which is what Netflix is known for, just like YouTube and Spotify and Roblox, I think all of those platforms become huge beneficiaries because people are going to have so much more content on those platforms than they do today. So I think it ultimately becomes an advantage, hurts one side of the business, but I think as you look at it holistically, becomes a big long term advantage because the platform, big platforms have the eyeballs. Like that's where people turn to, right? Like you come home from work and you turn on Netflix, you turn on YouTube. Like those are the fundamental winning platforms and they're battling it out. And I think AI makes that battle, even both of those companies even more powerful.

32:21

And so sounds like if I could summarize, the place that you're getting to is maybe not believing that Paramount and the Ellisons should just lever up to this insane degree when they could potentially take some more, take maybe still tens of billions of dollars, but just allocate it across a wider variety of deals. Is that, is that, is that one path that you're seeing?

33:24

I think this is they're stretching, right? Like when they tried to buy this thing at 19 and there were no other bidders, it was a brilliant idea. Maybe even at 23, it was a brilliant idea. As you start getting, you start to realize that like you're not winning at 30. That's pretty, I would assume to both of you and me, like, they're not winning at 30.

33:50

Yeah, right.

34:08

Like, so they're going to have to bump and they're going to have to bump meaningfully. And so to pay 34, $35, like you start to get to valuation levels, I don't think you can put any more debt on this. And so you got to come up with 10, 12, $14 billion more equity, like, and even that is too levered. Like I wouldn't want, I would not want to be in this transformative AI media world sitting there with seven times leverage. That is not a great place to be. And so I truly believe David Ellison has a huge opportunity over the next decade. But I don't think over levering to buy Warner Brothers is the best path forward.

34:08

Also, it's worth noting that State Oracle at The beginning of Q4 was trading at $313 a share. It's now trading at $174 a share. And so that has to be a factor too in terms of everyone's confidence around like, hey, how much, how much.

34:46

Cash do you have available?

35:07

Yeah, look, Larry's got plenty of capital. If you wanted to buy this for all cash, he could. But remember, he's not. This is $100 billion transaction. The Ellison family is putting in 12.

35:08

Yeah.

35:19

You know, so, like, there is a limit on how much. I think the Ellison family, they put in, you know, $8 billion or, sorry, they put in 6 of the 8 billion of cash into the Paramount transaction. Redbird put in the other two. So six plus 12. That's an $18 billion commitment between the two deals. It's a big number. But on the flip side, you know, given the size of this transaction, you know, I think there is a limit to how much they probably want to invest in this sector. And again, I just think leverage is not your friend. Like, I look at sort of how you win over the long term in this media space, especially given how much change. And I think you two do a great job highlighting sort of how technology is changing. What's happening. Being levered is the last thing you want. And I think what was really interesting is that when Skydance bought Paramount, they injected cash onto Paramount's balance sheet to delever it because they didn't want to be levered. And so I think even they understood that they don't want to be levered. They're being forced to lever to try to compete against Netflix. And I think that's actually the mistake. And I just think that there's other ways for them to win long term. Overstretching the balance sheet does not, to me, seem like the right path.

35:20

Explain the cable assets that are sitting within Warner Brothers discovery. You have cnn, I believe there's a few other.

36:38

Yeah. And when some of the numbers are floating out, CNN doing something like projecting $600 million of EBITDA.

36:46

Yeah. Seem like a good business. My question is, you have this Netflix fight over the Warner Brothers, like that, the studio, the IP library. And to Netflix, it feels like the TV assets are less valuable. Certainly they were carved out of the initial offer. Are those more important to Paramount, Skydance, the Ellison family? Is there a world where Netflix takes Warner Bros. The studio and the cable assets go to Paramount.

36:54

They're critically important to Paramount because Paramount, as I said, doesn't generate free cash flow or doesn't have the earnings capacity of Netflix. And so in order to finance this, to lever this up seven times, you need all of that cash flow. If they were only buying what Netflix was buying, John, this would be levered over nine times. And so you have to have those assets. If if you're going to finance it the way they're doing it.

37:25

Okay.

37:48

And not do this as a cash deal.

37:48

Yeah.

37:50

In order to do debt, they need those assets. This becomes really important assets. And so for Paramount, they have to stop the split of this company which is supposed to happen this summer. If Warner Brothers splits, Paramount can't do this deal because they can't finance it the way they currently are financing it. The reality is these assets like cnn. Do I think public market investors will like these assets? No, I don't think so at all. I don't think these are growth businesses. I don't think investors are going to be excited at all. But I do think that there are strategic buyers of these assets where there is the influence and power of owning a news network that will far outweigh its public market trading value. And I think that's really the opportunity. If you put this up for sale, someone's going to come in and buy CNN because it actually has important strategic value. You think about like a nexstar which is trying to build their own news network. Trying to build out. They're in the process of making a major acquisition of Tegna. There is certainly an opportunity where I think you'll see multiple buyers come out of the woodwork. Maybe even billionaires who want. Or hundreds of billionaires who see this as a way to influence. Because remember, you could change the direction of CNN from the perception of liberal to. You could move it more center. Right. If you wanted to. And so I think there's a lot of people that would look to buy CNN if it was available.

37:50

What about. Is there any political saving grace for the Ellison's? Is there any chance that the Netflix deal gets blocked? Is that the angle? Like if they're not increasing their offer, I'm sure they're back channeling trying to get it killed. That would be my assumption.

39:19

I'm sure they are. But. But you guys are the. It's funny doing what we're doing right now.

39:40

Yeah.

39:46

Is the single greatest reason the deal should be approved.

39:47

Yeah.

39:49

And I mean that sincerely.

39:50

Yeah.

39:51

Like, look at what we're doing.

39:52

Yeah.

39:53

We are sitting on the Internet.

39:53

Yep.

39:55

On. On X. Broadcasting live.

39:56

Yep.

39:59

No cable subscription required. But you are absolutely competing against the NBC. And like there is no doubt in my mind that what you're doing is a real comparable, you know, a true competitor to linear television. So trying to tell me that you have to define, you know, streaming TV, subscription TV, let alone free streaming TV like YouTube and X, that we're going to put up artificial barriers and say this is somehow different than watching CNBC or CNN or Fox News. Like, look, you can make the argument. Will it hold up in court? I would love to see how a judge is going to isolate those categories.

40:00

Yeah, yeah.

40:43

You mentioned Take Two Interactive. It's a $44 billion company. GTA 6 is coming at some point. How does that make any sense for either Netflix or Paramount to own? It feels like for both of them, it would be a massive. Just different business line to enter. Obviously, like, in the attention economy, they do compete. If you're playing GTA 6, you're not watching CNN. But they just feel like fundamentally very different businesses to me. Is there some synergy that I'm missing?

40:44

Well, I'd say for Paramount specifically, I mean, Skydance. Skydance Games is a real focus. I mean, Ellison, you know, I think if there's anyone in the traditional or legacy media world that bridges Northern California and Southern California, it would be David Ellison. You know, he grew up friends. Not just, obviously his dad is Larry Ellison at Oracle, but he grew up friends with Steve Jobs. Right. And, like, so if you sort of look at sort of his understanding of technology, you have to believe that he understands the importance and power of gaming. It's why Skydance Games has such big ambitions. And so, you know, I look at sort of the power of that GTA franchise and Red Dead, and, like, you look at sort of like the future and you go, sure, it'd be great to make more movies, and it'd be great to have a bigger studio with Warner Brothers, but there's many ways to win in entertainment, and I think video gaming is a huge category. And while I don't think Take Two and Strauss is looking to sell anytime soon, look, it would be a much smaller transaction relative to Warner Brothers, and you would get some incredible IP and you would shift away. Rather than owning more TV and more sort of legacy studio, you would shift the focus of the company in an entirely different direction, which I think could be really exciting.

41:19

Yeah.

42:37

Why did Jared Kushner pull out of the deal?

42:37

You would have to ask Jared Kushner. I have absolutely no idea. But, I mean, I'm. I'm suspecting that just politically, it didn't look great. I mean, it might have been one thing when this was a friendly, you know, straightforward acquisition, but when you were getting into a competitive bidding situation and it was sort of being looked at as sort of, hey, this is the. You know, this is someone tied to the president that's influencing it. It Might have just been this did not look like he thought it was going to look. But again, I honestly don't know. I've never. I've never met him.

42:44

Yeah, that makes sense.

43:15

How are different video generation AI models? Are any of these factors showing up in earnings and financial reports and conference calls? How are management teams messaging around generative AI? Is it just a growth opportunity or is it actually becoming something that moves the needle either on the cost side or the revenue side for anyone?

43:18

I would really keep your eyes on Amazon. I think they're the first mover here and really trying to make a dent here.

43:42

Yeah.

43:48

And it's not to make less content, you know, or.

43:49

Sorry.

43:51

To make content cheaper and spend less. Amazon's literally looking at this as, can we actually use AI to bring the cost down 20, 30% and then make more content for the same budget? And so this sort of speaks to what you guys were sort of hinting at before. This is a way to increase the total amount of content. And obviously more content means more engagement. And so I really think Amazon's taking a lead here. Netflix has their own AI playbook. I think you're going to start to see it up from Netflix and of course, the other company that's talked about it. We haven't seen anything. But the other company that's most vocal about it would obviously be Paramount, where Ellison certainly believes in the power of technology to change the cost structure of this business.

43:51

What about Disney? Disney did the big deal with OpenAI.

44:35

Yeah. I wanted to ask if you're hearing about other IP holders rushing to try to do. Like, when we saw that deal happen, I was like, wow, this having like a one year, maybe more exclusive on all the Disney ip, I feel like could be pretty significant. Once it's live, it's going to be incredibly viral. Parents are going to jordy.

44:37

What's the last time you used sora?

44:57

I was never bullish on Sora specifically.

44:59

But that's the problem. Right. Like, so all of this is.

45:03

I don't know what I was gonna say. What I was gonna say I was bullish on the. On the Disney IP deal specifically because parents are gonna be willing to pay to bring joy to their children by turning their life into Disney scenes.

45:08

Yeah. Or I mean, do kids just love watching Bluey? Cause they love bluey. Like, I don't. Do you want to watch yourself in a, you know, I guess for the creative parent that can actually tell an interesting story. I mean, I get the gimmick. Like, I remember when Sora first came out, we all did it, right? Like we were all trading these around, sending around. I remember I was in a Knicks jersey before they put the copyright controls in. I was in a Knicks jersey on the court dunking like it was. I was catching the game winning pass and you know, for the Giants, like. And I think, I think, I think.

45:26

You'Re a little, I think you're a little bit jaded. I mean Disney has a whole business of just like have lunch with a character, right? Parents are spending.

45:57

I mean it goes back to, you're at Disneyland. The kid puts their face through the cardboard cutout and they appear to be Superman.

46:04

Anecdotally, when I first took an image of my son and I turned the two of us into dinosaurs. He has never forgotten that moment. And he'll bring it up all the time, turn us into dinosaurs. And I'm really like, we're not using AI right now. We're gonna go, we're going to go to the beach.

46:12

Look, I think it all depends on how sort of how preconceived it is. Like, is it just turning it in and you have to go figure out how to make something interesting and compelling or is it sort of already pre engineered and preset where you stick in your faces and the scene comes to life? Like, I think it's got to not be a lot of work because I think, you know, people don't. I think I will say I am not a terribly creative from like a writing stories. Like, you know, people need to be taught how to do it. I think that initial how to prompt how to make it good is not easy. And so it'll be interesting to see how easy they make it for parents to make this really a great experience. So I like the idea and look, I give Disney credit. You would not have expected Disney to be the one to jump out in front on AI given how much they're focused on sort of protecting their ip. So I give them a lot of credit for trying. And I think the key is Sora needs to become something that really becomes less of a gimmick and more where you actually want to sit down and actually like watch the content. Because I still feel like it right now. It feels very gimmicky.

46:31

Yeah, yeah, yeah, I do. I wonder, I wonder what OpenAI is getting out of it, right? Disney has 150 million people visit the parks annually. If they're if in the line, there's like a QR code that's like, turn yourself into such and such character while you're waiting. I think like right now OpenAI has a challenge which is that free LLMs are pretty great. Right. And they have this gap.

47:37

Free is great.

48:05

Yeah.

48:06

They have this real gap between turning on commerce and turning on ads and really unlocking that from a revenue standpoint. And I think if they can making these sort of like magical products that are unique to open AI that you're not going to be able to get at at through Gemini or any of these other LLMs, I think it could buy them a little bit of time to turn on some of the other monetization. But again, I think you're right on Sora. I wonder how they're doing. See if they're still in the app store top 25.

48:06

They've been in the top 100. I think they're around 60ish or something. And I agree with you, I haven't used the Sora app very often, but I am seeing clearly content that has been generated with Sora distributed on Instagram. And so there is a world where you create a prosumer tool that's subscription that people are paying for and then the really creative people are distributing that content elsewhere. They've already said that some of it will go on Disney. And so you could have someone who's a creative storyteller tell a story that takes place in Disney IP as the world that they're interacting with and then that does do well and it goes out. And maybe it's not every single person using the Sora app to generate their own videos, but that you're seeing content that was generated through that partnership and it's monetizing in a variety of ways. I don't know.

48:40

This all gets back to what we said before. Yeah, there's going to be a heck of a lot more content.

49:33

There's going to be more content for.

49:36

Sure over the next five years. There's going to be an exponential explosion.

49:37

Yes.

49:41

Of content.

49:41

Yes.

49:42

And it's funny you mentioned Instagram. Another example. Platforms, yeah, platforms win. That is the key. If you own and control a platform and the cost of creation comes down, yeah. You're going to be a long term winner. And so that's why Spotify, why Netflix, why YouTube, why Roblox? Like that's how we're positioned is picking platforms are going to be the big winners over the next five years.

49:42

So maybe that means that Paramount plus Warner Brothers and HBO is more of a platform than separate and there is actually accretive value to putting those together and overpaying. And so maybe There is some rhyme to the reason behind David Ellison's aggression here.

50:05

Building a platform, a daily use platform is really, really hard. I'm not saying it's impossible.

50:22

If you have Looney Tunes and Superman and Batman and you're on there as a kid, you grow up and it's one of the key apps that you open next to Netflix, next to Disney.

50:28

The funny thing is in what you're saying is it's all the reasons why Disney should be bidding on Warner Brothers studio as well.

50:42

I feel like that one just aesthetically actually does trigger ftc. Just because Superman and Spider man can never cross paths. Batman and Iron man, it's too similar. They're billionaires, they're, they're tech guys. It doesn't work. They gotta be separate universes.

50:49

I'm just saying they should want a lot more content.

51:06

They should, oh, 100% if they could get it done. And they could make the same argument that Disney Plus Warner Brothers is no more powerful than YouTube and Spotify and Instagram and all the other platforms that are soaking up people's time, including what.

51:08

Are you paying attention to this earnings cycle?

51:27

I mean, look, I think the, you know, this is obviously the 26 outlook for a lot of these companies. And I think, you know, you're going to hear sort of like in a world where so much is changing so fast. I mean, we came out with our and we do a quarterly earnings scorecard where we sort of highlight phrases that are used and it'll be interesting to see. While I dipped in terms of mentions last quarter, do we see even more companies talking about AI as we move through the year? I think that's going to be really interesting to watch. And is it in a positive or negative light and like who's being disrupted by or helped by I think is going to be really interesting to watch. And then of course just going back to where we started, you know, whether it's Netflix or Paramount, someone's going to win Warner Brothers. Well then the question is what does everyone else do? Because everybody else looks really small. And so sort of seeing, you know, the reverberations of what happens because of all of this is going to be really important to listen as you move through earnings.

51:32

What do you think about some of the private movie studios? Do you think any of those will get picked up? A24 did around at 3.5 billion thrive capital mid 2024, certainly building a lot of IP. They found an interesting mix of content. I really like a lot of their films. Do you think anyone will try and acquire some smaller studios.

52:28

Look, the biggest problem is smaller studios don't move the needle, which is why I think, you know, something like Lionsgate hasn't been bought. But why Warner Brothers? You're seeing this massive interest. I don't know about the smaller studios. I certainly think like as you look at something, you know, you look at Sony Pictures or Sony, you know, Sony, you know, pictures and tv, there are assets out there that, you know, look like they are ripe for, you know, M and A. Whether it's sale or some form of combinations, it does feel like you're going to see more combinations in the year ahead.

52:53

Yeah, but again, you can also just license their content, which is what Netflix and many of the other streamers have done. Well, thank you so much for taking the time to come chat with us. Always a good time to hang out.

53:24

Great to hang.

53:33

Have a great rest of your week. We'll talk to you soon. Rich Goodbye Plaid Plaid powers the apps you use to spend, save, borrow and invest securely. Connecting bank accounts to move money, fight fraud and improve lending.

53:34

Now with AI, Back to the timeline.

