a16z’s $15B Raise, Tim Cook Exit Rumors, Meta Goes Nuclear | Ben Horowitz, David George, Alex Rampell, Jen Kha, Jeremie Eliahou

You're watching TVPN.

0:00

Today is Friday, January 9, 2026. We are live from the TVPN Ultradome. The temple of technology, the fortress of finance, the capital of capital. Let me tell you about ramp.com Time is money. Say both easy use, corporate cards, bill pay, accounting and a whole lot more all in one place. I forgot in the Vanity Fair profile that we were pitching Julia ramp so much. She actually put it in the profile. It was very, very funny. Anyway, if you haven't always read, we were in Vanity Fair yesterday should. It's a fun piece, little whirlwind tour of what a show that happened maybe six months ago. So a lot of things have changed, but it's a good Snapchat.

0:02

But she got up to speed.

0:41

Yeah, it's fun. Yeah, it's fun. Anyway, we have a massive show for you today, folks. $15 billion raised by Andreessen Horowitz. We have a bunch of folks. We have four members.

0:41

I'm gonna hit the gong. Just because.

0:53

Hit the gong. Warm it up. You gotta warm it up. Cause we gotta hit it 15 times. When we have Jen, Alex, David, George and of course Ben coming on the show. Hit that app loving Don Jordy. Linear, of course, Meet the system for modern software development. Linear's purpose built tool for planning and building products. We also have Jeremy Antiveros from Semianlysis coming on to explain energy, to explain data center build out and the gas turbine infrastructure that's going into those. He did a great interview with Ben Thompson that dropped yesterday. We're going to dig in and go deep into. Apparently there's a bunch of fascinating things. Apparently there's like 10 terawatts of requests for data center capacity, which is like way more than anyone would ever build. But it's because of this weird dynamic of you have to ask for more than you need because you might not get it. And there's a whole bunch of interesting things. He of course broke the story that Meta had completely changed their data center design. They were optimizing for sort of energy efficiency before. Now they're much more focused on speed of development and scale. And so we're going to be taking you all over the place today. But we're going to start with Steve Jobs, Apple. We're going back into Cupertino because there's a rumor that Tim Cook might step down sooner than expected. So this comes from Tim Cook, his compensation, We've talked about it a lot. 74.29 million per year. His salary's 3 million. Stock awards 57 million non equity incentive compensation. He gets a $12 million bonus. If he does well, he gets 21,000 in 401k personal use of private jet. 800k on that. That's nice to see. Only 800k vacation cash out of 56k security expenses. They're paying $900,000 a year to secure him. That's whole team of people. Probably some jacked tier one operators following him everywhere he goes. But he is rumored to be out. AppleTrack says Apple CEO Tim Cook has told senior leaders that he is tired and would like to reduce his workload. Rumors suggest he could.

0:55

I doubt they wanted that quote specifically to leak.

3:16

But it did via the New York Times. So rumors suggest he could announce a plan to retire as early as this year. Of course, the rumor is that John Ternus might step into that role. MacRumors has a story here. With Tim Cook having recently turned 65 years old and a number of other senior Apple executives having already departed in recent months or heading for the exits, there has been a significant focus on Apple's plans for who will succeed Cook as CEO. I was hoping for a Warren Buffett third act from Cook. I was hoping for him to just say, I'm just hitting my stride 65 to 95. That's where I'm going to do my best work.

3:23

That's my window. You haven't seen any compounding yet.

4:04

It's a completely underrated era for business leaders. If you can stay in the game and continue to compound from 65 to 95, that's where the sweet spot is. Just get ready to lock in, not check out. But he might be.

4:06

He loves a joke about him being underpaid. I actually think he is, or he has been serious for how big of.

4:22

A company is and what he's done.

4:27

To the determinism of Tim Cook coming in and just absolutely cooking for as long as he has, it will always be remembered.

4:28

Yeah, so. Several recent reports have identified Apple's senior vice president of hardware engineering John Ternus as likely to be named the next Apple CEO. And the New York Times has now shared a profile of Turnus with some context on his expertise and how he's viewed within the company. According to sources who spoke with the New York Times, Apple has begun accelerating its planning for Tim Cook's succession last year, with Cook having expressed a desire to reduce his workload, while software chief Craig Federicki, services chief Eddie Q, marketing head Greg Jaws, and retail HR chief Deidre o' Brien have all reportedly been seen as Potential candidates. Turnus appears to have shot to the front of the pack, with Cook likely to remain as chairman of the company's board of directors. So he's not completely out to pasture. He'll be in the boardroom. Ternus is known for his expertise as an engineer, having worked on many of Apple's devices through although he is known more for maintaining products than developing new ones. Big question about what the next decade or two of Apple's product roadmap actually looks like. How many more new products do they need? They sort of have. They sort of have one thing in every category for if we go through a major form factor shift, that could be an issue. But in general, if you have someone who's really good at maintaining products and keeping dominant market share, driving up margins, that could be the right person for the job. Quote about John Ternus He's a nice guy. Let's hear it from nice guy. Sometimes nice guys finish first. They always say nice guys finish last. I think that's a bit of fake news. This is from former Apple engineer Cameron Rogers. Quote about John Ternus. He's someone you want to hang out with. I love it. He's just a good hang. Everyone loves him because he's great. Has he made any hard decisions? No. Taking shots at your boy? Hey, we just like hanging out with the guy. We just like hanging out. Has he had to do any real work ever?

4:38

No.

6:36

Has he made any a single hard decision in his life? No. I'm sure that's not true, but it does characterize his role. I guess he hasn't been in the CEO seat, so he probably hasn't had to make crazy decisions like should we launch Apple Vision Pro now or later? He's not the one. He's just like, you told me to launch it. I got it done right. That's his role.

6:37

Should we make the iPhone less durable?

6:57

That's a hard decision. Are there hard problems he solved in hardware also?

7:00

No.

7:05

What?

7:06

This is an insane quote. Wow. Ternus and others may quibble with that assessment, however, as Ternus has been involved with a number of innovative products over the years, including spearheading effort to develop the iPhone Air and working on the upcoming foldable iPhone. That's exciting. Ternus is seen as a natural successor to cook, with an even temperament, strong attention to detail, and an intimate knowledge of Apple's supply chain. That's obviously very good. But he may not bring the visionary focus and willingness to take risks that Steve Jobs had, leading to debate among Apple employees about exactly what type of leaders.

7:06

We need to get the Germinator on.

7:45

We do, yeah. Yeah.

7:46

Nick, can you reach out to Mark Gurman to try to get him on the show Monday?

7:48

Let's talk through all of these things. There's so much to talk about here. Before we continue our conversation, let me tell you about the New York Stock Exchange. Want to change the world, Raise capital at the New York Stock Exchange. So there's this question. Will John Ternus, if he steps into the role of CEO of Apple, will he bring the visionary focus and willingness to take risks? That's Steve Jobs had. That's a tall, tall order. I think Tim Cook's executed extremely well. He hasn't even. It doesn't really seem like he's tried to bring a visionary focus. He's been the operator.

7:52

He's a supply chain visionary.

8:28

Yeah, visionary in his own way. But you were thinking, and we've been discussing this need for a Steve Jobs of AI, a visionary leader in AI, we have a number of household name type CEO, Sam Altman, Elon Musk, Dario Amadei Demis at Google DeepMind, but we don't quite have that Steve Jobs. Maybe that's too tall of an order, but you still think it's necessary. So walk me through your thinking. Before you do, let me tell you about Shopify. Shopify is the commerce platform that grows with your business and lets you sell in seconds online, in store, on mobile, on social media, on marketplaces, and now with AI agents.

8:30

Yeah. Everybody's worried about not having a job because of AI. Well, AI needs a jobs too. They need a Steve Jobs.

9:13

Yeah. Oh, I didn't get that. That's good.

9:20

So, yeah, we've talked about this a little bit this week. I tried to summarize it today in the newsletter and I'll kind of read through a little bit. So I went back and looked at the history of the phrase techlash. It was originally coined by Adrian Wooldridge and the Economist in 2013. He correctly predicted that, quote, the big developments of 2014 will be the growing peasants revolt against the sovereigns of cyberspace. The Silicon elite will cease to be regarded as geeks who happen to be filthy rich and become filthy rich people who happen to be geeks over the coming years. He was entirely correct. It was actually in 2018 that Techlash was the runner up word of the year.

9:22

No way.

10:03

So he called it perfectly. Obviously you had the Cambridge Analytica scandal, which is actually finally going to be dramatized this year with the social network too.

10:04

That's coming out this Year. Do we have a release date yet for that?

10:14

I don't think so.

10:16

Okay.

10:17

But it is in the works. And then, yeah, just growing concerns about monopoly power, privacy, democracy, censorship.

10:18

Really quickly, Tyler. October 9, 2026.

10:27

Yeah, that's what I'm seeing as well.

10:30

There we go.