53:47

The messy drama that dealt a blow to one of AI's hottest startups. This is an exclusive in the Wall Street Journal After a relationship with a colleague, a thinking machine's co founder had his role changed Months later he was fired after a contentious meeting. Megan Bobrowski has the story in the Wall Street Journal alongside Keech Hagee. Mira Moradi's meeting with her co founders Going off the rails Murati, the chief executive of AI startup Thinking Machines Lab, had shown up for work on Monday last week expecting to have a one on one with Barrett Zof, her chief technology officer, according to people familiar with the matter. Last summer she had learned that Zof was in relationship with a colleague in the months since she had expressed repeated concerns about his lack of productivity, according to the People. She was invited instead to an impromptu meeting with Zof, another co founder and a third employee. The three told her they disagreed with the direction of the company and that they were considering leaving. They asked Zof to be given charge of all technical decision making, according to the People. Moradi responded that Zof was already CTO and asked why he hadn't been doing his job for months. So they're clearly beefing. That's a very funny, it's a funny request. Like what was he doing as CTO if he didn't have full control? But I mean obviously CEO can override all sorts of stuff and there's other people around the table and titles only mean so much. There's soft power all over the place. So two days later, Zof was fired. Within hours, all three had signed offers to rejoin OpenAI, the AI lab that they ditched a year ago to join Marathi's flight fledgling startup. The departures are a sign of how the heated AI race that is consuming hundreds of billions of dollars and transforming the economy is as much a battle for talent as technology. For all the high tech advancements AI startups are springing to develop, they are ultimately at the mercy of the humans powering.

53:52

So do you think this was a three hour talent acquisition process? Do you think OpenAI just, just looked at, checked the timeline and said, hey, we got three people that are on the market, maybe we should hire them.

55:52

It seemed like they were talking before.

56:06

They entered the portal and then they.

56:08

Immediately entered the portal.

56:09

The trade portal for sure. That's exactly what happened.

56:11

No, I think, I think very obvious that the conversations had been ongoing. It's very likely. I would assume they had already agreed to terms. Yeah, it seemed like it prior to the news going out.

56:13

Well, let's continue. But first, Vibe Co, where DTC brands, B2B startups and AI companies advertise on streaming TV, pick channels, target audiences and measure sales, just like on Meta. So Zoe's firing and decision to rejoin OpenAI with colleagues also marks a pendulum swing for a company that Marathi, the startup that. For a startup that Marathi, that startup's former chief technology. Former technology chief had founded with 20 former OpenAI employees. Employees. So 20 people from OpenAI left during the Thinking Machines Foundation. Addressing Zof's departure to Thinking Machines employees, Maratti said there had been multiple issues with his performance, trust and conduct. According to an internal message viewed by the Wall Street Journal, Zof said she fired him after he exposed an intent to take a job elsewhere. Thinking Machines terminated my employment only after it learned I would be leaving the company, full stop. At no time did Thinking Machines Lab cite me on any performance reasons or unethical.

56:24

John, I'm breaking up with you. You're like, no, you're fired.

57:25

I'm bringing up you. It does feel like a little bit of that going on. But yeah, there's a war for narrative here between Barrett and Mira. Clearly they didn't. So he says Thinking Machines didn't cite any performance reasons or unethical conduct as part of the reason for the termination. And any suggestion otherwise is false and defamatory. Zof said in a statement to the Journal. The exits, coupled with fellow co founder Andrew Tullock's decampment to Meta last fall leave Thinking machines with just three of its original six founders. Murati spent six years at OpenAI, where she earned a reputation for emotional intelligence and lack of ego, and was named interim CEO during the brief period when CEO Sam Altman was deposed. She helped launch its first product and ran almost every aspect of the the company before starting Thinking Machines last February. Many of the early researchers, including Zof, that she hired came from OpenAI's Post Training Team, the division that built ChatGPT, and was tasked with teaching AI models how to communicate with humans. Murati's issues with ZOF started over the summer when she began to suspect he was having a relationship with a colleague with whom he had lobbied to bring over from OpenAI, according to people familiar with the situation. In responses to questions from the journal, Zof said that many people at Thinking Machines wanted to hire the woman, including Maratti. At the time, Marati was in the process of raising one of the largest seed rounds in Silicon Valley history. The company ultimately raised 2 billion at a $12 billion valuation. When she confronted Zof about the possibility of an undisclosed relationship with a female employee who was junior to him at the company but did not report to him, he initially denied it, according to people familiar with the situation. By June, however, both Zof and the woman had told Maratti about the relationship, which had begun when they were colleagues at OpenAI. People with knowledge of those discussions said the woman then left the company and returned to OpenAI, which is sort of a wrinkle in this. Zof told his boss that he had been manipulated by the woman into a relationship, according to people familiar with the matter. Shortly after that conversation, he took a break from work.

57:29

Okay, I gotta. I gentle. Generally think that the Mira.

59:32

She told me it was cuffing season. I didn't know what it meant, so I just said, okay, I got cuffed. Is that what happened?

59:38

Yeah, I mean, I think just saying, like I was manipulated into becoming the significant other. You know, you gotta take a little responsibility.

59:45

You know, everyone is assuming that this is a romantic relationship. They never reported that this is a romantic relationship. They just said that he had a relationship with someone else and it was undisclosed. And I think that more companies need to be clear about the rules around disclosures of relationships. Like if you and Tyler start an esports team, for example, like you would have a relationship, you would be teammates on playing Call of Duty. And if you didn't disclose that to me, I would be. I would feel left Out. I'd be like, why don't I. Why am I not on the team? If I find out that. That Tyler and Scott are going off and drinking a bunch of athletic brews every night without me? That's a relationship.

59:54

They're bros. And it wasn't disclosed.

1:00:36

It wasn't disclosed.

1:00:38

And you might be very angry. I might be angry that they didn't disclose.

1:00:39

I think relationship disclosures need to go beyond romantic relationship disclosures. If you're just broing down with people, that actually.

1:00:43

That actually is kind of a real thing. Sometimes you're catching up Monday after the. The weekend, somebody's like, oh, yeah, I was hanging with so and so on Saturday.

1:00:52

Like, wait, where was my invite?

1:00:59

You guys hang out on Saturday?

1:01:00

What? You didn't disclose that Relationship.

1:01:02

Korean Barbecue.

1:01:05

Yeah. Oh, so you have a relationship where you go to Korean barbecue together.

1:01:06

Okay.

1:01:10

And you just didn't think to tell the rest of the company that?

1:01:10

Kbbq.

1:01:13

Yeah.

1:01:13

What's your thoughts?

1:01:16

I was gonna say, I think there's. You know, I like to imagine there's some kind of, like, Shakespearean tragedy here, right? It's like the Montagues and the Capulets. That's OpenAI and thinking machine.

1:01:17

Oh, yes.

1:01:25

And it's these forbidden lovers who can't.

1:01:26

Be together, but now they are. They're united. Because everyone is an open AI. Even though, I mean, if you go to the Journal and you say, I've been manipulated by that woman who's now my colleague. Because they both work at OpenAI now, right? So that's odd.

1:01:28

Wait, do we know that.

1:01:43

The woman left the company Thinking Machines and return to OpenAI? So Barrett and the woman in this relationship, which might be romantic, they might just be playing board games together or playing COD professionally together. Who knows? But they were both at OpenAI. They both went to Thinking Machines, and then they both went back to OpenAI. And so they're all in the Montague camp now, or the Capulet camp. I don't know who's who. I haven't studied enough to assign roles. Have you read Shakespeare recently?

1:01:46

Well, at the end, they both die.

1:02:18

They both die, and they get steamrolled by a Neolab. Is that what happens in Romeo and Juliet? I don't know.

1:02:22

Okay, one thing is rewinding to November 13th. Bloomberg reported thinking machines in funding talks at $50 billion valuation. So that's a big question.

1:02:28

12.

1:02:42

Yeah.

1:02:43

The big question now is, did any of this financing talks actually convert into real investment? And if not, I think that is very Little chance that a round gets done in that territory anymore.

1:02:43

Yeah, yeah. It is like, certainly not the best thing to have hit in the Journal if you're out on the fundraising track at the same time, these mega rounds, it just feels like they take a long time. We've been hearing about the anthropic round for a really long time. And we've also been hearing that the anthropic round, even though it's been going on for, I don't know, six months or something, it's going really well and everyone's excited and there was a whole bunch of FUD around the Xai round. Is it going to happen? Have they hit benchmarks? Is there too much fear around all the different hallucinations and whatnot? Are they seeing real traction? The round got done and so I. This is not like it's not over because these rounds can just take a long time. So it's not like, oh, you need to have the round closed in three months. Yeah, but if you're raising it.

1:02:57

But yes. Mirrors out there. Fundraising.

1:03:47

Yeah, but this is obvious.

1:03:49

Yeah, we've three covers. We're figuring a lot of stuff out. You know, we have a lot of momentum. Really excited about what we're doing. And then you have articles in the Journal that are saying your own co founders disagree with the direction of the company. I think most investors are going to be like, hey, we should probably chat before we give you another $5 billion.

1:03:50

But Mira is going to say they only disagree because I fired them for having relationships.

1:04:12

He said, she said, we're solid.

1:04:17

Yeah, it is a he should. And it has Silicon Valley talking. Over the next several months, Thinking executives at Thinking Machines witnessed a drop off in Zof's performance after he was cycled back to a new technical contributor role with reduced executive and managerial responsibilities, and they were tracking his performance. They said his usage of slack, the main arena where the company did its work, declined precipitously in the following months. He was locked in. I don't know. I don't see a problem with. If you go from a manager to an individual contributor, you don't need to be slacking as much. You need to be locked in. You need to be grinding writing prompts. In his responses, Zoef said he worked as an individual contributor for two months and also worked on projects that included recruiting and retention of talent, roadmap planning and the release of the company's model training product, Tinker. He said he was also out of the office for parts of November and December due to illness And a death in the family. Very tragic. That's very sad. In addition to the holidays. So he's like, I'm not on slack. I was with my family on Christmas in October. Shortly after Tulok left for Meta, where he had previously worked, Zof reached out to Altman to talk about coming back to OpenAI. By the time of their meeting with Muradi last week, ZOF co founder Luke Metz and third employee Samson Holtz had been unhappy with the direction of the company for months and had been holding Talks with both OpenAI and Meta in recent weeks about joining either company as a trio, according to people familiar with the matter. Interesting that Meta, amid the crazy buying spree that they've been on in the talent wars, did not jump in with something, you know, so much higher. They already have Andrew Tulak. They've already got one thingy Machines co founder. They're basically acquiring the whole team piece by piece. They're going to wind up paying more than the 12 billion price.

1:04:19

Yeah.

1:06:10

And I don't know, I mean it wasn't that long ago that people were speculating, okay, what, what is thinking machines worth in an Aqua Hire scenario. And based just on that type of analysis, it's worth a lot less today than it was less talent a few months ago.

1:06:10

I wonder what they mean when they say unhappy with the direction of the company. What do they want? Are they excited about the ads product at ChatGPT? Are they excited about Sora? Are they excited about OpenAI for health or for science research? Particular technical direction? Just the GPU scaling, the build out, just how rich OpenAI will be with the backing of Oracle and all that. Yeah, it's 1.4 trillion. Little bit nerve wracking if you're an investor, but if you're an AI scientist, you're like, I'm going to have unlimited resources. And so it could be that. I don't know.

1:06:25

I think, I think the big question is obviously what is. What's Albania's move? Standby Mira ride with Mira potentially lead this round at 50, right. Albania, the government of Albania isn't investing $11 million no way into thinking machines. No way they could decide to take a bigger swing though. The GDP of Albania is somewhere around 30 billion USD.

1:07:01

30 billion.

1:07:27

And anyways they might take a Trump intel kind of dynamic and say we need national AI champion and potentially nationalize thinking machines. They could have like a joint venture.

1:07:28

Between nationalize a company that's based in America. If you're not an American joint venture.

1:07:39

Between A16Z and Albania.

1:07:44

Yeah. Yeah. I saw a vibrio of Jamie Dimon talking about, like, how important it is that his bank is big. And he says JP Morgan has to be big because we bank, you know, all of the Fortune 500. We bank governments. We bank the US government. That's what he said. It's such a good line, banking government. I really want to know what's involved in that. Anyway.

1:07:46

Yeah, I think my takeaway from this is the pressure for Thinking Machines to deliver went from pretty extreme to maybe slightly more extreme, but it's actually not that much of a difference. I feel like they were already under a lot of pressure, similar to Meta, where dense talent concentration haven't shipped anything, you know, truly remarkable yet. Obviously, everyone's looking to see if they're going to be a real horse in the race.

1:08:06

Yeah, yeah, yeah, yeah.

1:08:39

The other thing is, like, this team was created during an insane period of drama.

1:08:43

Yeah.

1:08:48

So it's not all that surprising that born in the drama, they were born in the drama. Now they're living in the drama.

1:08:49

Yeah.

1:08:55

And. But they can consolidate. They can refocus, they can realign the team. They still have a bunch of super talented people, and I'm sure they can still figure out kind of a niche and deliver.

1:08:56

Yeah. Anyway, Okta. Okta helps you assign every AI agent a trusted identity. So you get the power of AI without the risk. Secure every agent. Secure any agent. Okta.

1:09:09

Well said.

1:09:22

After the meeting. So the three co founders, the three Thinking Machines employees that have this meeting, Mira asks them, have you already committed to jobs elsewhere? Metz and Schoenholtz said they hadn't. Zof didn't answer. According to some people that were in this meeting. The day after the meeting, Zoaf had dinner at a pizzeria with Meta executives Alex Wang and Nat Friedman. He confirmed on Wednesday he was fired. Maratti posted on accident, Thinking Machines had, quote, parted ways with Soph less than an hour later. OpenAI CEO for applications. Is it CEO for applications? Fiji CMO? I thought it was CEO of applications. Fiji CMO. Anyway, she's running the application. She posted that Zof, Metz and Schoenholtz were returning to the company and that negotiations had been underway, quote, for several weeks. That's an interesting thing to put in your. Put in your hiring post because there's clearly gonna be a narrative about how sudden was this, when was this decided, blah, blah, blah. But she is saying proudly that we've been talking to these folks for weeks. She said Zoph would report to her while the other two would report to him. So the craziness with thinking machines, end.

1:09:23

Of the day, you got Mira, you got John Shulman.

1:10:37

Yeah.

1:10:39

Like, still.

1:10:40

And it feels like there's. Yeah, it feels like there's some sort of niche that they could carve out. I mean, when we first heard about the fine tuning for LLMs for business, for B2B companies, that seemed like, not. It was not the AGI Super Intelligence pitch. It was also not the, okay, we're going to niche down and give consumers one specific thing. We're going to go after audio, we're going to go after video, we're going to go after. We're going to make an AI version of Excel or AI version of PowerPoint. There's so many niche AI companies that have done very well. We're going to do AI for legal, AI for health. Right. Like, there's all these ways that you can niche down. They kept it sort of broad, but. But it felt like going into the enterprise with her connections and her background and the team seemed like something that could ramp revenue very, very quickly.

1:10:41

Yeah.

1:11:30

And the raw talent of the team.

1:11:30

Totally. Yeah.

1:11:32

A lot of which still exists.

1:11:32

Yeah. So. But I mean, really, like, the strongest way to counter, like, the vibe or that's happening right now is just show some good numbers. Because if the business is working, even if you have a little bit of tumult on the technical side and there's, you know, a CTO leaving, like, if the business.

1:11:34

Yeah.

1:11:51

Like with. With OpenAI, it's like, you know, some of the craziest drama ever surrounding a company. And then you look at, you know, the revenue ramp and it's like, you.

1:11:52

Know, and the DAUs are just through the roof and you're like, okay, well, clearly the user was completely unaffected. I bet if you look at the ChatGPT Daily Active User Growth chart, you cannot see Sam Altman getting fired and then rehired. It's not like people were like, well, I'm not using this app today and I'm firing it back up. Once he said, like, no, there was no blip in the graph, I'm sure, around that. Because this was like Silicon Valley drama. And for the average user, it's like, did it deliver or not? And that's just something that happens all the time when there's these drama elements that burst onto the scene and take over the timeline for a few days.

1:11:59

Yeah. Another possible route. And things would have to get really bad over there. They'd really have to be directionless, which, again, the team clearly was debating on if the direction was correct, but it sounds like they do have a direction. But I still think this is a company that would have a pretty meaningful valuation in an acquihire context. Right. Like, if Amazon were to come and buy them, how much would Amazon pay to get. John Shulman and Mira Morati at Amazon.

1:12:36

Should be able to clear the pref stack for sure.

1:13:10

Yeah, by like a long shot. You would imagine so.

1:13:12

Anyway, MongoDB choose a database built for flexibility and scale with best in class embedding models and re rankers. MongoDB has what you need to build. What's next? I wanted to look at this photo of Davos. It's AI generated, but it captures a good vibe. A whole bunch of private planes parked there. That's not real, but I like that Kevin Kwok contextualizes it anyway, saying, this looks like Hoth Echo Base. And he shared one further down. That's Davos transformed into a Star wars scene. Anyway, over in Davos, Trey says we.