10:31

Okay, we do book the tickets now. It's going to be an interesting. Probably great.

10:31

This would be a good. We should organize. We should, we should. Actually, I don't know, I'm not sure that this movie is. I expect this movie to hit like 10% or potentially negative in comparison to the social network one.

10:35

I agree.

10:49

And so I think it might be the kind of thing you get a bunch of people that go. And it's just like, okay, that was the Social Network.

10:49

The original movie is a really good Rorschach test for are you gonna have a good time in tech? Like, if you ask someone who is thinking about working at a company or tech startup, what'd you think of the Social Network? And they're like, oh, I thought it was awful. And I hated all of. And there were no heroes. Well, they're probably not going to enjoy tech. But if they came away from it being like, oh, well, it's actually really ironic. I'm going to start now because he just coded a thing in his dorm room that became really big. And yes, there was drama and fights over who gets what on the cap table, but even Eduardo Saverin became a multi billionaire. So, you know, sort of an aim for the moon, land amongst the stars situation. So, yeah, I mean, you could read it both ways. But I think most people, most tech insiders, if you ask them about the Social Network, they were like, that was inspiring. I listened to the music all the time. It inspired me to grind harder, basically.

10:55

Real quick, Happy birthday to the chief in the chat, day one. Happy birthday to you.

11:49

Gemini 3 Pro, Google's most intelligent model yet. State of the art reasoning, next level vibe coding and deep multimodal understanding.

11:58

Anyways, I'll continue. So first Tech Lash is all about how is this impacting our mental health? How is this impacting our democracy, the foundations of our country, society, privacy, censorship, et cetera. The second Tech Lash has begun. Feels like it started last year. This is one of the things, yeah, you don't really know. Sometimes it takes a while to realize, okay, we're in this thing now, that we can look back and see how public opinion has been forming around this. So I believe the average American believes that technology and now AI is now like a threat to their Way of life. So I was looking at.

12:07

There were rumors of the tech lash in 2024, when the image generators came out, a lot of the arts community were saying, this is really, really bad. It's going to put artists out of jobs. The thumbnail community on YouTube was upset. But this year it's solidified around. There's like three or four key points, key talking points. If you talk, why don't you like AI? Well, it's stealing copyrighted information, it's slop. It's putting people out of jobs, it's stealing water and stealing power and each one of those.

12:44

So the first tech lash was like, okay, these tech platforms, our lives are now existing in these platforms, and they are in some ways more powerful than the government. And now. So I had Maslow's hierarchy of needs pulled up and I was just like going through and looking at physiological needs. Right. Air, water, food, shelter, sleep, clothing, reproduction, safety, personal security, employment, resources, health. Right. All these different things. And then you just go up and you can see that there's good reason for the average American to just kind of believe, like, AI is going to mess all of this up. So starting at the bottom, Americans have heard that data centers use a lot of water.

13:16

I need water.

13:52

It's not necessarily factual. Yep, sure, water is used in the process, but we're not like blowing through water at the rates that the public perceives.

13:52

Totally. I was joking about this online. I was hypothetically debating with AI Doomer about water usage. Well, are they long water stocks? Because if you believe that AI is going to use all the power and you bought GE Verona, you did very, very well. But the water stocks have not mooned. So. Hey, D cells who think AI is going to use all the water, maybe you got to put on a long position.

14:02

Yeah. Also privatize a public utility, become a monopolist.

14:25

Also, I mean, we're talking to Jeremy at Semianalysis, who's sort of their power expert. Obviously, AI does use a lot of power and there's a lot of investment theses that can be built on top of the Semianalysis energy model. Why doesn't Semianalysis have a water model? Oh, because it's actually not a bottleneck to anything.

14:29

Yeah. So the power thing is more real. People now just assume, like if they. I have to imagine people are reading an article, oh, your power bill might be going up. If your power bill just goes up because it's a winner, you're like, oh, thanks, AI.

14:46

Yep, yep.

15:00

I didn't ask for this. So they've seen Terminator 2. So they can imagine kind of like the sci fi scenario playing out. That's one factor. If they're super online, they might have heard, like, the Casey Hanmer or other people have talked about this, like solar panels, you know, an AI system. Just saying like, hey, actually, this farmland, I could use it better than you humans.

15:03

Remember that Ilya video? So he did an interview with the San Francisco Chronicle. It was this video. It was like a video documentary almost, where they were interviewing him, but there was no questions. So you never saw who was asking the questions, but he was giving his answers. And he's sort of like, sadly walking around on a gloomy beach. It's, like, very moody.