1:13:15

Need to know what those machines are thinking.

1:13:54

You call it Thinking Machines. Intelligence. Thinking Machine Intelligence. TMI Labs.

1:13:56

Yeah.

1:14:01

I mean, we don't know. It's possible that when they think, when they're there, they were debating the product direction. They were talking about, should we make branded 45s, right? We know they make. We know they have branded plates.

1:14:02

Oh, yes, yes, yes, yes. But maybe that was the technical direction, like do you lift heavy or do you do lower weight, higher reps? And there was a massive debate over that. Anyways, I mean, everybody is another relationship that you need to disclose. If you're spotting each other, I want to know about it. I want to know who's spotting who, because that's an important relationship that's happening outside of the workplace. I got to know about it if you're going to be lifting together.

1:14:14

Anyways, I'm rooting for thinking man machines and excited to see what they do.

1:14:40

Demis and Dario are apparently just brought in Davos. I love it. Lisan Al Gaib says Demis Hassabis is CEO of Google DeepMind at Davos. Quote, when told that Dario Amadei was here earlier today, his face lit up. Two minutes later, he was talking about a CERN like collaboration for AI Once the last are close to AGI. They're just stoked on each other. I'm on pretty good terms with pretty much all of the other leaders at the leading labs. I love Ilya and we're good friends. You love it. Just some positivity hanging out At Davos, Demis said he thinks entry level jobs and internships might fall away due to AI sees consistency as the biggest problem with current agents. Advice for college graduates get incredibly proficient with these new tools. AI agents. And so, yeah, we were debating this earlier, like, the role of a junior engineer is going away, but is it possible that it just changes what makes a senior engineer that much better with an AI tool that they don't need anyone else who can potentially use AI tools as well? How much of this is something where, I mean, to go back to the Chinese young hiring boom, there's an element where I'm sure there's companies where you have older software engineers who are more resistant to adopt AI technology and you'd actually be better off with a young junior.

1:14:46

Here's a tough question for you, Tyler. What are you fantastic at? That AI is bad at.

1:16:19

Walking around being physical. Physically embodied.

1:16:31

Embodied. That's the last one.

1:16:33

So maybe that's like, for China, it's like, maybe the, you know, Silicon Valley has kind of found like, pederasty from first principles. Maybe it's just a similar thing over there.

1:16:35

Right.

1:16:44

There's actually a lot of value, I think, like, so generally, like, I'm kind of, I've always gone between like, is our jobs actually going to go away from this or are they just going to get like, more fake? Yeah, like Jordi and I were talking about this earlier where it's like, are there just going to be no junior jobs in like five years or are they just going to be like, you're just kind of. It's like, you know, you're in the office because, like, it's like people like having other people in the office and stuff like this. So I don't really know where I sit, but I think I'm probably more in the camp of like, you just don't need to hire new people.

1:16:45

Yeah.

1:17:18

Well, Ken Griffin. What did Ken Griffin say on a tear?

1:17:19

Okay, let me tell you about console consul builds AI agents that automate 70% of it. HR and finance support, giving employees instant resolution for access requests and password resets. Let's play.

1:17:24

So. So the post here. One up for the team. Citadel CEO says AI has re empowered technology departments in every business, but claims that 50% of entry level white caller jobs will disappear due to AI in 5 years is hype. He's throwing some cold water on Dario's talk track, but let's play the video.

1:17:37

Let's play the video. Are we getting audio? Where's Ken Griffin on this one? Is this the right one from Disclose tv? Citadel CEO says AI has re empowered technology. Is that the one, Jordy? Yeah, let's play it.

1:18:00

And we were, we are, I guess, in the market for an AI agent that can play videos on our.

1:18:20

Let me tell you about figma. Figma make isn't your average vibe coding tool. It lives in Figma so outputs look good, feel real, and stay connected to how teams build, create codeback prototypes. Fast 20s were an extraordinary period.

1:18:26

First of all, it's a pleasure to be on the global loom and doom panel.

1:18:38

Yeah, we're not.

1:18:42

We're not. 1920s were an extraordinary period. As I said at the beginning, it's.

1:18:43

Not preordained that it has to have.

1:18:48

The endnote of the 1920s, which was.

1:18:50

Of course, the Great Depression.

1:18:52

Andrew.

1:18:53

So let's take a step back and talk about where we are right here, right now. The area of recklessness is the spending of governments around the world, who are all, with little exception, all spending well beyond their means. That's the recklessness of this moment in history. This is not a parallel to the 1920s in terms of the recklessness of the, of the private capital markets. It's a story of the recklessness of government spending within the private sector. There's a huge question as to where I will take us. And I was carefully taking notes and listening to what Larry has to say or to what Madame Lagarde has to say, because this is one of the big issues of our moment. Will AI create the productivity acceleration that is honestly this hoped for in Washington and in the halls of government around the world as a ways to overcome the profit spending that we're currently engaged in? The world, the world needs a savior. And the hope is that AI is the savior that we need for productivity. And the challenge with this is, is it may or may not be. We just don't know yet. Now, there's a tremendous amount of hype around AI and in some sense the large AI companies need to, to create that hype, to raise the tens or actually hundreds of billions of dollars.

1:18:54

Hear that, Tyler?

1:20:31

That are going to the field, you wouldn't be able to raise hundreds of billions of dollars. We'll spend, and Larry can help correct me on this, but roughly $600 billion this year in capex for data centers the United States, I think it's a bit larger.

1:20:33

But does that mean that it's getting hyped up too much or it's just the hype is recorded, acquired as a, as a sales mechanism.

1:20:47

Go Ahead.

1:20:54

Larry's backing you up, Tyler. All right, let's play the next video, which was the one I was originally talking about, which is talking about job loss.

1:20:55

What AI has done is it has re empowered the head of technology in every business in the United States business, and it has pushed budgetary resources into the hands of, of the chief technology officer. So what you're seeing across American business is actually the impact of American businesses spending more on digitization writ large, of which AI is just one component. It's one component, but it's in some sense it's the COVID story that creates the space for American companies to really embrace technology in a much more profound way. I was in. I was in China about a year and a half ago with a group of prominent global executives, and they were talking about how AI was changing their business was so fun. And I said, let's go around the table and let's each share a story as to how we're using AI to make our business better. I heard five or six great stories. Not one involved generative AI.

1:21:08

Does that mean that the product is mogged?

1:22:17

So you're telling me a big company is inefficient in implementing technology that, you.

1:22:21

Know, as Dario Amadeus says, you know, half of all entry level white collar.

1:22:26

Jobs will, you know, be gone in.

1:22:30

The next five years. Is it that kind of change?

1:22:32

Or, you know, put yourself in his shoes. How, how much money does the AI community need to raise over the next five?

1:22:35

He's wrong about this, right? Demos doesn't spend the United States this year over half.

1:22:40

Google's doing this with cash flow over $500 billion.

1:22:44

Like that's true for every other lab lead except for Demos. You would expect to make a promise.

1:22:48

You'Re going to profoundly about Dario.

1:22:54

Oh, Dario. Okay, so is it hype? Yes, of course, of course, of course.

1:22:56

How else are you going to people to write $500 billion of checks just this year alone?

1:23:03

Right.

1:23:10

There needs to be a level of like, AI is your savior almost. And the question is, where will I land in productivity gains at the end of the day? In certain areas, we know it's going to be profound, whether it's call centers, whether it's helping to improve the productivity of software engineers. But in a number of white collar jobs. There was a recent Harvard paper on this. They called it AI Workslop, that it looks good, but if you sort of peel back the onion, the substance isn't there. I was with one of my colleagues who runs our commodities business, and they. He handed a report on that we were generated with an AI engine. Doesn't matter what the topic was. The first few sentences like, wow, that's, that's really insightful. And then you go down below that and it's all garbage.

1:23:10

So that's going to. And we are, we are running out of time. But that suggests that AI is not going to come.

1:23:57

Come to the rescue to make up.

1:24:02

For the various negatives that we discussed about in the first part of this conversation. So in 10 seconds.

1:24:03

So, but remember, spending on technology writ large is having a clear positive impact on the economy.

1:24:08

So which economy in the world benefits most from this conflagration of geopolitics, economic policy and technology change?

1:24:18

I mean, who benefits the most? That's a really interesting way to put it.

1:24:25

In 10 seconds.

1:24:33

In 10 seconds. You know, when it's all said and done, it's somewhat ironic, but it's, it's. The United States and China are the two big beneficiaries of this moment.

1:24:34

Yeah, so. So one thing, yeah, my. The way that I process this is as somebody who's not indexed to AI scaling and AI progress, but a lot of the most talented people he's ever worked with have ended up working in AI. They've gone from the east coast and gone over to the west coast and started working at the labs. And so while he's not working in a laboratory, I do feel like, you know, his. And you can argue like he's not AGI pilled. He clearly is not processing models developing like more advanced, like reasoning and agency and all these things that are going to.

1:24:50

We could be here next year and he could be like, oh, like I got a report from a commodities trader. It was AI generated. It was amazing. That's happened in so many other fields where people have said, oh, like I'll always be able to clock generative imagery. And then they look at the latest nanobanana Pro output and they're like, yeah, okay, actually, I can't. I have updated. That certainly happens. The interesting thing about his. So I like the point about the incentives of lab leads to make grand proclamations about progress in order to underwrite larger financing rounds. That's clearly true. That's clearly something that does happen. But explain Demis sounding exactly like Dario in terms of timelines and impact. Like a lot of their rhetoric is very, very similar. And Demis is in a position to just say, oh yeah, like we're going a lot slower because we're doing this off of cash flow and we don't need to raise and it's going to be longer. That actually might be to his benefit to sort of like pour cold water on all the competition that's out there trying to raise all this money. If he comes out and says, yeah, you don't need to lever up and build a trillion dollars worth of data centers because progress isn't going to be that fast, that would be probably beneficial to Google maybe. I don't know. But it certainly doesn't seem like he has to message that maybe to the Google shareholders somewhat, but not really. It's not really showing up in Google's financials yet.

1:25:33

It's not.

1:27:05

Not really changing.

1:27:05

Well, Demis might wait until anthropic and OpenAI get their S1s out and then decide, all right, it's time. It's actually open.

1:27:05

What do you think about that? As a counterpoint to Ken Griffin's CEO of Citadel, he's saying that all of the AI hype is about fundraising for these labs. Demis doesn't count there, right?

1:27:14

Yeah. I mean, doesn't he still have to convince Sundar to.

1:27:30

Yeah.

1:27:35

And the internal markets.

1:27:36

But I think that's probably true. But it's like he had the example of like, oh, he talked to business leaders and how are you improving your company with AI? And none of the examples were AI. You just know those are like, it's from some consulting firm that gave them the model that was like three years ago.

1:27:38

Yeah. They need some younger people at these companies that would be using.

1:27:52

Yeah.

1:27:58

Generative AI in transformative ways as opposed to. Yeah. I mean, it feels like those two things cannot be true. You cannot have a sclerotic, you know, large business that is failing to take advantage of AI and simultaneously not be hiring young people who are more AI native, it feels like you, you have to, you have to, you know, embrace one or the other.

1:27:58

Yeah. And he's discounting any concept of progress. Right. He's like, oh, when you read something, it's like total slop. Now that's exactly how code was like two years ago. Now there's software agents that can pretty autonomously build entire systems. That's certainly not unlikely to happen in just general writing.

1:28:25

Yeah. Yeah. Ken Griffin needs to vibe code a mobile game or something.

1:28:44

Yeah. I think calling out that saying, like labor replacement theories and pitches are like catnip for venture capitalists. They're doing white collar work and if they get the pitch that, you know, some technology is going to like, if you go to, if you went to a VC and you gave a really Compelling pitch of how your product is going to replace all VCs in the world. And it actually made some sense and you could kind of imagine how it play out. I'm sure you'd get a lot of investor interest in something like that, Right, because you're just trying to hedge. So it is, it is like the most compelling narrative. Narrative.

1:28:48

Yeah. FIN AI, the number one agent for customer service. If you want AI to handle your customer support, go to FIN AI. Let's click over to Jamie Dimon at the World Economic Forum in Davos. He's talking about raising taxes and special interest groups and the audience that makes.

1:29:26

Society work better for everybody.

1:29:45

And would you raise taxes to pay for that?

1:29:47

I don't think you'd have to. I think it would drive a lot of growth. I think I did the numbers at one point be $60 billion of spending. I think it would probably create more than that of growth and taxes, you know, and if you have to raise taxes a little bit, that's fine. But again, I don't want to have the binary argument. I don't know anyone. Okay. And you guys in the room, it'd be Democrats, Republicans, who thinks the spending that, that sending another trillion dollars to Washington D.C. will actually improve anything. So when you say raise taxes, if you said raise taxes and directly give it to the people need it, do it.

1:29:50

Of course. No, we are running out of time.

1:30:23

All these interest groups, you know, and they give it to their friends and all that. And which is why. Which is why the people are considered a swamp. It's kind of a swamp, you know, the 17,000 lobbying groups. But bank companies are guilty too. They're just fighting for their one self interest as opposed to what's good for my country. But, you know, that's what happens in Congress, you know, and you see these, how these bills get spent. Like the CHIPS act was a good idea until, you know, was it was had to be unions, place bays, child care, diapers, you know, what the hell are we doing? And then.

1:30:27

And we do it over and then.

1:30:58

We spend more money because the problem.

1:31:00

Is we just spent enough money.

1:31:03

Yeah, Jamie Dimon's great. Let me tell you about Sentry. Sentry shows developers what's broken and helps them fix it fast. That's why 150,000 organizations use it to keep their apps working. Demis, we've been talking about him and his blockbuster performance at Davos. He had a number of interviews, this one with Emily Chang, talking about a pause on AI and whether or not he supports it. A very revealing quote. Very, you know, says a lot about his timelines and his p. Doom in some ways. Let's watch this clip from Emily Chang interviewing Demis. To give regulation time to catch up.

1:31:04

To give society time to sort of.

1:31:47

Adjust to some of these changes. In a perfect world, you knew that.

1:31:49

Every other company would pause if every country would pause.

1:31:55

Would you advocate for that?

1:31:58

I think so. I mean, I've been on record saying what I'd like to see happen. It was always my dream of the kind of. The roadmap, at least I had when I started out DeepMind 15 years ago and started working on AI, you know, 25 years ago now, was that as we got close to this moment, this threshold moment of AGI arriving, we would maybe collaborate, you know, in a scientific way. I sometimes talk about setting up an international CERN equivalent for AI where all the best minds in the world would collaborate together and do the final steps in a very rigorous scientific way involving all of society, maybe philosophers and social scientists and economists as well as technologists, to kind of figure out what we want from this technology and how to utilize it in a. In a way that benefits all of humanity. And I think that's what's at stake. Unfortunately, it kind of needs international collaboration, though, because even if one company, or even one nation or even the west, decided to do that, it has no use unless the whole world agrees, at least on some kind of minimum standards. And, you know, international corporations, a little bit tricky at the moment. So that's going to have to change if we want to. To have that kind of rigorous scientific approach to the final steps to AGI pod.

1:31:59

Friendly guy in the YouTube chat says Damas is the Steve job of AI.

1:33:21

You're wearing the TVP and turtleneck today. Maybe we need to make a turtleneck and award it to the who we crown the Steve Jobs. No, he's a fantastic communicator. Demis is inspiring, balanced, measured, knowledgeable, just not. He's not completely accelerationist at all costs, but also not completely doomerist. I don't know. It's a really great take. I don't know. What do you think, Tyler?

1:33:26

Yeah, I think there's maybe an interesting way to look at this. Like the question of pausing where. So if you take. Dario says that he's not competing with DeepMind or OpenAI because he's an enterprise. They're a consumer.

1:33:51

Yeah. Wait, wait. Sorry, sorry. Clarify who's who there.

1:34:04

So Dario.

1:34:08

Dario at anthropic. Anthropic is enterprise is saying we're not competing with DeepMind because that's a consumer company. Is that what he said?

1:34:08

Yeah.

1:34:14

And OpenAI. That's what he said yesterday. Like broadly, he was like, we don't have to monetize billion users, whatever.

1:34:15

He's like, just ignore those companies and their enterprise businesses. They don't matter.

1:34:19

Yeah, but so if you take that and then.

1:34:23

Sure.

1:34:25

So that means, you know, there's some way in which, you know, DeepMind's real only competitors. OpenAI. We were just talking about how DeepMind doesn't have to raise money. Right. They're funded by cash flow. So basically there's a way to look at it where you can either pause for safety reasons so we can catch up with regulation or whatever, or it's just because if we basically pause all AI, it's actually fine. Philanthropic. Right. Because they don't have a super direct competitor.

1:34:26

Right.

1:34:52

OpenAI is trying to get more into enterprise. Maybe DeepMind is as well too, but they have fairly strong position perhaps. And then it's going to hurt OpenAI a lot.