15:23

And I would say he was Aura Farming.

15:42

He looked sick.

15:44

You're leaning over the.

15:45

He did Aura Farm San Francisco a little bit. But as I was getting dressed up as him for Halloween, we were playing that video and the makeup artists who were applying the Ilya Sutskever, all the makeup, to me were watching that being like, that's not inspiring at all.

15:46

Okay. Yeah. And I didn't even include that in here. But that's like the reaction, like, every time people hear leaders at labs talk.

16:06

Yes.

16:13

They're like, turn it off.

16:13

As opposed to turn it off. You could show someone an Apple ad or Steve Jobs clip, and it would be like, oh, dancing on your wired headphones with your ipod. Like, I love music. They're making music available.

16:14

Great.

16:27

I love it. And there were so many things that were just inspiring. So continue.

16:28

Yeah, continuing. So, yeah, moving up the pyramid. People have been told that AI is coming for their jobs. Some people have, like, actually had an experience that made them feel like, whoa, I thought what I did was unique and special, but I'm watching AI do my job, kind of do it on my own computer, maybe. You know, imagine somebody that's driving for Uber and Lyft, and all day long they're driving and they're just seeing, wait, they're sitting next to Waymos in traffic, and you're looking over and there's no one in the seat. Like, yeah, that's ominous. That's. That's. That's going to be scary if that's how. If somebody puts food on their table. And then every single CEO last year was saying, like, we have, you know, fortunately, we have increased efficiency due to AI.

16:33

Yeah.

17:10

And so we've laid off 10,000 people. Right. And so a lot of that is just kind of like spin marketing, etc. But that's what people are hearing. Right? And Then you look at what are the leaders are actually saying. So Ilya Talking for 10 minutes, people are like, whoa, that doesn't seem good. Ilya is saying like, he's saying, let's not do that.

17:11

He's trying to prevent that bad scenario. But it still reads like, whoa, I didn't realize they were taking that seriously.

17:31

Yeah. So you look at the quotes, just, you could easily look up quotes from Dario. Obviously had his quote, AI could wipe out half of all entry level white collar jobs and spike unemployment to 10 to 20% in the next one to five years. Elon had a good quote from over a decade ago. He said, with AI, we are summoning the demon.

17:37

Some people today might say the demon has been fully summoned.

17:56

Fully summoned. And Sam obviously said at one point, I will probably most likely lead to the end of the world, but in the meantime there'll be great companies. And so this kind of messaging credit to them, it's like super effective for fundraising. Right. If somebody's saying like, all jobs will be wiped out, the world will be destroyed, but in the meantime, there's a.

17:59

Lot of funds that are long demon.

18:21

You know, you're just like the demon.

18:23

Thesis, the demon, the free cash flow from demons.

18:24

Yeah. So it's like if you're sitting there being like, if AI is going to eliminate my job, I want to own a piece of it. So, you know, maybe, maybe I can benefit from it. So, yeah, so the big issue is like, anybody that's hearing all these, like, why would they actually be excited about AI, Right. Even though, even though it is so incredible in so many ways. Right. I gave the example earlier this week of like AI just being like a free sleep consultant for a toddler that just one shots it and it's free.

18:27

Even ChatGPT Health that launched this week, that's very good news for a lot of people that can't see a doctor that often or just don't have the time or don't have the money. There's a million reasons why that might be advantageous. And yet it's not that they're not pitching it. Like Steve Jobs pitched GarageBand, which was like, now anyone can be a musician, now anyone can be their own doctor is inspiring, but it's just like they are fighting an uphill battle because of those other quotes.

18:54

Yeah. So, yeah, and I was thinking about it, it's like, if you want to, if somebody is kind of like generally scared of AI, who do. Who do you. Who like, what do you, what content do you point them towards? Typically you'd want to point them towards the people building it.

19:22

Yeah.

19:35

And say. Or people around it. But even like Dwarkesh is like a little bit maybe like too high level or not high level. Half actually. Right. It's like a little bit too ethereal. Right. Talking about all these different potentials. But if you have them listen to like a Joe Rogan episode with one of these guys, it's gonna be.

19:36

To be fair not to debunk your piece, but I do think Demis is pretty good.

19:54

Yeah, Demis is great.

19:59

He doesn't have one of those crazy quotes. And there's been two documentaries about him. Both of them are incredibly inspiring. And when I hear him talk about AI curing cancer or humans curing cancer with AI, it hits a lot different because he literally has a Nobel Prize.