1:34:52

Right.

1:35:01

They need to raise a lot of money to compete with DeepMind. So if you can basically make this like almost cartel looking thing group with Anthropic and DeepMind, you can basically try to take out OpenAI.

1:35:01

No, it's kind of like, I mean.

1:35:14

That'S a cynical way to look at it.

1:35:16

Yeah, yeah, yeah, yeah, yeah, I hear you. The other interesting thing is that he says he would like maybe potentially support a pause, but then what he describes is not, oh, all the researchers just focus on productization of the current models, like stop training, stop doing reinforcement learning, stop optimizing models and instead like let's build more great SaaS. Like that's not what he's saying. He's saying like, let's all still do.

1:35:18

Research to focus on ads and AI progress.

1:35:45

That would fire me up. But that's not what he's saying. What he's saying is like, get all the research together and in cern and, and CERN is not pausing scientific research. If anything, CERN is a way to accelerate AI research, but in a collaborative way. So he's really just calling for more collaboration and a pause on competition or a pause. And I mean, I feel like everyone in the scientific community, in the research community definitely laments the days of yore where there were more open research papers, more sharing of knowledge. It wasn't all constantly just backdooring through hiring and poaching. Some of it was just open. You could just download the Transformer Paper.

1:35:49

Yeah.

1:36:28

But even in that case, DeepMind is still on the defensive because OpenAI is trying to take. Google is like, oh, search is dying. Right. OpenAI is trying to take it away. If you can basically get everyone to be like, oh, we're going to collaborate, everyone's going to share the research, that's probably better for people.

1:36:29

Yeah. So you would never expect the leader to say, hey, let's pause the race.

1:36:48

You wouldn't expect OpenAI to pause the race because they're the ones who have so much more to gain from attacking the other.

1:36:52

Just continuing pushing further. Yeah. Okay. I mean, depends on what you mean by pause. Because I feel like if there was like a model freeze and you only have Opus 4.5 GPT, 5.2 Gemini 3 Pro, you have those models, like OpenAI could still compound on the DAU front, right?

1:36:57

Yeah. I mean, yeah, there's a bunch of ways it could actually go down, but there's different ways to look at it, I think, rather than just pure safety.

1:37:21

Yeah. There's some extra context here from Andrew Curran. He says Darius said the same thing during the day after AGI discussion this morning. They were both asked for their timelines. Demis said five years, Darius said two. Later in the discussion, Darius said that if he had the option to slow things down, he would because it would give us more time to absorb all the changes. He said that if Anthropic and DeepMind were the only two groups in the race, he would meet with themis right now and agree to slow down. But there is no cooperation or coordination between all the different groups involved. So no one can agree on anything. And it's very funny thinking about this cern, like bring all the researchers together in the context of the craziness of OpenAI. The craziness, the thinking machines, all the interpersonal aspects, the human element of AI progress that you get everyone in and it's like, okay, we need to pick a leader. Okay, who's in charge? Who decides what do all these people get along? Because some of them have some pretty crazy bad blood. He's like, hey, you kicked me out of this lab and then you quit when I thought you were going to be my co founder and then you were having this relationship with this person. Creating harmony in the barnyard might not be the easiest thing.

1:37:27

Interesting point from Deepak in the chat. No, OpenAI@davos?

1:38:38

There is. So I believe Sarah Fryer did an interview there. The interviews are on like a delay. A lot of them happen at 10pm Davos time. Also the first day of the conference was yesterday, so we will be getting more Davos content today. I did feel that though. But a lot of it's just about when different segments happened. So I mean, amazing move by Dario and Demis to come out with absolute heater on day one very early, establish some ground game, some ball control because Satya Nadella is there. He's going on the all in podcast and he's already given quotes and there's some clips from him, but we haven't seen him. He hasn't been aura farming Microsoft, hasn't been aura farming Davos as effectively as Google and Anthropic. So, you know, little bit of tale of the tape there.

1:38:44

Well, let's Skip ahead to 11 labs.

1:39:35

Build intelligent real time conversational agents. Reimagine human technology interaction with ElevenLabs. And I believe the CEO of Eleven Labs is there at Davos. But what do you want to move on to?

1:39:38

Jordy?

1:39:48

This is, I mean this was probably the story that got reported on the least out of Davos over the last 24 hours, but is arguably the most important, important, the most expensive car in the world. Everyone wants to know. Of course it is the car that Jensen bought his parents. So let's pull up this clip. I love it.

1:39:48

This is great. We have many more years of that journey ahead of us.

1:40:08

Thank you.

1:40:13

I appreciate that. My only regret was at the ipo. After the ipo, I wanted to buy my parents something nice. And so I sold Nvidia stock at a valuation of $300 million. The company was at a valuation $300 million and I bought them a Mercedes S Class. It is the most expensive car in the world. Yeah. They regret it. Do they still have it? Oh sure, yeah, they still have it. Yeah.

1:40:14

Good.

1:40:46

Jensen, congratulations on that journey. Yeah. What is 5?

1:40:47

So it's currently a $4.5 trillion. Four and a half and an S Class. When Trillion. They went public in 1999.

1:40:51

So what was NASCAR cost like 60k something back then? I don't know, 80k.

1:41:04

How much? What did I mean, you're talking about.

1:41:08

A maybe a billion dollar car. I don't know if I have the.

1:41:11

Math right there, but it seems like, yeah, around 65k.

1:41:16

It seems like the stock might be up to 15,000x. I don't know. It's definitely up 1,000x.

1:41:20

Right.

1:41:28

Because it's 300 million. 300 billion and then you get another zero. So 10,000x. Wow, that is brutal. Brutal. Worse than the TV I bought with a Bitcoin.

1:41:28

Hey, worth it. Worth it.

1:41:39

I bought a tv with a bitcoin. Terrible. Not even 4k things in the trash now. It was like $1000 back then. It was ridiculous anyway. Cisco. The Davos of Cisco. Cisco AI Summit. It's February 3rd. It brings together leaders from Nvidia, OpenAI, AWS and more to discuss the future of the AI economy. The whole thing will be live streamed and will be there for a giga stream. Mark your calendars. February 3rd, Cisco AI Summit.

1:41:40

Many of you there were very excited.

1:42:10

President Trump had some comments in Davos. Alex Heath broke the story, took it to the timeline. Alex Heath says a quote from Trump. Mark Zuckerberg showed me a plant where he put it over map the size of Manhattan, miles long, miles wide. It literally covered most of the island of Manhattan. It's called a big plant. I think he's talking about a data center, but I like that he's calling a plant that is a big plant, a power plant, essentially a plant of things. And President Trump also said that the stock market dip is peanuts. That the stock market is going to be doubled and boring business quotes it and says the President is literally telling you that the stock market will double from here and you are bearish. It's a crazy, crazy line. What else did Trump say here? He said our stock market dip is peanuts. The Dow is going to hit 50,000 and double to 100,000 in a relatively short period of time. Anyway, let me tell you about Applovin. Profitable advertising made Easy with Axon AI. Get access to over 1 billion daily active users and grow your business today.

1:42:12

Someone should try to follow up and figure out, ask the President, are you giving financial advice and what's driving it?

1:43:23

Are multiples going to double or is our earnings going to double or are we going to see a little bit of earnings growth and then a little bit of multiple expansion? What's driving the valuation increase? I'd love to know that. Let's get to the bottom of it.

1:43:31

Semaphore made a print newspaper for Davos.

1:43:45

I mean, first we need to confirm that Philip Johnston, the founder of Star Cloud, former guest on the show, he is there at Davos. He's dropping takeaways, he's asking questions. He was the only person that got to ask Demis a question. Right. That's crazy.

1:43:50

I mean, makes you ask some questions.

1:44:08

It does, it does, yeah. Yeah. I mean, what a run from Philip Johnston. But it's been fun seeing like the on the ground. Obviously there's a lot of polished Clips that are going out from the media organizations. But he's doing citizen journalism over there and we thank him for his service. Bringing clips to the timeline saying that Jensen Huang announced that we need more energy. And also he broke the story, at least to me, that Jensen's parents own the most expensive Mercedes in the world. I wonder if they still have it. Anyway, let's go to 704. First, let me tell you about Gemini 3 Pro, Google's most intelligent model yet. State of the art reasoning, next level vibe coding and deep multimodal understanding. I like that sound cue. It's a good one.

1:44:09

It's the best.

1:44:52

Anyway, Reid over at Semaphore says it's fun to work for a print newspaper again. Although I missed the print deadline for some changes which brought back some fond Wall Street Journal memories and traumas.

1:44:53

We really missed our window to do the TVPN print newspaper. Somebody sent us a fully built out TVPN print newspaper. It looked absolutely fantastic. Last year we wanted to roll out at least a weekly, maybe daily print newsletter, but couldn't get around to it.

1:45:05

This is a very interesting lead too. I can't find this actual article. I'll have to find a digital version of this. But he says techies ignore the geopolitics on their Davos home turf. Quote, this reminds me of Burning man, but not how you think. I'm like just zooming in to try and read this. AI has finally been eclipsed by geopolitics. At Davos, there are essentially two conferences happening. One involves Greenland, Venezuela and the rewriting of the world order. And the other is really just a tech conference centered around selling and deal making. That's the one I want to go to. Yeah, we were texting with a journalist who is writing a story about tech's involvement in Davos. And my point was like, it does feel like Davos is coming back and the tech companies are there. Although from the outside I think a lot of people, people think of Davos purely as a meeting of world leaders and geopolitical and they're following the Greenland story and they're very much looking at the front page of the Wall Street Journal. But if you go to the front page of X and the timeline, it's all Damas and Dario clips and Ken Griffin and a few others. But mostly like the stuff that I've been seeing has been focused very much so on tech. But there is a whole different story going on all around Greenland and whatnot. But we got to talk about AI timelines. It's More interesting to me. I'm just so excited.

1:45:23

Should we talk about why Elon is racing to take SpaceX public?

1:46:41

Absolutely. Absolutely. But first, speaking of public, public.com, investing for those who take it seriously, they got stocks, options, bonds, crypto, treasuries and more with great customer service. So why is Elon racing to take SpaceX public?

1:46:45

Putting data centers in space Push the bean air toward an IPO, sources say. According to the Wall Street Journal, SpaceX resisted going public for years. Then came the rise of artificial intelligence. Elon's rocket maker became one of the country's most valuable private companies due in part to its ability to develop risky space businesses outside the scrutiny of public investors. Its executives like to say the company wouldn't IPO until its rockets were literally regularly flying to Mars. That was before the rush to build data centers for AI computing prompted Musk, Jeff Bezos and others to propose putting them in space. The idea has prompted skepticism from many engineers given the technical challenges poised by building solar powered AI data centers that zip around the earth. But it has continued to gain traction, and Musk has become obsessed with the idea of SpaceX being the first to do it. Such a feat would be hard to attempt without billions of dollars in capital an IPO could deliver in one fell swoop. The billionaire it's so funny to just refer to Elon as like just a billionaire. The billionaire also sees the SpaceX IPO.

1:47:01

Correction like two months.

1:48:04

Yeah, two months. Once he's a trillionaire.

1:48:05

Teen Air.

1:48:07

Teen Air. Teen Air is a new I'm excited to be throwing that around. Sees a SpaceX IPO as a way to help his AI company, Xai catch up to rivals. Musk has a long running rivalry with OpenAI CEO Sam Altman, who last year explored buying a rocket company to deploy satellites with AI computing capabilities into space. Two of XAI's competitors, OpenAI Anthropic, are eyeing their own IPOs this year, and Musk seems eager for SpaceX to hit the public market first. SpaceX is expected to select banks to lead the stock offering soon. Musk has told people he wants to complete the IPO by July. He wants to get it done so he can have a nice summer.

1:48:09

Philip Johnson needs to spac star cloud like today. He spacs it it's going to 20 billion like overnight. Like it's going to be the biggest meme stock Anytime that there's a, there's a like a public sparked comp to an Elon company that's like private. It's Just like everyone wants it because they want the. They're like, well, if the Elon thing works like this one will take 2% of the market. So maths out certainly be able to raise a lot of money.

1:48:48

Musk's apparent change of tune on SpaceX IPO plans in the middle of last year surprised many. The beaner, soon to be Teen Air, has loudly complained about running his existing public company, Tesla, and is repeatedly tangled with regulators in the courts over issues like his compensation.

1:49:18

Everyone thought like Elon ran the biggest AB test on what it's like to run a large cap company in the public markets and the private markets. He had a terrible experience in the public markets with Tesla getting sued and dealing with all that, dealing with complicated comp packages. And so it felt like he would definitely prefer to be private all the time. He wanted to take Tesla private, remember at Funding secured, Funding secured. He's never been a fan of the public markets, but he has changed his tune.

1:49:31

So anyways, building and launching thousands of satellites would be technically demanding, costly. And SpaceX officials quickly decided the easiest way to raise the tens of billions of dollars it required was to dip into the newly thawed US market for IPOs. Meanwhile, Xai was trailing rivals including OpenAI and Google's Gemini by key metrics such as revenue and user base. A SpaceX IPO was seen by some of its investors as a potential cash cow that could supercharge Xai's growth and in turn help SpaceX. If SpaceX succeeds with putting data centers in space, its investors expect XAI to become a customer. Some investors believe SpaceX could purchase a percentage of XAI. Or Musk, who owns more than 40% of SpaceX, could tap that stake to invest in his other companies, including xai. Plus, a publicly traded stock creates a capital safety net of sorts and would be reassuring to investors. So anyways, the two companies have existing ties.

1:50:01

SpaceX has invested 2 billion in XAI as part of 5 billion financing and then XAI raised a bunch more money. So very exciting. Anyway, CrowdStrike, your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. And we have our first guest in the TVP Ultra, Jan Srimak, the founder and CEO of California Forever. Good to have you on the show again. Welcome back. How are you doing? Sorry, let me get my diet.

1:50:57

How are you?

1:51:29

How's the, how's the new year going? How is your break? How are the holidays?

1:51:30

We have three young kids under four and we dragged them nine hours. We dragged Them Nine hours to see my mom and dad in the Czech Republic. Okay. Nine. Nine hours time with their friends there and then nine hours back.

1:51:34

Yeah.

1:51:44

So you can imagine how my Christmas was more.

1:51:45

More like. Like a day of travel with young children is. Is more stressful in many ways than 99% of work days.

1:51:48

I went to Europe over the break. I went to Austria, but didn't bring the kids. So it was very.

1:51:58

It's a big difference. My new pet peeve is single people, couples on airplanes who are complaining about the long flight. I'm like, really?

1:52:04

Oh, shit.

1:52:11

You have to sit there for 12 hours and drink wine and watch Netflix. That was terrible.

1:52:11

Well, I used to think when a kid was crying, it meant the parent was doing something. Something bad. So, as you know, I'd be on a flight and then kids crying, and I'm like, wow, this parent is just, like, being abusive or whatever. Why is it. Why is the kid so upset?

1:52:15

Yeah.

1:52:29

Turns out kids will cry for all sorts of reasons. For a million reasons.

1:52:29

Yeah. Anyway, give us the update. How are things going? Just sort of reset for us on the project. How long have you been working in California? Forever. Where are you in the process?

1:52:32

I've been working on this for almost a decade.

1:52:44

Almost.

1:52:46

Almost a decade. So 2016 started in 2017. There's about seven years of land assembly, and then we introduced it publicly in about two years ago.

1:52:46

Were you full time back in 2017?

1:52:58

Yeah, I was full time. I was the. I was the only employee for seven and a half years.

1:53:00

Okay.

1:53:06

And obviously confidentiality was very important for us. And so that meant. Is when I say I was the only employee, I meant I did everything. We raised a billion dollars in that period. And I did everything from bookkeeping to being everything else in the company. I literally did.

1:53:06

And the confidentiality was that in part because if people knew what you were doing, they would come in and start trying to bid against you on land stuff. And realizing he's already.

1:53:18

This plot, so he's willing to pay more. And that is the game theory.

1:53:27

Yeah.

1:53:30

And it always happens. I mean, even if you have university, if you have public universities buying land in a city to build a building, they don't come in. And like, we are MIT and we would like to buy this plot of.

1:53:31

Land, use some kind of companies.

1:53:41

That was why.

1:53:42

So what was the. What was the structure of that fundraising? Because you probably had to draw a salary. Is it like a search fund where a venture capitalist might raise some money, pull some management fees to fund their salary and Operational overhead. And then, like, when there's a deal to be done, there's a capital. There's a capital commitment that happens. A transfer. Is that what it's like?

1:53:43

We structured it as a partnership that raised money over time at different valuations as we made progress. And then I just took salary like any other five. It was pretty similar to a normal tech company. There's some nuance because it's real estate and because of just the sheer amount of capital that we needed and how long it took to do it. But other than that, pretty standard.

1:54:04

Yeah. And what was the first pitch? What was the first plot of land that you bought? What was that negotiation, Mike?

1:54:24

Well, the first few were. The first few are listed. So when we came to the market, there were four or five properties that just been listed and sitting on the market for two years or something.

1:54:30

Was there anyone else thinking about doing anything in this particular region?

1:54:41

I don't know of anyone else. I mean, not right now. But the great irony of the situation is lots of people thought this was a really good idea in 1950.

1:54:46

Oh, yeah.

1:54:56

Seriously?

1:54:57

Yeah. Yeah. You showed me some of these. There is this.

1:54:58

In 1958, the Federal Commerce Department and the Army Corps of Engineers got together. This is before almost before computers. And they wrote this thick binder. They call it the Future development of the San Francisco Bay Area, 1958 through 2020. And decade by decade, they planned where growth should go in the Bay Area. And you open these maps, and they did it decade by decade. And you look at the first one, and Santa Clara is still orchards and so on. Then they correctly predicted this explosion in places like Pleasanton and Dublin in the east bay in the 80s and in the 90s. But the part of Solano where we're building remained open. And then you flip open the 2010 map, and where I thought there should be a city, they had a city.

1:55:01

There you go.

1:55:47

And I was sitting there in 2015. I'm like, okay, maybe this isn't just Jan's insane idea. Maybe some smart people thought about it. And then you look at other regional planning documents in the Bay Area, something called the association of Bay Area Governments, which is the regional planning body, had a plan called 1972, 1970-1990. And in the plan, they literally said, by the end of the 90s, we are going to be running out of space in the Bay and we're going to need to build some new cities. Otherwise they region is going to become unaffordable. And there are some areas where you could Build cities that are really good for it. And the part of the Bay that's the best place to do it, that is five times bigger than anywhere else is the area of which we've acquired about 90% over the last 10 years. So this is a very old idea that we've dusted off. And obviously we're building something different that they would have built there in the 50s, but lots of routes in regional planning.

1:55:48

So in terms of development, it feels like, you know, AI has been the megatrend over the last few years. Most of the building that's happening around AI is happening in Abilene, Texas and just wherever the natural gas plant lines are and sort of all over America in pretty low population areas. No one's really thinking that, oh, you know, Meta needs a data center right next to their Menlo park campus. But the AI Megatrend also might drive up housing prices in San Francisco a ton. A lot of people are predicting this based on the IPO schedule. If OpenAI anthropic get out, there's a trillion plus in liquidity going loosely into the San Francisco housing market. You could see some really crazy dislocations in that market. Is that something you're tracking as an accelerant of your plans?

1:56:39

Yeah, 100%.

1:57:33

Okay.

1:57:34

And I think it's. And I think it's, it's. Honestly, it's more from the policy perspective, because the last time, I mean, this has happened in the Bay Area periodically for the last 20 years. The problem is we haven't done anything about it in 50 years and every time it gets worse and then there's a bit of a pullback and like, oh, it's going to be better and then it just comes back stronger than before.

1:57:34

Right.

1:57:53

And.com boom, then social crypto, now AI. And so from a policy perspective, the argument that we've made is we need to get ahead of this because if we don't, this is going to be way worse than what we saw in 2017, 2018, and when it last happened in 2017, 2018. The problem is tech is going to get blamed for this.

1:57:53

Oh, yeah, of course.

1:58:15

It'S wrong and it makes no sense because you just create all of these really well paid jobs, but you're going to get blamed for it or we going to get blamed for it. And so we as an industry need to get ahead of it. And you saw attempts at this in 2017 and 2018, and Google and Apple and Facebook and I think Amazon up in Seattle all made these 501 billion $2 billion commitments to go and build affordable housing. But what happened when you look at the numbers is they would maybe build a few hundred units or a few thousand units because building an existing cities is just so time consuming, so expensive. It costs $1 million to build a small two bed affordable apartment in San Jose. Yeah, a million dollars. And it takes 11 years of planning. Whereas our.

1:58:18

Affordable housing for millionaires.

1:59:02

Yeah, exactly. Heavily subsidized by the taxpayer. Whereas our argument is if you create a place where you can build, you can get the private capital markets to build it. And we are proposing to build 174,000 homes. For context, 174,000 homes is more homes than San Francisco has built since 1950. Since 1950. And sure, we should build a lot more housing in San Francisco. The idea that you can solve the Bay Area housing problem, the trends are the trends in San Francisco or in Menlo park is insane. No city in the world has ever solved housing affordability issue without spreading out a little bit. And then the question is, do you spread out just because by spreading out with sprawl or do you spread out and actually thoughtfully plan it and be able to build something really magical?

1:59:03

What is the number of homes that the California government has projected are needed to sort of solve the housing crisis?

1:59:52

Three million.

2:00:01

Three million, yeah. Just in California.

2:00:02

Just in California, yeah. And the way that they calculate wires.

2:00:04

So like in some ways you could say like, okay, 174,000 and like it's starting to make a dent. It was less than 10%, but it shows that you need projects at this scale to ever solve it. Because you're not going to solve this problem by like, oh yeah, there's an empty lot behind this in this neighborhood. Let's put up eight units here. Right. It just doesn't. That really doesn't make it down.

2:00:07

Right.

2:00:28

Yeah, I mean, I think that's going to be a part of the solution. But you'll need, I think California will need multiple California forevers. And our hope is that by blazing the path and showing how you can, how you can finance it, how you can buy it, how you can work with the community to get the buy in and to convince people that you actually care about it and you want to do it right and you want to do it well, we make it easier for others to do it. I will say we do have what I think is by far, it's a very unique site. The idea that you can put together, I mean, we own 70,000 acres at this point. It's 110 square miles. It's two and a half times bigger than San Francisco. It's five times bigger than Manhattan. It's bigger than Washington D.C. by one and a half times. And so the idea that you can find that, by the way, I drove from my meeting in downtown San Francisco to the site in 55 minutes the other day. Now, that's not the case, depending on how you time it with traffic, but it's pretty close. So it's going to be hard to do it at this scale, but we certainly should be finding 500 acres at the edge of or a thousand acres at the age of Sacramento and all different parts of LA and so on, and building it at medium density and walkable and some industry and jobs. But I think In California it's 3 million units, it's about 2 million in Southern California, it's about a million in Northern California. And so we come close to solving 20% of it. And that's just one of our projects. That's the expansion of a city called Suisun. We also looking at doing a second project on lands that we own on the west side, sorry, on the east side of our property around a city called Rio Vista. That could add tens of thousands of additional units. So we get close to solving 20% of Northern California's deficit, which is a big deal.

2:00:29

Yeah. Talk to me about the transportation issue. 55 minutes when there's not that much traffic, could be more.

2:02:08

As an LA resident, I'm thinking, yeah, it's easy. That's my daily commute.

2:02:14

Yeah.

2:02:21

So.

2:02:22

But yeah. What has the Bay Area done recently? I know the bridge, the Oakland Bridge was improved, but what's on the table in terms of like, toll roads, congestion pricing, building new bridges, adding ferries? Like, are there lessons to be learned from Manhattan and how they've solved things? I've seen people be very pro congestion pricing, very anti congestion pricing. All over the place, place. What's your vision for what you would advocate for in terms of modernizing transportation in the Bay Area?

2:02:22

It's a great question and there's so many layers to it. I'll try to keep it brief. I'll give a slightly nerdy answer. The first step in any kind of transportation planning is land use planning. Where do you put things and how do you lay them out to kind of eliminate the needs to go that much into first place?

2:02:57

Okay, yeah, you don't want to have to go to San Francisco just to get a burger. There should be a burger joint locally. And same thing for, you know, all the different amenities and jobs in particular.

2:03:15

Because in Northern California we have this really, really dumb distribution of jobs where the Northern California mega region increasingly works as one place.

2:03:26

Yeah.

2:03:34

From Sacramento all the way to San Jose. I mean, you have people who live in Sacramento, they work in the bay two days a week. They do the commute. Right.

2:03:34

There's a ton of people that live north of the Golden Gate Bridge and commute to San Francisco. And then there's a ton of people that are really far south and commute up to the tech companies, whatnot.

2:03:41

Exactly. And right now the challenge is basically all of the jobs are between San Francisco and San Jose and then up in Sacramento. And then you have 4 million people in places like East Contra Costa county and Solano and Stockton and San Joaquin. There's basically no jobs.

2:03:49

Yeah.

2:04:03

So what happens in the morning is all of those people get in a car and they have a two hour commute to Menlo Park. The number of people who live in Solano county who work in Menlo park and have a two hour commute each way every day would blow your socks off. It's half. More than half of the county commute outside of the county. So it's completely insane. So the first solution, and this is actually our big contribution to the region, is to take a bunch of jobs and put them in the middle of the region, which makes life better for everyone who lives in Palo Alto, San Francisco, because you don't have all of those people sitting on the 101 or.

2:04:03

So big office buildings for tech companies or big companies, big employers, big manufacturing employees, big manufacturers. Have you talked to companies? Are they receptive to that? Because there's sort of like the first mover is going to have a rougher time, I imagine, than if they're already, hey, we're already in Menlo park. And most of our employees are right here. At least the executives are. Yeah, there's a lot of people that can commute, but maybe 50% of the employees only commute less than 30 minutes. That's probably a pretty common stat for a tech company in the Bay. How are you thinking about pitching this to companies to go and take that leap and be the first one we start with?

2:04:38

There has to be a reason for them to be in Solano and generally.

2:05:16

The ones not just pricing.

2:05:19

Not just pricing. And the biggest one that we found, two big ones. The first one is manufacturing.

2:05:20

Sure.

2:05:26

You have this whole wave of companies doing manufacturing. A bunch of them are headquartered in the Bay Area. Yeah. And then I think the sweet spot Is you are a company, you figure out a new way to do x. You've raised 20 million, you have 20 people, you've perfected it now you've raised 100 million. And you need to build a factory. And the factory is going to have 100 or 200 employees and you want to build it somewhere. Right. Now what happens for virtually every single company, and we've spoken to 400 of them, is they say we really want to build it in California because we still figuring out how the factory works.

2:05:27

Yeah, the factory is the product.

2:05:57

The factory is the product. And if every time that our software engineers or AI engineers need to go to the factory, they need to go on a plane and fly to another state, that's a three day trip. Yeah, it's like, oh, it's three hours to go to Austin. No, it's not three hours. Yeah, yeah, it's an hour and a half to the airport, then it's tsa, then it's late, then you fly in, then it's to get there, then. So it's a three day trip.

2:05:58

Right.

2:06:19

And our value prop is put the factory in Solano. Because if you get in a car in Menlo park in the morning, you can be in the factory of 80 minutes and you can be back home and have dinner with your kids. And that really works. Now what means, what that means is if you put the first factory there, then you go and you raise $500 million and you need to build factory number two. Our value prop is we have so much land that if you want to build factory one, which is 100,000 square feet, we can reserve enough land around it for five or 10 years that you can put factory two, three, four, five next to it and you can have them all in the same place instead of having to solve this problem over and over and over again in a different place. Now that means that we get the factory that employs 500 or 1,000 people. Now, if you want to build a gigafactory that employs 10,000 people, where you are really, really cost sensitive, maybe that still goes to Ohio, but that's okay. But a lot of the cutting edge stuff would be in Solano. And to your point about AI, that's becoming even more important.

2:06:20

We already see this down in la, by the way. There's a lot of startups that start in El Segundo and they have, you know, 5,000 square feet. They're doing some R and D and then once they start building larger factories, they move just outside of El Segundo. Just to drive. They might still have, you know, SpaceX was started in El Segundo, but eventually move to Hawthorne and grew out of that. And then eventually Starbase in Texas. And so for that interim step in the San Francisco hard tech, deep tech community that can make a ton.

2:07:17

They don't have that. Exactly. San Francisco doesn't have. And then. And even in la, at least when we talk to a bunch of those companies, Louisiana is becoming a kind of. El Segundo and Hawthorne are becoming. And Torrens are becoming victims of their own success.

2:07:44

Oh sure.

2:07:56

Because they are fully built out. There's no available land. So a bunch of these companies will tell you we can't really build another factory here. And we still have R and D in the Bay. So it might be an even outlet for companies that started out here. They have a factory here, they have a bunch of talent in the Bay. And then to your point about AI, everyone talks about building the factories of the future. How are you going to build the factories of the future? You're going to build them with AI talent. Where is 85% of the market cap in AI in the Bay Area? Do you think you're going to convince those engineers easily to spend every week flying to some other state? It's a real comfortable advice.

2:07:57

Hey, I know you can go work at Meta for $100 million a year, but how about if you work for less than that in Utah?

2:08:35

Yeah.

2:08:41

And then you fly all the time.

2:08:42

What does success look like for this year? One thing that I just appreciate so much about this project and this company is that most people come up with an idea like to make a new city or make, you know, and they start running the calculations and they realize, okay, not only is this going to be hard, but it's going to take 10 plus years. So I'm not even going to start right. Because I want to be doing something where maybe I can see more tangible progress on a faster timeline. Or I need a billion users like two years from now.

2:08:44

Not I need 100 million ARR today. The bar is three months now.

2:09:16

So you've just chunked it out just like 10 years in, which is probably the moment where it really is hopefully more exciting than ever. What does this year look like in terms of just pushing the ball forward?

2:09:22

Well, hopefully we'll break ground this year. And so that would have seemed ambitious a year ago, I think, but I think there's a few things that kind of aligned. And so if I just. There's this really interesting thing that's happened Where a lot of things have come together at the same time for us, where we've announced it, what is it? Two years ago. And of course it was very controversial. Right. You come to a community and it's a guy with a funny accent and we want to build a major new city which we haven't done in America in 100 years. So nobody even thinks that you can build new cities. And it's like, oh, it's not just going to be a city, it's going to be a massive manufacturing process. And by the way, we'll build the biggest shipyard in America on the south end of the property. And everyone is like, you guys are crazy and we don't trust you. But what's happened over the last few years is we've actually done the work, we've put together this incredible team and so we've built a lot of trust and now there's a tremendous amount of support for it in the community. And then what's happened at the same time at the state level and at a national level. I mean think back to where we were as a country two years ago. Abundance wasn't really a thing, build wasn't really a thing. We didn't just have an election on affordability. Re industrialization was kind of beginning. And so at a macro level there's been this massive shift towards we need to do this. And that exists in California. And so right now where we are is we had a big announcement this morning that we can talk about but basically the, the goal for this year is the stretch goal for this year, I'll say that is to break ground. And there's lots of reasons for why that makes sense for Solano county, it makes sense for California, it makes sense for the country.

2:09:40

Sorry, going back, closing the loop on congestion pricing.

2:11:22

Yes.

2:11:25

Sorry, are you pro con? What do you think?

2:11:26

I think it is, I think the, the direction that I'm most excited about is we should use congestion pricing to build a lot more capacity. I think it's really politically complicated to come and say we are going to take this existing road that right now is free and make it paid.

2:11:29

Yeah.

2:11:49

But I think people want a lot more capacity and paying for it with congestion pricing is great. New York has proven that. It, I mean I lived in London when they introduced congestion pricing. It clearly worked. We've known that this works for a long time. It was a big fight to do it in New York. I think that self driving cars are going to force us to do it because there's an induced demand with self driving cars, when the cost of driving a car is just the financial cost, but you can be watching Netflix or doing email or whatever, people are going to drive more, we're going to send packages back and forth.

2:11:49

If you believe the self driving car car narrative, you have to be long suburbs.

2:12:26

You have to be long suburbs. And then you have to believe that it's going to force us to price at least some of the roads and some of the capacity. Because without it, it's induced demand and there's just going to be more and more driving. Right. Because if the cost of your doordash Delivery drops by 70% because you don't need to have the driver, you're going to get more, you're going to get more doordash.

2:12:30

And yes, there's some intelligent routing that you can do exactly have multiple people in the same car, but that's what congestion pricing incentivizes.

2:12:50

I think the political opening is to say we're going to price these roads, we're going to use part of that revenue or all of the revenue to fund the construction of them. And that also I think is a way to get it through politically. Because in California you have a environmental interest group that doesn't really want to build highways, but if you price those highways, they are much more open to it. And so I think that's the grand bargain that you can drive. The other thing that I would say is what we hope the city will do at the scale of the mega region is if you look at the California mega region, from San Jose to Sacramento, it's kind of like a donut. In the middle of the donut is all of the property that we own. And part of the issue we have is that this used to be two regions. It used to be the Bay Area and Sacramento. And they've become one and they've grown and they've become what would be. I mean, if the Bay Area was a state, if Northern California was a state, there's 12 million people who live there, it would be a seventh largest state in America. And right now we haven't really tied it together in terms of transportation infrastructure the way that we need to. That's why when you try to go from Palo Alto to Tahoe on a Friday night, it's such a disaster. And so we need to build more capacity there. And putting a major new city in the heart of the region is how you help pay for some of that.