20:02

And is sold the problem. If you look at, if you just count up the views that Elon, Sam and Dario have gotten in comparison to.

20:20

Demis, he's much less of a household name. And he's also not the CEO of a big company because he's running the biggest lab in a big company. So there's that.

20:27

Yeah. So I've just been feeling like there's this gap, gap. Steve Jobs sized hole. Right. He had plenty of concerns about technology. He shared them freely. Somebody once asked him, so your kids must love the iPad. Then he said, my kids haven't used it. He just said, we limit how much technology we have in the home. He did talk about losing the PC race to International Business Machines. He said, if for some reason we make some big mistake and IBM wins, my personal feeling is that we are going to enter a computer dark age for about 20 years. You can imagine, like Sam saying something like that around, like, you don't want. And you've seen the internal sort of messages between him and Elon talking about, like losing to Google.

20:36

To Google. Yeah.

21:18

Like, oh, we don't want Google to control the AI God. Right.

21:19

Yeah.

21:22

He also had a. He had a 1994 Rolling Stones interview. Rolling Stone interview. The interviewer said, nevertheless, you've often talked about how technology can empower people, how it can change their lives. Do you still have as much faith in technology today as you did when you started out 20 years ago? Steve says, oh, sure, it's not a faith in technology, it's faith in people. Technology is nothing. What's important is that you have faith in people, that they're basically good and smart and if you give them tools, they'll do wonderful things with them. It's not the tools that you have faith in. Tools are just tools. They work or they don't work. It's people you have faith in or not. Yeah, sure. I'm still optimistic. I mean, I get pessimistic sometimes, but not for long. And part of this I just, it feels like people have like. Part of it's like fundraising, part of it's just how excited we are about AI, but like you're leaving humanity out. So there's this quote from Sam. He says if AI stays on the trajectory that we think it will, then amazing things will be possible. Maybe with 10 gigawatts of computer, AI can figure out how to cure cancer. So it's like, that's a fine quote. There's some way to look at it and be like, okay, this is abundance. This is super exciting. I'm maybe more optimistic about AI, but he happens to. He's saying that AI is going to figure out how to cure cancer. Right. And if you've used these tools today and talk to people that are at the labs, the reality is it feels much more likely that humans will use AI to cure cancer. Right.

21:24

Thesis.

22:56

Like Steve would have said, if AI stays on the trajectory that we think it will, then amazing things will be possible. Maybe with 10 gigawatts of compute, humans can use AI to cure cancer. Like small, small tweak humanity. It's a big difference. And so I wrote the facts of the fact. Steve Jobs was not one to shy away from impressive specs and massive scale. But flipping the final line from AI will cure cancer to humans will use AI to cure cancer makes all the difference. Apple put human centrality at the heart of everything they did. Even when they were talking about something like a CNC to mill an aluminum block into a MacBook Pro, the focus was not on the CNC, it was on what it allowed the human being to do. Right.

22:56

CNC is literally a robot. It's computer numerical control. But when you watch that unibody presentation, it puts Jony I've at the center and it's like I used the tool to create something beautiful out of this amazing material that I could never do with just my normal tools. Like, I could never chisel out an aluminum unibody. I need a CNC for that. I have it and I can create something beautiful.

23:36

Yeah. So yeah, at the end of this, I just said, like, I think AI has a massive narrative problem right now. The narrative is working within the industry. It's not working for people that are outside the industry. And I just don't. I really don't think it has to be this way. Right. Like, I think that there is a human central, there is an empowering way to pitch this technology in this future and we're not doing it right now. And I expect that everybody will. You know, Elon has his own style of pitching all things and I don't think he's going to change. I think that other folks, maybe like Dario and Sam, can make small tweaks that will go a long way. And obviously there's founders that we don't even know their names yet that are going to be huge players in all of this as well.