2:12:58

What are you seeing on the. Are you tracking the rebuild of the Palisades in Altadena in la. Obviously we experienced the fires down here. There's a huge amount of construction. The hope is that it's really quick construction. We've already talked to a friend who's I think finishing rebuilding this summer, which is pretty crazy, pretty fast. Are there any lessons learnings just supply shocks to demand for building labor? Are assets actually getting reallocated from Northern California to Southern California to enable the rebuild? Or is that something that's not really changing your plans at all? Are there any lessons that you've learned from that?

2:14:17

Yeah, I mean, I'm not deeply involved in the rebuild. We're far away from it. But my, from my vantage point, we are still at a point where permitting and coordination is the bottleneck. It's not the labor, it's not the construction. It might become that I think what it illustrates is just the complexity of building infill in general. Just building an existing cities because you have thousands of people who own the land and they need to all agree on construction schedules and redoing it and so on. I will say that there's some pretty impressive success stories. I'll give a shout out. I mean Alexis Rivascover has been doing great work in terms of both documenting the permitting challenges and then I think that they delivered several units already. Yep. And so I'm a huge fan of what they're doing. I think the product is beautiful.

2:14:59

Yeah, it's a factory built ADU company.

2:15:44

Yeah, exactly. I think they started out as an ADU and now increasingly single family. And then I think they have ambitions beyond that. So I think that's an interesting area. But I do think what the Palisades did, if there is one silver lining to it, is that, and it was, and it's partially because of when it came out with Abundance and other books of the topic is it exposed a lot of people to just how insane the system is that we've built for permitting something in California. And my favorite example that I give is I think a lot of the rules that we've put in place in California were genuinely, well, intentionally intention. And I mean it. I mean some of them were just bad people using a pretense to put in place a rule that they knew exactly what it would do. But most of it was well intentioned. But I think having two glasses of wine is a pretty good idea. I think having 50 glasses of wine is a pretty bad idea. What's the analogy here and regulations in California?

2:15:47

Oh, glass of wine worth regulation.

2:16:46

Every single one of those regulations you look at it's like. It's a good idea. Like, it's a good idea to have more standards on heating.

2:16:49

Sure, sure.

2:16:56

It's a good idea that the community should be able to have another review on the design. This is a good idea. This is a good idea. But this is 50 years of legislators who each felt like they wanted to fix an issue, but they've created this layer cake of regulations that's just very hard to get through, which actually is one of the things that we can deal with better than individual homeowners, because in some sense, we are permitting the entirety of the city at once.

2:16:56

Yeah.

2:17:25

And what that means is we're doing it wholesale, which means we have to spend an insane amount of money on lawyers and consultants to do all of that work. But we can drive down the costs of the housing because we are spreading it out over so many units.

2:17:25

Yeah.

2:17:39

And a big factor to permitting is speed, right?

2:17:39

Exactly.

2:17:43

Even. Even if you're still dealing with the same number of regulations, if you can just turn approvals over faster throughout the build process, you can still have a tremendous impact.

2:17:44

I want, like, a synthetic permitting office in, like, embedded in CAD or something. So as you're designing your structure, you put a window there and it runs a simulation to say, hey, if you were to submit this, the permitting office would say no to that. Or. And then you slide the window over 2 inches and says, okay, now you're good. And you just have like, sort of red, green lights on everything you're designing. So you're getting feedback instead of having to submit, wait for them to come back with the revision. What you submit should be almost always approved and accepted, but there's still all of this uncertainty when you submit permits.

2:17:55

I think that's much closer than it might seem.

2:18:31

Yeah.

2:18:33

And I think there's a. And even before we get to it, I'm really optimistic about AI tools that you basically embed in cities.

2:18:34

Sure.

2:18:41

And then as they get these applications come in, instead of people manually review them, you just do a pre scan. And I actually had discussions with legislators in Sacramento who want to run those bills that you could basically. Actually, it's very much what you said. It's. You can submit your application and it just runs it through, and it says, we think that this is acceptable, or here are some issues to fix. So that's kind of the front end to the client, to the architect or to the builder, but then even to the employees who are working those jobs and improving it. You could see a system where it comes to them and is like recommendation is that you approve it. And the five areas where me, the AI have the least confidence is these five. Go check them. Because I think it's easy to vilify some of the permitting authorities and some of the people who work there, but they just trying to do their best. And some of it is real. I mean, I don't know the details of it, but the tragedy in the nightclub in Switzerland, I was talking to a friend who is on the ground and who knows it. And part of the issue, allegedly there is you basically had a situation where it was a small town, there were some relationships. And the fire chief, My understanding is the fire chief or whoever was in charge of doing the inspections was friends with the club owner and they didn't want to deal with the costs of revising the material they had on the floor. And then they had a fire exit and they locked it because they didn't want the teens to be sneaking into the nightclub.

2:18:41

Interesting.

2:20:14

And so you like, it's very easy to be really, really upset about your fire department. Be like, this is stupid. We should get rid of it. But if you want to be able to walk in a restaurant or you have kids walking in a restaurant or walk into an office and feel like, feel secure in the environment, I think you need a balance. And I'm hopeful that AI can actually give us the best of both worlds.

2:20:14

Yeah, me too. Jordy, next question.

2:20:38

On the customer side for California Forever, customers being businesses that would come in and build facilities, are you within the range now where you can move? You. You've said you've talked to hundreds of companies that it would be potentially leasing or buying or building in California Forever? Given that you're hoping to break ground this year, are you at a point where you can actually get more serious with these companies and say, hey, let's actually put this into. Not go from concept to actual planning around building factories, ship yards, etc.

2:20:41

Yeah, I mean, we've. We've so. So we have two timelines. We have a default timeline, which is we going through the usual process when we would break around in 2028. And that's what we've told the companies. We said, this is, this is what we feel pretty strongly that we can deliver. And actually part of where the 2029 breaking ground, sorry, 2026 breaking ground is coming from is a bunch of those companies have come back and said, you know, we left Solano as a location. We left putting the facility in the middle of what you're building. But we need to place it and break ground in 20, 26. So if Solano can deliver and if you can break ground in 26, we want to put it there. If you can't, unfortunately, it's going to have to go to name another state, generally speaking. And so we've brought that to the local elected officials, and we've brought that to kind of economic development agencies at the regional level and said, can we work on this? Can we figure out a way to do the permitting in parallel?

2:21:16

Because you don't want to lose these jobs.

2:22:08

Tax revenue and job jobs will just go to other places, and we have an opportunity to keep it here in California. Exactly how has the support been from local officials and even Sacramento?

2:22:09

I mean, it's night and day from two years ago. And as I mentioned, kind of in the beginning, obviously very controversial, and people didn't know what to make of it. You look at it today, there's an open letter that was put up this morning on our website in connection with a labor announcement we made that is basically this. I mean, honestly, it's a love letter to California that is signed by the construction unions, by us, by a bunch of other people saying, hey, this is the state that used to be able to build the Golden Gate Bridge in four years. We built Hollywood, and we built the California water project. Surely we can break ground on this quickly. And there's a moment, and if you look at the signatures, it's the former speaker of the California Assembly, Bob Hertzberg, who wrote the legislation around how cities would expand. It's Alain Bertot, who I think is one of the most respected urban planners in the world. It's former senator Bill Dodd, who was senator for our district until two years ago. It's countless local mayors, former supervisors. And so there is a really incredible coalition that's come together to say, hey, we can do this. We can. We don't have to kind of skip steps. We can just do them in parallel and break ground. And part of that is companies saying that. Part of that is construction workers saying, I'm sick and tired of commuting for two hours. I want to have a job 20 minutes from home. Part of that is Solano had some old industries closed down in the last couple of months. There was a Budweiser brewery in furfield. It was 50 years old, that's closing. There's a refinery that's closing, closing. And so there's a need to replace these jobs. And so you see the outlines of this compelling argument to say, hey, this can be a solution. To the very practical local issues of tax revenue and jobs. But at a much bigger level, this can be a real moment for California where the state can show, hey, we can do big projects. This isn't just high speed rail. And I think there's something deeply poetic about doing that in 2026. Yeah, I mean, it's 250th anniversary. Right?

2:22:21

I love it.

2:24:27

Growing up as a, Growing up as a. As a kid, I can't tell you how many times I, like, saw a piece of land and thought out loud with whoever I was with, what if we put this there? What if we put that? And every single time the messaging was, oh, like, yeah, technically you could, but it'll take so long, like, it's not.

2:24:28

It's not worth.

2:24:46

Worth doing. And so I'm so optimistic that getting California forever off the ground, active, a real place that people live and work and build, will just inspire, could be like, really a real turning point for the whole state.

2:24:46

So how many acres do you have now? Your empire's growing. What's the total amount of acreage for California?

2:25:00

Total acreage.

2:25:07

Give us the number.

2:25:07

69,000 acres.

2:25:09

Boom.

2:25:10

Congratulations. Thank you so much for coming.

2:25:14

Always in the studio.

2:25:17

Yeah, yeah.

2:25:19

I'll come back when we break ground.

2:25:19

We can't wait.

2:25:20

We can't wait for our first episode.

2:25:21

From there, we'll do it.

2:25:23

You can come over. We'll do it from there.

2:25:24

Thank you so much. We'll talk to you soon.

2:25:25

Goodbye.

2:25:28

The New York Stock Exchange want to change the world. Raise capital at the New York Stock Exchange. Cha Ching. We have some news here from OpenAI. This is sort of leaked through Foxconn, but from Dan Neistat says OpenAI will unveil its first AI earbuds, dubbed Sweet pea, in September this year. And shipments are expected to reach 40 to 50 million units in 2027. So this is citing OpenAI's Chief Global affairs officer, Chris Lehane. Taiwan's Foxconn will do the assembly for these buds. Makes a ton of sense. You know, you're chatting with OpenAI, you're chatting with ChatGPT, you got Jony, I've on the team, former Apple. And you put something together that's your always on AI assistant that you're chatting with. But 40 to 50 million units, that's a lot. So I wanted to put some context around how. How massive that would be. So the Amazon echo in year two sold 11 million units. The Xbox series X X S. This is the sequel to the Xbox. There's already an Xbox install base 16 million units in year two. The iPhone. The iPhone in year two, 17 million units. The Apple Watch did 20 million units in year two. The PS5, massive success. Hugely anticipated product. The PS4 was almost a decade old. Lots of people upgrading. The PS5 shipped 30 million units in year two. The Nintendo Switch did between 35 and 37 million units in year two. AirPods. Now AirPods did do 50 million. So if the ChatGPT OpenAI earbuds dubbed Sweetpea perform as well as AirPods, you could see 50 million units. And the iPad. The iPad sold 55 million units in year two. But it's a tall order from the go. It's a very, very tall order.

2:25:29

Consumer hardware. There is more news from Wayne Ma and Keanir Loo. Apple is developing an AI powered wearable pin the size of an AirTag that is equipped with multiple cameras, a speaker, microphones and wireless charging.

2:27:37

Shots fired.

2:27:48

The device could be released as early as 2027.

2:27:49

Yeah. Wait, wait. Why are you reacting?

2:27:52

I'm just like, like, I feel like a lot of reason why people like do the pin or something is because they can't get all the features from the iPhone, but like they already have the iPhone, so why don't they just make that good? I guess they probably will, but. Do you think this is a good idea, Jordy? I feel like I'd rather just have it in my phone.

2:27:53

Yeah. It is crazy. If you're in the Apple ecosystem and you have a MacBook, an iPad, a watch, earbuds, AirPods and the phone look and the Vision Pro, it's like they got 90% of your body covered with tech. Like you're pretty good.

2:28:07

You're talking to a wired headphone guy. Yeah, I've been rocking Apple wired headphones. Yeah.

2:28:23

Yeah. So if this was wired in, you'd.

2:28:27

Be good to go probably five years now. No, no, I'm not in the, in.

2:28:29

The market for this.

2:28:37

I'm not in the market for this, but I gotta wait and see. I'll continue here. So Apple's trying to release their device as early as 2027 as well. Such a product would position Apple to compete more effectively with OpenAI, which is planning its own AI powered devices, and Meta, which is already selling smart glasses that offer access to its AI assistant. Google also plans to release smart glasses in conjunction with Samsung. Apple's development is still in the early stages and could still be canceled. Still, Apple is endeavoring to move faster than usual to try to stay competitive because of OpenAI. It's planning to manufacture roughly 20 million units at launch. The person added. Apple's pin joins a growing portfolio of AI powered products the tech giant has under development, including AirPods equipped with enhanced sensors. Like, yeah, it's kind of a.

2:28:37

It'S.

2:29:27

Like, are people really going to have like their pin on and their AirPods in and have their phone? Just how many AI enabled devices do we need?

2:29:27

Shoulder pads, the smart cane, you want the breastplate.

2:29:39

And there's the, there's the Apple, there's the Apple home device that's going to.

2:29:44

Be like, okay, I actually do we have to get to our next guest, but I do have a hot take that there might be a future wearable that looks like the Pip Boy from Fallout. Are you familiar with the Fallout series? I'm actually not kidding about this. And I'll unpack it. So basically in Fallout you wear a device on your arm and you turn it and it's basically a phone that's glued to your arm and then you can access your menus through here. Right? And it was always like something fun that like the, you know, the cosplay community would build out. Yeah, there's the Pip Boy and people have made real versions of this. It has this cool retro aesthetic. It's never taken off. But I've noticed that if you look at the live streamers like Ishowspeed Kaisana, when they're out in the world, they often have phones strapped to their arms to read chat while they're going around. Like when I think it was Speed that went to China and he went on this world tour, he toured America. He keeps a phone strapped to his arm so he can use both hands. And it feels like weird in cyberpunk, but it feels like a glimpse of the future. Like we're all going to be living like speed in the future maybe and maybe we're going to have phones or something like it glued to our. That could just be an accessory that actually helps you glue the phone to the arm. It could be some sort of modified device that's curved to wrap around your arm. But I feel like the end state is like more technology on your body constantly and that's like an easy one that people could adopt. You've already seen, you've seen Apple sells the sock that you can keep your phone and a lot of people are wearing like almost necklaces with their phones. So their phone never goes in their bag, never goes in their pockets. It's always just here and then they can just pull it out and Take a picture, do whatever. Because they're so glued to their phone. I think like an interim wearable is just something that makes your phone, like permanently dangle here. Because the phone is powerful, it has a lot going on in that, but you just never want to put it in your pocket. You're obsessed. What do you think?

2:29:48

Yeah, I mean, the next step is just like a pair of glasses.

2:31:51

Yeah, we'll get there.

2:31:53

You mount the phone. The phone uses the camera to show you the.

2:31:54

Google made that. Google Cardboard. You could take a cardboard box, fold it up and put it. Put the phone on your face anyway.

2:31:57

Yeah, I'm so. I'm so curious to see if. If OpenAI and Apple just happen to simultaneously kind of land on the very similar form factor.

2:32:04

Form factor.

2:32:17

And it actually becomes good enough that.

2:32:18

That they're in a race.

2:32:20

I'm not currently, you know, using LLMs, like multiple times a day for search and research or whatever. I'm not having these moments where I'm like, oh, if I could only just not get my phone right now.

2:32:22

Yeah, especially because of the. I mean, these prompts, they take five, ten minutes. I still feel like going into an agentic coding experience or even just a deep research report. I'm going to spend a couple minutes there. And so the two seconds it takes me to pull up my phone is not the rate limit. Maybe if the models get faster, but expectations are sky high, and it'll be interesting to see the reveal. How it's marketed, how it's pitched, and then ultimately how it sells. Let me tell you about Cognition. They're the makers of Devon, the AI software engineer. Crush your backlog with your personal AI engineering team. And I'm also going to tell you about Railway. Railway simplifies software deployment. Web apps, servers and databases run in one place with scaling, monitoring, and 30 built in. We have our next guest, George from Testudo in the Restream waiting room. Let's bring him into the TVP and ultradome. George, how are you doing?

2:32:35

Well, good afternoon, gentlemen.

2:33:28

Good afternoon.

2:33:29

Afternoon.

2:33:30

Great to see you.

2:33:31

Good afternoon.

2:33:31

We got.

2:33:32

We got it. We got to ask first, where does the name come from?

2:33:33

Well, it's Latin. It's Roman military history. It's all things that we love.

2:33:36

Okay.

2:33:41

It basically is Latin for Roman military formation, where the shields are on all sides, the front, the back. It's been in Lord of the Rings, it's been in many gladiator films. And that is exactly what our insurance is for the enterprise metaphor we work with. We are that Protective coverage from litigation that they might experience.

2:33:41

Yeah. Talk to us about the early customers who you're going after with this insurance product and then we can get into how AI plays a piece into your business.

2:34:03

Yeah. So this is a brand new category of insurance product. This is generative AI insurance. So excited to announce and launch that category today.

2:34:12

Previous creator category creators.

2:34:22

Come on, we've got to get a gong there, John.

2:34:25

Oh yeah, of course. And former Goldman Sachs vp. So gong, gong for that.

2:34:27

Yeah, there we go.

2:34:32

A lot of gong worthy moments. Love it.

2:34:35

Yeah, exactly. So brand new category. You'll have heard of cyber insurance.

2:34:38

Yep.

2:34:42

This is AI insurance.

2:34:43

Yeah. So, so walk me through is this. I build a tool with generative AI and then at some point a lawsuit comes along because of a hallucination. What are some of the downside risks that folks are ensuring against?

2:34:44

Yeah. So enterprises are deploying AI for as you know, a huge range of different things. The enterprise is our target customer. Any company in any industry deploying AI for anything. Quite a large market for us there. So our focus is to cover these enterprises with an insurance product. So if they are sued, for example for copyright risk, for kind of bodily injury, personal injury, all sorts of different types of risks. We cover the legal costs, damages and settlements associated with any liability risk that they might be exposed to.

2:34:59

How do you possibly underwrite this? There's so little data. There's the anthropic like billion dollar settlement open. I might have some massive settlement. At the same time there's like two data points. How are you actually.

2:35:34

Yeah, yeah, I mean I'm sure a lot of enterprises are like I want this. It's one of those interesting challenges where like, you know, if you're, if, if you get, you kind of need to balance that. Like if demand is too extreme, maybe you're underpricing it. But I'm sure you have a lot of comps from again, cyber insurance and maybe other insurance categories that were.

2:35:46

Exactly. Yeah. So that's the secret sauce and that's what's taken us over a year to figure out how to price this insurance without needing to connect or invasively audit the enterprise that we're working with so we can turn a quote around in the same day. You know, companies like Goldman are not going to allow third parties to invasively integrate or audit their proprietary systems. We can price and give quote indicative terms within the same day. So that really is key to getting this product out via brokers distributed to these enterprises. And the pricing obviously is what we spend a huge amount of time doing. We're using some very special new data that we have sourced ourselves. We're using lawsuits and basically have ingested the federal and state US lawsuit databases. We've taken all of the generative AI lawsuits and then we break down those lawsuits in specific causes of action so we know exactly what's driving the losses where in the US of that data, we then took it to Lloyd's of London, the largest specialty insurance market in the world, where we source all our capacity and essentially presented our view of the frequency and severity of the generative AI liability risk in the market. And, you know, we're super excited that we're launching the product today with nearly $1 billion of risk capital backed by. By Apollo Specialty in Lloyd's.

2:36:08

Let's go another gong.

2:37:28

Hit two gongs.

2:37:31

Love it.

2:37:33

I mean, obviously, I'm sure every policy is different and you guys are acting as the. You guys are the broker. Right. So you're not actually deploying your own capital or how does that work?

2:37:36

Yeah. So we're called a Lloyd's cover holder. In the US a similar structure is called an mga. It means we have full delegated underwriting authority. We have the pen, essentially. So the insurer has the balance sheet and the risk capital. We essentially have the pen. We do all the technology, the infrastructure, the pricing, the data, the monitoring, the portfolio controls. And we take submissions from insurance brokers. And it's our decision. Yes, no. As to whether we cover that company or not. Again, up to $10 million or eight and a half million dollars of risk for those different areas. And very excited to be working with enterprises deploying.

2:37:49

And so how does a claims process work? If an enterprise is running some gen AI product and they end up getting sued because of that, what typically qualifies versus just a company just being generally negligent? Right. So like if somebody is driving a car and they're drunk and they crash into someone, the insurance company is maybe going to.

2:38:28

The CEO says like, please go download all of Netflix to train a video model. That's pretty blatant infringement. I imagine that wouldn't be covered.

2:38:54

Yeah. So I'm hoping we're not going to be doing too much of paying out claims, but obviously we will be paying some to prove that the product is useful, but the claims are being managed by our capacity provider, by the insurer and Lloyds. And essentially you're right, we'll be getting kind of logs, we'll be getting all the information associated with exactly why there's been a lawsuit. What's really interesting from all our legal Data is only 4.9% of all of the generative AI lawsuits in the US market have been caused by model hallucinations or performance related issues. So you would think that the model performing incorrectly is driving a lot of the litigation, but that's not actually the case in the data. Obviously a lot of copyright risk, a lot of patent infringement, a lot of DNO claims, things like that. So if you over index too much on the performance of the model rather than where that enterprise is deploying the AI or which industry or, or what type of revenue that company has, you will miss the driver of the liability. So we're very excited to you know, talk through our approach.

2:39:05

Scenario in the future where people have a bunch of agents deployed working on behalf of a company and let's say an agent goes a little haywire, starts bribing a bunch of officials. How much are you thinking about the sci fi future where companies have more basically robots working for the company than.

2:40:12

Yeah, if AI kills everyone, I definitely want AI payout. That's important to me because I'll be on a spaceship going lightspeed away from the, from the gray goo boom. And I expect you to send the wire immediately.

2:40:33

Exactly. Product two for us is AI agents as you said. Product three is AI and robotics. And so we're getting in balance for both of those things. Now we are going to be multi line across, you know, the whole economy. And I really believe that insurance is the biggest horizontal play picks and shovels for the economy. You know everything in, in every industry is insured up to a certain level. So you know, we think this is going to be a massive category. We're excited to be defining and leading it and you know that agent product is number two for us, but we don't see a huge amount of enterprise deployment of agents yet. We think obviously these types of insurance products can unlock that adoption. And you heard last week with JD from with coverage, companies like Coalition have done a really good job on the cyber insurance side. They're a great comparable for us and huge fans of their work. This is not cyber insurance, this is AI insurance. We're very excited about the future roadmap and the category that we're creating.

2:40:49

Talk about policy concentration, diversity of revenue and policy size for you. I imagine that some of the big labs would probably love a billion dollar policy. They'd take 100% of what you have to allocate. Are you looking for 100 to spread that across 100 companies? A $10 million policy cap 1,000 at $1 million. What's the right shape of diversification for your business?

2:41:50

Yeah, that's a genius question and exactly what we're focused on. Obviously coming from a capital markets perspective, you know, diversification across jurisdiction, industry types, revenue bands, AI models, deployments. That is exactly right. We are weaving a basket of, of diversified clients, which means we can manage risk on the, on the portfolio, on the, on the whole portfolio.

2:42:16

And then walk me through a YC company for example, they raise $5 million $50 million valuation or something, they have a couple customers, they're making some money, they're still early, they're deploying a bunch of generative AI products. Should they get AI insurance at that point? There's sort of a normal function when you do a series A of getting DNO insurance, a couple other pieces of insurance when you leave the garage. How quick on the life cycle of a startup should someone consider this?

2:42:40

Yeah, so unfortunately this is not a product for startups. This is for kind of mid market enterprises who are deploying AI. We are not currently working with AI vendors or the base model providers. We think there's too much concentration, risk. We've seen the number of lawsuits that have hit, especially the base model providers. So as you say like if we were ensuring some of the big guys up to $1 billion, I'm sure we'd be paying a hell of a lot of that out in claims already. So we are focused on enterprises, you know, pings that we're getting from architectural companies to law firms to you know, banks to you know, any, anything under the sun really. So those are the target customers and we start you know, in the mid market.

2:43:16

What about non legal negative outcomes from AI? I'm thinking of, I run a, I run a company, I let a chat bot talk to some customers. The customers prompt inject the, the bot and they get refunds or discounts that I didn't want to give them. And it happens sort of quickly and I'm like ah, they, they, they took like some scammers came in, hacked the bots and they took 20 GS off me. Like can you do anything for me?

2:43:56

Yeah, I think we absolutely could do something for you. It depends whether this is a, like a first party loss. So you say like a financial loss that you've suffered yourself.

2:44:24

Yeah.

2:44:32

Or you've been sued. Right. This is a liability product so this is protection from litigation. So first party is definitely in the hopper for us. But this first product is third party litigation and the liability exposures that Enterprises are exposed to kind of out of the box when they're deploying these AI systems.

2:44:32

Fantastic.

2:44:52

Well, have you announced any funding yet or what's the story there?

2:44:53

We don't announce funding. We don't believe that's the purpose of this company. I mean, you guys have seen David Senra on all the time and he says the focus is building a durable business. Honestly, that's what I'm focused on with our moat within. Within the data, within the licenses, within the capacity, the balance sheet, and obviously our incredible team and honestly hit the.

2:44:58

Gong for not announcing that. He's not announcing it, but building a durable business. I'm reading that as we're building a durable business that has raised a little bit of money. Maybe a lot of money.

2:45:18

Maybe a lot.

2:45:32

But I love it.

2:45:32

I love it.

2:45:33

When you're ready with other announcements, please come back on. This is a lot of fun.

2:45:34

Yeah, yeah. Very, very cool.

2:45:38

And congratulations. I mean, honestly, like, how you're approaching this. Having the backing of Lloyds is like probably harder to get bigger, more important to the core business than grabbing a couple of checks from some Silicon Valley investors. So that's why we lean in the gong.

2:45:39

Yeah, yeah. Good luck to any YC startup going into Lloyds of London. It is a den of underwriters.

2:45:54

Good luck going to Lloyd's London without an accent like this.

2:46:02

I'm serious. And if you wear the wrong colored shoes or the wrong type of suit, you will be laughed out of that building.

2:46:06

So that's your mo, that's your moat. What's your moat? The mode is I know exactly what kind of suit to wear and AI can't disrupt that.

2:46:13

AI cannot disrupt those relationships. You both are invited to Lloyd's of London and a tour of the inner sanctum of underwriting by. By our head of insurance and my co founder Mark. So we'd love to have you there.

2:46:23

Thank you.

2:46:36

If you can drink Guinnesses at high velocity, we'd also love to welcome you to Leadenhall Market, the home of insurance boozing. But you won't find us there. We are hard at work, 247 around the clock and excited to launch the category.

2:46:37

That's amazing. Congratulations.

2:46:52

Where are you based?

2:46:54

We're in between SF and London, but looking to expand into New York and other. Other cities as well.

2:46:56

Amazing, amazing. So great to meet you, George. Congrats to the whole team.

2:47:03

Congratulations.

2:47:06

Not congrats on the funding. Congrats on the incredible partnership and all the progress.

2:47:07

Yeah.

2:47:12

Have you back on.

2:47:13

Great. Thanks, gents.

2:47:13

Have A great rest of your day. We'll talk to you soon. Goodbye, Gusto. It's the unified platform for payroll, benefits and hr built to evolve with modern small and medium sized businesses. Get on the backbone and then also head over to Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social, on marketplaces, and now with AI agents in seconds.

2:47:14

It's so fast.

2:47:39

Yes.

2:47:40

I will routinely just buy the products of our guests if they're on Shopify. I'll just buy it during the interview. Yes, it's one of my favorite pastimes. I really love it.

2:47:41

Well, we have our second in person guest in the TVP and Ultradome, Ara Kharazian from ramp. He is the economist. We'll bring him on down. He's here in the TVPN Ultradome. So welcome to the stream. Ara, good to see you. How you doing? Good to see you. Welcome to the Ultra Dome. Thank you so much. For those who have been living under a data center, please introduce yourself for everyone.

2:47:51

My name is Ara Kharazn. I'm the economist at Ramp Time and Money Save.

2:48:17

Both.

2:48:20

Let's go.

2:48:20

I research data on where businesses spend, where they're investing, where they're pulling back and what's growing. Our flagship research is called Ramp AI Index tracks the size of the AI economy.

2:48:22

Okay, well, we're going to have you. I think Tyler might have some beef with you. We'll see. Maybe you agree, because there's this article in the Wall Street Journal that I want to go through. It says CEOs say AI is making work more efficient, but employees tell a different story. Tyler took to the timeline to say auto manufacturers say cars are making more travel. Inefficient horses tell a different story. And so he's having fun with it. But there is an interesting stat in this Wall Street Journal article. I don't know if you've had a chance to read this, but basically they asked workers and C Suite executives, how much time do you think you're saving each week by using AI? And the C Suite said only 2% of C Suite executives that were surveyed by the Wall Street Journal said, no time at all. I'm getting no benefit from AI workers. 40% of workers said AI is not doing anything for me. Conversely, 19% of executives, the C Suite said they were saving more than 12 hours a week. And only 2% of workers said they were saving more than 12 hours a week. So you have this huge disparity at least in the reporting from the Wall Street Journal. How are you interpreting this piece?

2:48:31

There's two things going on here, right? You've got either the fact that the surveys may be the improper way of measuring this, because surveys don't always correspond to people's revealed preferences. So if you look at actual spend on AI models, it doesn't tend to reflect, you know, this kind of hesitance to use them or any lack of adoption. Yeah, especially because when we track spend, we're not just seeing that people are blowing a ton of money on models. We see that the spend that they're doing is becoming larger and they're renewing their contracts more, they're doubling down on them, they're typically extending those contracts out. That's not the type of spend that you do or investment that you make unless you're actually seeing some benefit from it.

2:49:42

And specifically, I think a year ago, sort of like mid AI boom post ChatGPT, there were a lot of startups that came out and had, we're doing vertical AI for this, vertical AI for that, AI SaaS. And a lot of people saw the charts that went from zero to $100 million in 12 months, eight months, two months, two days. And people were wondering, is it going to be sticky? Are the venture capitalists going to get burned? Are these going to be flashing? The pans are going to get steamrolled. But from the data that you've pulled, it seems like a lot of these companies, probably not all of them, but a lot of them, have become sticky. The spend has been recurring and it's growing. But tell me more about what you're saying.

2:50:18

And. Well, that's the second part of why these surveys may not be the best way to measure this is that the, this kind of adoption takes a long time to actually take place. We're still in our own tech bubble.

2:50:53

Yeah.

2:51:02

Where I just posted this data about cursor and anthropic, and every time I post about cursor, anthropic, it gets way less engagement.

2:51:03

Why?

2:51:10

Because a lot of people don't know what cursor and anthropic are.

2:51:11

Oh, interesting.

2:51:13

And not to no shade about cursor, Anthropic, like great technologies, but when I.

2:51:14

I literally thought you were going to say the opposite. That like the cursor fanboys and the anthropic fanboys are like downranking you or something.

2:51:18

Social media marketing content about Anthropic is just significantly lower because a lot of my followers, particularly those in the accounting space, even People involved in procurement are still new to understanding, well, what are the dynamics in this market, even for Anthropic, which is second largest player.

2:51:25

So Jordy had this thesis. We didn't really fully run the analysis, but I'd love to sync it up with what you're seeing because he was saying that there's a vibe war on X about the hottest lab. Everyone over the break was obsessed with cloud code. And then Jordy was saying let's try and marry that with what we're seeing in the app Store and look at where things are ranking. There's the discussion that's happening. But then you go back to the app Store and you see a very different picture. What stuck out to you about the vibes versus public perception in the consumer market and then maybe we can marry that to what we're seeing in the ramp data.

2:51:41

Yeah, I mean that was really consumer just, just, just because you have consumers on ax, just saying this company's cook that cooked. And then you just look at like the actual reality in the charts which shows acceleration and it just shows that social media reactions in real time are not even actually driving. Like yeah, macro acceleration in the even downloads of different apps.

2:52:17

Sure, sure.

2:52:40

That has less to do with enterprise.

2:52:41

Yeah. And then, and then what are you seeing in the spend data?

2:52:42

A lot of those consumer trends start to reflect in business trends as well. So in our data set, we see that OpenAI has the highest adoption rate amongst US businesses, approaching about 35%. And to be clear, only about 46% of businesses on our platform are paying for any AI model service at all. Yeah, the vast majority of them are.

2:52:45

Probably using open 50% just being like horses. Yeah, horses are actually fantastic.

2:53:02

Unpack that. Is it, is it that they're just not paying but they're using free tiers? Is it that maybe they're paying but on a personal basis and the CEO is paying for a chat, CBT Pro subscription or cloud code or something. And then when they go to work, they're not expensing it, so it's not showing up on a ramp card.

2:53:09

There's a couple of dynamics again. Yeah, like we already got criticism for having a data set that's probably a.

2:53:28

Little bit more tech forward than other.

2:53:35

Data sets and we're transparent about that and we think that's an important part of doing this kind of research. At the same time, I think that we're likely underestimating AI adoption in some ways because we don't capture free usage, we don't cap because we Only see anything that's associated with ramp card spend. We don't capture employees using their personal accounts. And then even when you do capture all that, where do you draw the line on what AI usage actually is? I mean this is a big debate happening in the sort of AI economics research, particularly as far as how we're going to measure the labor market impact. Because the real question you want to answer is are companies using AI? And then secondarily, how intensely are they using AI? It's very difficult to measure. You can't just use dollars to measure AI intensity because the models are getting cheaper over time.

2:53:38

Sure.