23:58

Definitely. Before we move on, let me tell you about 11 labs. Build intelligent, real time conversational agents. Reimagine human technology interaction with 11 labs. So there's some massive news from Meta. They are doing a big deal with OKLO to build nuclear power plants. We've been following the energy story very closely this week. Obviously we're talking to Jeremy in just a little bit. We have some exciting guests next week digging into how we are going to generate more power in this country. The headline from the Wall Street Journal is Meta Unveils Sweeping Nuclear Power Plan to Fuel Its AI Ambitions. And we'll read through a little bit of this. Meta Platforms on Friday unveiled a series of agreements that would make it an anchor customer for new and existing nuclear power in the United States, where it needs city sized amounts of electricity for its artificial intelligence data centers. The Facebook parent said it would back new reactor projects with the developers Terr and Oklo, and has struck a deal with the power producer Vistra, which is up 11% today, to purchase and expand the generation output of three existing nuclear power plants in Ohio and Pennsylvania. So they already exist. There's probably some work already done on the permitting side. They're probably deeper in, but Facebook's just stepping up and saying, hey, we're opening the pocketbooks. We got the debt, we got the cash flow, we got the money to power this and take it across the finish line. So Vista and Oklo, both their shares rose about 15% after the stock market opened. TerraPower is still privately held, so no movement there. But you imagine the secondary market is booming right now. Meta aims to see the first new reactors delivered as early as 2030 and 2032, which feels like it won't matter because superintelligence will be here by then, certainly AI 2027. We're now less than 12 months away from the superintelligence, if you believe the most aggressive Possible scenario.

24:46

I mean, to give. To give AI 2027 credit.

26:48

They didn't.

26:50

To date, it's been. It's been fairly on point.

26:51

It has, it has. And no one will correct you more quickly than Tyler Cosgrove over there. The ultimate maxi. I do think we were trying to do the number of days till AGI on the ticker, and I think we got to go analog. I've been enjoying moving the goalposts and I think we need a massive flip board.

26:55

It's like the Doomsday clock.

27:14

Yeah, well, you know, how do you remember back in the old days when there was a movie theater and they would put up the letters on each of the like, if it was like Avengers, they would take an A and they would take a little sticker, sticks like suction cup on the end of a pole and they would put it up on the marquee one letter at a time. I feel like we need something much more analog to change the number of days till the singularity as we monitor it here on tvpn, its purchase of nuclear power. Yes, we're very eyes wide open that the schedule is challenging, but we think it's important to be bold. Said the director of global energy at Meta. What a gig. That's a great hitting.

27:15

I'm the director of global energy around the globe. I'm the energy director. I'm the chief energy officer.

28:00

You want to power this globe, you're going through me soon. I mean, you got to be angling.

28:06

For a promotion there because you Meta as like a nation state.

28:12

Yeah, well, so the problem is the globe. It's impressive. Meta operates all over the globe. But why aren't you thinking bigger? Who's the director of solar system level energy development, Galactic energy production, universal energy production. You should be producing energy all over the universe, Meta, you're thinking too small. Merely focusing on the globe. Hitting those timelines for new reactors would require the companies to quickly select sites that would be acceptable to nuclear regulators, start working with utilities to secure grid connections and get their manufacturing operations up and running. She said. But it would also mean they have a chance to meet the urgent demand for more electricity to fuel AI computing. And so if you think about 2020, 2032, this stuff comes online, that's great. But that feels like 2027, 2028. We're going to see a mismatch in demand relative to production. So we'll talk to Jeremy from some analysis about how we can solve that. In the interim, let me tell you about Figma before we Move on. 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 code back prototypes and apps fast Continuing OKLO and meta making this announcement 1.2 gigawatts is the total size of this nuclear campus in Pike County, Ohio. The agreement includes binding prepayment to support fuel procurement, enabling OKLO to advance early project work and secure fuel, adding new clean, reliable power to the grid.

28:16

So yeah, OKLO opened super high this morning and Then is at 115 and then it's been trading down. So it's up 7% today, but up up 28.7% in the last five days. So it's almost like somebody knew this was coming. But this was a fun article in the Journal. Were you happy to hear that AI is mining our trash for treasure?

29:49

Trash economy. Trash economy. We're going to be using cubes of trash.

30:14

We're all going to be getting trash post AGI.

30:19

Yes, yes. We're going to be using cubes of trash to. Everyone will be rich because everyone will have a cube of trash in the trash economy. The headline from the Journal is AI is mining our trash for treasure.

30:21

Plus hospitals embrace AI for better and worse. And scientists create a robot smaller than a grain of sand.

30:36

So Waste management, the largest U.S. trash hauler and recycler, is spending on building and automating recycling facilities. You have to go back and imagine a Sopranos like scenario where the mob is running trash management and just vibe coding and being like, what model should we use to detect what's in the trash? But what's actually happening here is that they have to sort out the recyclables because there are valuable things that get thrown away in the trash. And the more that you can route things to different places, the better. So it's a difficult job that pays workers little and it's hard for companies to fill. Who really wants the job of trash sorter?

30:42

That was like my first job. Picking up cigarette butts.

31:18

Yes, but you didn't have to sort them.

31:21

Maybe I should have. Maybe I should have found the one.