2:54:20

So comparing those dollars spent over time is not going to be helpful to you as a researcher. And then second, where do you draw the line on what AI usage is? And then what kind of job can actually be automated? So far the research has tried to define job families with characteristics that are exposed to AI and likely automated. That research has, it has gotten plenty of criticism itself, especially as the AI models develop and they start to be capable of doing more and more. So now the research community is coming around to, well, we need to figure out which companies are using AI, get a data set. Ideally that's something that private sector is working on. Why we put out our ramp AI index research tracks which businesses are using AI and spending on it in some meaningful way and how they're using it and then ideally be able to marry that with payroll statistics and see, well, are there shifts happening in this part of the labor force or others?

2:54:21

Yeah. The thing that is going to be like, I don't care if a CEO does a layoff and says like we're doing this because we're getting efficiency from, from AI because the incentive is just to blame it on AI efficiency instead of we over hired or businesses and growing as fast as we thought it was. And all this stuff you actually want for the labor replacement kind of theories to be proven, you have to see somebody actively switching spend from payroll to tokens.

2:55:13

Yeah.

2:55:45

And you're going to have to see it over a long period of time. I agree that when you say, see these layoff announcements, a CEO may or may not say that it's about AI, but that's not going to matter when you know, hundreds of thousands of people get laid off all the time. It's sort of a drop in the massive size of the labor market. You really want to track these changes over time as they happen, particularly as they affect other sectors.

2:55:46

Sure.

2:56:05

And whether or not those sectors that are seeing the concentrated employment effects are.

2:56:06

Having AI story, have you been surprised at all at the types of individual companies that are adopting AI? You know, are you seeing a difference? Have you broken out like professional services firms, so like accounting, tax prep, legal versus SMBs, like retail and, or you know, like, let's say like car dealerships, things like that?

2:56:11

A lot of the early adoption is where you'd expect. I mean the vast majority of tech companies, for example, we do track their adoption and it's been high for a long time because the vast majority are raising some AI models. Finance directly follows that and I think that makes sense especially for tech. Right. These are the tools that were developed by engineers for engineers and therefore the best productivity enhancing tools were things like cloud code or the AI code editors, things that would directly help the people who were building the technology in the first place. What we're going to start to see soon though is how it's going to affect other parts of the labor force as especially now cloud cowork is out, increasingly there are AI agent tools that can automate and have access to the broader context of a worker outside of a coding tool. So you can start to imagine a more typical office worker benefit from these tools now. And that's where we're going to see the greatest productivity gains as those tools proliferate. And we're starting to see that change in business strategy now from the model companies themselves. Anthropic is doubling down on subscriptions, increasingly growing their subscription revenue. They used to be sort of seen as like, oh, it's just about APIs, that's their business model. No, they're like really leaning into the consumer side and the employee subscription side. I think we're going to see that from OpenAI a lot of competitors too.

2:56:33

Yeah.

2:57:46

Can you help me understand a little bit more about how you would theorize about the substitutability between payroll dollars and AI spend? Because if I see a public company CEO lay off a thousand people like that might be half a billion dollars of payroll spend over a year. And if I'm not seeing, well, a quarter billion dollars of AI spend, I would not attribute it to AI just instinctually because I feel like if there was a substitute of effect, it would be, well, you know, AI. An AI worker is half the price. Not an AI worker is 1000th the price. So sure, yeah, get, you know, spend, spend $100,000 on AI and lay off 1,000 people. That was costing you $100 million. That just doesn't clock for me. But am I off there economically or do you think there's something to that?

2:57:46

Our early research right now shows that the job profiles that are going to be affected most by this transformation are those most marginally connected to the workforce. So you can imagine something like a labor market marketplace, like an Upwork or Fiverr. Yeah, this is where people go to. It's anything from designers to software engineers to.

2:58:41

Yeah, we gotta check in. We gotta check in on. On some of those businesses. Because I used to think like, oh, you want a song made for like a video? Like, go to.

2:59:00

Go to.

2:59:09

Yeah, go to.

2:59:09

Fiverr.

2:59:10

Go to.

2:59:10

Upwork. A blog post written for SEO. Like a lot of SEO, blog post writing was highly scoped.

2:59:10

One off task.

2:59:17

Yeah. Well, yes.

2:59:18

And instructing a human being to do that task is not dissimilar from instructing an AI comprehensively.

2:59:19

No, the job description is the prompt and then you're done. And there's very little back and forth. It's just, turn it in and I'll pay you or I'll grade you.

2:59:24

Yeah.

2:59:32

On RAMP spend, we have seen a significant decline in spend on those labor marketplaces and specifically among firms that have replaced that spend with AI spend. But it's not one for one. Okay, so there are significant savings.

2:59:32

Is the ratio 50% or more than 50?

2:59:43

But unfinalized numbers a lot.

2:59:46

Yeah, but it's not 1000th maybe or something. I don't know.

2:59:49

It's somehow. So Fiverr is down 50% over the last 12 months. Upwork is somehow up 20% over that same period.

2:59:53

Yeah, but over the past couple of years, I think they're down over the last couple years.

3:00:04

Yeah.

3:00:09

How do you think about this? Going back to this Wall Street Journal article, How do you think about AI tools diffusing through organizations? Either from your experience inside of RAMP versus the companies that work with ramp? There is a world where you could see C suite executives. They're like, I definitely need to be doing deep research reports on every business question that comes to me. Maybe I'm not Vibe Coding Solutions myself, but I'm a power user of these tools because I take my job seriously. Alternatively, there's a lot of young people who are just AI native entering, and they're more on the junior side who come in and sort of bring these AI tools with them and might demand an AI tool. So how do you think about AI adoption across levels of seniority within organizations?

3:00:09

It often starts with one team at a company that might be an early adopter or might be the same adoption dynamics that happen in a larger Market where it starts with, with early adopting companies happen at the micro level at a company too, where it might be a couple people at a company that then evangelize it or bring it to a team, then you might see a larger pilot program. I've criticized the idea of pilot programs because I just think you end up trying to sign a multi year contract for one thing, instead of doing a more constant evaluation of the types of products coming to market, especially for enterprise lately. So the diffusion will always happen, will typically happen like that. It depends on the company, how much leadership there is from, from the executive team on, on increasing AI adoption. I think the vibe has also changed where more and more companies are not shying away from using AI. I think it's probably a net positive to talk about how you're using it effectively.

3:00:58

Yeah.

3:01:54

But it certainly depends on the industry because there is, I think a PR problem.

3:01:55

Oh, totally. Yeah.

3:02:00

But for the tasks that it does well, I mean we'll talk about how AI is the frontier is kind of jagged. We kind of know what it's good at doing. We're not entirely sure. But for the task that it's figured out pretty well, I think most companies are very receptive to evangelizing it fairly quickly.

3:02:01

Are you as part of your research, are you actively trying to replace your own work? You are an economist at ramp. Your job is to look at all of our data and turn it into easy to understand reports.

3:02:16

Yeah. What's your tool chain?

3:02:30

I mean, for a couple of reasons I just have to be on top of it all just so I can.

3:02:33

Talk about it effectively.

3:02:37

But for the experimental early stage research process, they're so good. I mean even a rudimentary aspect of most economic analysis is just doing like a seasonal adjustment on your data.

3:02:39

Sure.

3:02:51

And it's really just integrating a Python package. But there's a lot of different times where you might use this or that package and if before you might have to write that code yourself to say, hey, do this seasonal adjustment, you can now just instruct it to swap in and out packages and really check your results across different methodologies. So it really does make my work a lot better because it allows you to. Particularly for my job. Everything I do is public facing. So all my research is just out there. But you really only see 10% of what I ultimately worked on. A lot of the research threads that I go down, never see the light of day.

3:02:52

Sure.

3:03:24

And so AI allows me to experiment a lot more, check a lot of different results faster and ultimately get through the bad ideas more quickly so I can focus on the good ones.

3:03:24

There were some without AI you would need like a bigger team in order to put out the amount of research that you put out.

3:03:34

For sure, yeah. I mean for me it's certainly been a product of enhancing tool.

3:03:40

Sure.

3:03:44

But I haven't figured out yet how to do the more multimedia stuff Like I increasingly am making my own videos. I am still doing a lot of my own editing. I make my own promo images, I make my own charts. I haven't found the AIs to be.

3:03:45

For me I think they need to build the RL environment around Premiere or some video editing.

3:04:01

No, my AI girlfriend's always telling me to work on that aspect. I'm just thinking about you all the time.

3:04:07

There were some, there was some FUD around Open air towards the back half of last year. Is OpenAI cooked? What does the data say?

3:04:17

There's no definitive leader in terms of the race. Market's changing really quickly. OpenAI is a distinct advantage as far as distribution. It's at the most businesses now. Google also has a distinct advantage because it integrates Gemini for free across all of workspace.

3:04:25

Sure.

3:04:40

So in our data set I think we likely undercount gemini adoption. But OpenAI just by nature of being kind of this consumer default. If you're a company that is going to make some early business investment in an AI model for your employees, it's kind of a no brainer oftentimes to just go with hey, what do you know about as a consumer and what are your employees now?

3:04:41

Yeah.

3:05:01

And so for that reason OpenAI I think is in a very safe place.

3:05:01

Cool.

3:05:05

But it's a pretty tense race and we have seen particularly in the sort of product layer, you know, large incumbents being losing a lot of market share very quickly to like a new entrant GitHub co pilot a couple years ago that was the major. That was pretty much the only AI code editor available for enterprise. Six months later cursor comes down. It's at 50% market share.

3:05:06

Wow.

3:05:28

So I don't think any company should feel particularly comfortable even if they have the enterprise backing and distribution.

3:05:29

Yeah, only the paranoid survive. So OpenAI is growing. Anthropic is also growing 1.6 percentage points to 16.7% of businesses using Anthropic. Biggest spenders in the tech sector leaning heavily on API spend and then Google adoption growing as well. So basically it's all good news. There's still a little bit of a horse race but everyone's making it across the finish line. For the most part.

3:05:34

People were worried over the last couple months. Well, cloud code is getting so much better. Is it's going to automate away things like cursor or obviate it away. And what we just found in our latest research is that cloud code is definitely accelerating in its adoption. But cursor, you look at dollars spent like the pie is still getting bigger, too. So there are enough people entering the market such that. And also, by the way, enough people that are enough companies that are willing to buy multiple tools are still experimenting or at least finding that they can afford to have multiple tools at their disposal. So neither of them are losing revenues and they still continue to grow.

3:05:58

I imagine that as businesses scale, at a certain point, they want to go with an enterprise plan, they want a contract, they want billing. Do you think you'll maybe have to supplement your data with survey data? Because you might see a drop as a bunch of businesses say, hey, we don't want you training on our stuff. Let's just sign some big master service contract and we'll wire you the money or something. Or how are you thinking about adjusting to that?

3:06:37

I think that the value of ramp data is that we see actual money movements and you can go to another researcher if you want survey data. I think what we do well is that we're focused on what we uniquely have access to. There's no data set that tracks where and how American businesses spend, at least not one that's available to the general public. So our research tries to focus on that and track those.

3:07:02

That's great.

3:07:26

And you're pushing the ramp sales team to get even more businesses on the ramp platform. Feed me.

3:07:26

Feed me. Every time you sign up for Ramp, you build my data set. All anonymized, all aggregated. We don't do anything like that. But for that reason, this is the. At the start of this podcast, probably the first time I ever pitched ramp. Like, pretty separate from the sales team for that reason.

3:07:33

That's good. Yeah. The real nightmare is when Ramp rolls out enough AI tools that you have to classify being a ramp customer as an AI adopter. And then 100% of the data is AI adopters.

3:07:48

Yeah. This research at some point is going to become useless when it hits 100. I mean, imagine if there was like a ramp Internet index in the 90s.

3:08:00

True. It's like how many people have websites?

3:08:06

If I were still talking about that.

3:08:07

Yeah.

3:08:09

20, 30 years ago, everyone would be snoring at. People are still interested.

3:08:09

Yeah. Yeah.

3:08:12

Would be Would be useless research.

3:08:13

But at the same time there's a ton of interesting research to be done deeper in the Internet stack. Well, what clouds are you on? What databases are you using? What front end frameworks are you using?

3:08:14

We're hiring on Economics Lab. If you're. These are great ideas. If you're. I know you got your talent signing today.

3:08:23

Yeah.

3:08:29

John helped with the census at one point.

3:08:30

I did, I did work for the census at one point. Yeah. In college. It was a fun, fun gig.

3:08:32

They do great work.

3:08:36

They do, they do, yeah. I mean from, from working there and doing everything on pencil and paper. I was like the first person I think to ever use like Google.

3:08:37

Oh, were you a door to door census taker?

3:08:46

I was no.

3:08:48

He quickly worked his way up.

3:08:49

So I was a manager of a unit that would re interview people to make sure that the door to door census takers weren't like lying. Because if you're a door to door census taker, you get paid 20 bucks an hour. You just go up and you say I have to walk to all these doors and say I saw three people, I saw four people. But you could just write that and then just turn that in and not do the work. And so we'd go reality check all the day.

3:08:51

John was also 20 at the time, managing a team of hundreds.

3:09:14

It was a big team and they were all older than me. It was very crazy.

3:09:19

I want every tech person listening to recognize the value of government data and the checks that it always goes through.

3:09:22

Yeah.

3:09:27

And, and the great work that you did. There's some good stuff in your early career.

3:09:28

Yeah, it was a lot of fun. Anyway, thank you so much for coming on the show.

3:09:31

Great to have you here.

3:09:34

Always great. We'll wind down the show.

3:09:36

Thank you.

3:09:39

Thank you so much.

3:09:39

In a second.

3:09:40

And with that we will conclude today's show. We will be back tomorrow at 11am Pacific Sharp on the dot. Hopefully earlier. Is there anything in the timeline that we need to go through before we get out of here? Jordy?

3:09:41

Let's see.

3:09:55

We didn't even talk about CIA. We are officially signed with CIA. We went Hollywood mode. That's very exciting. I think everyone's excited.

3:09:57

Yeah, Very, very excited.

3:10:04

We're pleased to be partnered with them. We put up a trading card.

3:10:05

We're very excited. Team.

3:10:08

Yes.

3:10:10

They're going to be helping us navigate the entertainment landscape. Yeah, very excited.

3:10:12

Yeah. We went down, we took some headshots at CAAHQ also.

3:10:19

You know, I like to make them a little goofy.

3:10:24

The creative artists Agency. Sort of the anthropic of agencies. Do you know the lore? Did you actually get up to speed on this? So Creative Artists Agency was formed by five agents. They were at WME, they were at William Morris Agency in 1975. Who you got? There's a dinner. You got Michael Ovitz, Michael Rosenfeld, Ronald Meyer, and Roland Perkins, and William Haber. And they're at this dinner and they decide to create their own agency. What happens? They're like, okay, we need some financing. We're gonna start a rival agency. We're leaving wme. We're starting CIA. What happens before they can get financing? They all get fired. They all get fired. Seriously? This is real.

3:10:26

This is on the Wikipedia, like thinking machines dynamic.

3:11:07

Yeah. No, it really is like AI lab talent wars all over again. And I don't know if they're the anthropic or the thinking machines, but maybe they're the open eye on this. But the. The metaphor is extremely messy. But it's fun lore. Of course. CIA was incorporated in Delaware and had a $35,000 line of credit and a $21,000 bank loan. They rented a small office in Century City, and within a week they sold a game show called Rhyme and Reason, the Little Rich show, and the Jackson 5. An early plan was to form a medium sized full service agency, share proceeds equally and do without name plates on doors or formal titles or individual client lists with guidelines like be a team player and return phone calls promptly. I like those rules.

3:11:11

Great rules.

3:11:54

We're very happy to be partnered with them.

3:11:54

So I got to reread who is Michael Ovitz.

3:11:56

For sure.

3:11:58

For sure. Fantastic book.

3:11:59

Absolute legend.

3:12:00

And we. When we were at CA last week. Yeah, we or yeah, last week, we did confirm that the mail room still exists.

3:12:01

It does.

3:12:09

There was a lot of mail. A lot of people in it. They were moving the mail around. Apparently. They said we were the only. Only talent to ever.

3:12:11

That was crazy. I think they tell everyone. They probably tell everyone, but there was a lot of. There was a lot of positive energy. You can see that the next generation is already cooking, chomping at that. And. Yeah, getting ready to build. Anyway, thank you so much for tuning in. Is that the bomb? The bomb has been planted.

3:12:19

The bomb has been planted.

3:12:36

I love that.

3:12:37

Mark Gurman says the Siri chatbot is known internally as Campos. It'll likely be powered by Gemini, running on Google's own CPUs and Cloud Infra, meaning it's far more powerful than the one coming to.

3:12:37

I'm very excited for Gurman. He's coming back on the show. We're going to get him in person. Leave us five stars on Apple Podcasts, Spotify. Sign up for our newsletter tvpn.com meta.

3:12:46

Superintelligence lab delivered its first base models internally. Bosworth calls them very good.

3:12:56

More headlines, more headlines.

3:13:00

Work to do for post training.

3:13:02

Anyway, thank you so much for tuning in. We will see you tomorrow at 11am sharp. Goodbye. Nice work, brothers. I'll see you on the next one.

3:13:05