31:25

I never figure out that they just have that, you know that bucket that you put down and then you sweep into it. How did you not.

31:27

It was develop tools. It was a. It was like a dirt vacuum. It was like. It was like fine rock that was the ground. And so if I was doing that.

31:33

I would just be picking up rocks.

31:43

And then I'd have to be.

31:44

What about one of those Grabber tools. I still feel like at a certain point you weren't even operating at like monkey or dolphin level.

31:45

I was more like I was running. I know I actually was running around like a monkey. Like I like the speed at which I could just be like running around.

31:51

But the monkey and the dolphin, they developed tools and you, you were unable to develop tools and you suffered.

31:58

I was promoted fairly quickly.

32:02

Okay, you developed tools eventually. Thanks to recent advancements, some recyclers are now employing machines to do this dirty work. This week, Ryan December reports on the recycling companies using AI to find valuable commodities in the trash. So here's a job that computers can take without much complaint. Sorting recyclables. And before we read this, let me tell you about Lambda Lambda is the super intelligence cloud building AI supercomputers for training and inference that scale 1 GPU to 100.

32:04

By the way, do you have the honor of being serviced by Waste Management?

32:37

I don't know, actually. I think my town might have its own.

32:42

But every time I use, I interact with Waste Management.

32:45

It's good.

32:49

It's great, really. It makes me wish that all utilities were privatized.

32:50

How does that manifest?

32:54

I mean, it's just the website is great, the support is great. Everything about the experience is great.

32:55

When are you going to a website to use Interact with your trash?

33:02

Just like moving.

33:05

Oh, okay.

33:06

I need new.

33:06

If you move, there's all these things like interacting with California utilities.

33:07

My trash bin broke at one point and I needed replacement.

33:12

One click. I bet Waste Management will have an agent that Claude code get me a new.

33:15

Get me a new trash can. I suppose for humans it is a foul, laborious job that entails standing over a conveyor belt, plucking beer cans and detergent bottles from a stream of refus. The job pays little and is hard to fill. At Murphy Road Recycling's material recovery facility near Hartford, the machines are taking over the dirtiest jobs. A few workers remain on the line, mostly near the front, to watch for hazardous items. Otherwise, the system of conveyors, magnets, optical sorters and pneumatic blocks run largely unmanned. Watching over it all are computers that analyze material as it passes by. About seven miles an hour. The device is made by London based Gray Parrot. That's a good name for a company. Use artificial intelligence. The African Grey Parrot, I think is the smartest parrot, or maybe it lives the longest. Use artificial intelligence to identify recyclables, flag food grade material, gauge items, mass assess market value. They're doing DCF on every African grey.

33:21

Parrots can live to 70 or 80. Imagine just having the world's most wise parrot. Just incredible with you always, Tyler. You should get into birds.

34:24

Yeah, Bird guy.

34:33

I don't know if we're. I don't know if our office lease allows my budget.

34:35

I think an African grey parrot is expensive. I remember I was moving into an apartment building and I don't know why, but my friends were telling the landlord that we had that one of the friends that was moving in had a collection of African grey parrots, and they just thought it was the funniest thing to try and, and get this landlord to approve the African grey parrots in the apartment. It was very silly.

34:40

There's a store. There's a store in LA called the Perfect Parrot. Maybe you should go over there, Tom.

35:04

Maybe Gold Rock says Mike Rowe, is punching the air right now. Dirty jobs getting displaced so environmental concerns and the White House's push to boost domestic production of raw materials have turned attention to America's waste stream, with which is full of valuable commodities. The aluminum tariff, 50% has lifted demand for scrap metal. Meanwhile, pulp mill closures have left boxmakers reliant more than ever on old corrugated containers. And consumer good companies want to reclaim their bottles and jugs as states adopt extended producer responsibility laws aimed at reducing plastic pollution. That's good news. There's really a lot of value in a lot of recyclables and garbage, says the founder and chief technology officer at amp, a Colorado company that builds AI run recycling facilities. The problem has been that the cost of pulling those materials out is similar to or greater than the actual value of those materials. Recyclers believe that AI will allow them to efficiently mine our trash for treasure. Gray Parrot's analyzers were shown recyclables thousands of times in conditions ranging from crumpled to perfectly intact, until the computers could recognize materials. Perfect job for AI. The devices gather data about what is passing through the facility and which items aren't winding up where they belong. It's helping us make adjustments to the system. Murphy Road executives say the technology allows them to sort up to 60 tons an hour of curbside recycling. And some of these, some of the, some of the stuff that you can pull out of recycling is remarkable, especially if there's batteries that can be disassemble. There's a whole company. Redwood materials, founded by J.B. straubel, former Co founder of Tesla, early employee of Tesla, board member of Tesla, focused on recycling EV batteries because obviously we have the rare earth elements, we make the Magnets. We make the batteries and then we just kind of trash them. And if you can recycle them, that's obviously effective and valuable. Sort of an interesting second.

35:10

What do you think about the name redwood materials? Wouldn't you expect them to be working in wood?

37:09

Wood, I guess. I don't know. Yeah, I don't know why they call that. I think they're out in Vegas. They have a massive facility. I wonder how the business is doing. I know they raised a bunch of money. It was a massive capex intensive effort, but it seemed like something that was uniquely powered by seeing the rollout of Tesla. JB Straubel's working on Tesla, seeing how many cars they're shipping, watching all the batteries go out the door and just thinking, okay, well, something's going to happen with those in a couple of years. I should start building this business today. Well, Crowdstrike, Your business is AI. Their business is securing it. CrowdStrike secures AI and stops breaches. So let's move on to the big news of the day. Andreessen Horowitz raised $15 billion. Why are we here? Why did we raise $15 billion? Ben Horowitz wrote a piece on X. You can go and read it. He's also joined the show, by the way.

37:15

Redwood materials. The last round they raised was in October of 2025. I missed that.

38:09

October of 2025.

38:14

Yeah. So very recently raised 350 million.

38:15

Yes. So massive suite of new funds. The hall represents 18% of all venture capital dollars allocated in the United States in 2025.

38:18

Yeah. So that was something I was curious about. Does this get sort of, like, backdated? Like, this was entering into the data that we had from last year, and now that it's. Or was it already being counted in some way through filings? Right, because you add an incremental 15 billion funding.

38:28

Big change. I don't know, we'll have to ask them. But A16Z is now at 90 billion of AUM and it's split over a number of funds. The growth fund got 6.75 billion. American dynamism two got 1.776, I believe.

38:49

Right.

39:06

Or is it 1.176 billion? There's two different numbers here. Apps two got 1.7 billion. Infrastructure, the second fund got 1.7 billion. Bio and healthcare got 700 million. And other strategies got 3 billion. I wonder crypto is sort of missing here. I wonder if crypto is just on a different cycle or has some sort of different structure. We'll have to ask them about that. But very exciting and feels like despite the headline of venture capital fundraising declining, certainly plenty of money to go around.

39:07

Well, I think both, both Josh and Dalian this week talked about kind of the K shape and how emerging manager. Less emerging managers, less new funds. But the platforms have been doing just fine.

39:42

Yeah. Yeah. I'm curious about this. Oh, the. Okay, okay. So I understand it. So American Dynamism Fund 2 raised 1.176 billion. They already had a $600 million fund. So you add those together and you get the final amount of funding for American Dynamism across the two funds. 1.776 billion for American Dynamism. And some cool trading cards going up from Andreessen. David Ulavich shares one of an Anduril Fury drone flying across a mountain range for companies that support the national interest. Kathryn Boyle had a different graphic with a horse, which we love. Play that horse sound with an American flag. And the Andreessen new font, which is a very beveled and 3D looking metal texture. Very fun.

39:57

And Paki has a new piece on some of the history of the fund. The opening is quite fun. I'll just read it.

40:58

Before we do, let me tell you about Cognition. They are the makers of Devon, the AI software engineering. Crush your backlog with your personal AI engineering team.

41:10

Packy says Andreessen Horowitz hears your feedback that it's too loud, that it would shut up and dribble, that it should shut up and dribble, politically speaking, that you don't agree with a recent investment or two, that it's unbecoming to quote the Pope, that there is no way it will ever generate a reasonable return for LPs on such enormous funds. A16Z does hear you. It has been hearing you at this point for nearly two decades. Overnight success. And then he goes into a bunch of the history and the news. So I would encourage everybody to go to go. Specifically, there's a bunch of information on their actual returns, which is cool. But unfortunately this went out right before we joined, so I have not been able to read it yet.

41:19

But there was some spice. We gotta get to the drama. So Andrews and Horace, they put out this image why we raised 15 billion. We're all in on America. And what image do they use? They use Mount Everest. They're climbing Mount Everest. The metaphor is clear here. It's the tallest mountain. We're the tallest mountain in venture. We got the most money. We're the biggest firm. But a lot of people are saying, hey, Why'd you use a Chinese mountain? Why'd you use a Chinese mountain? It's Everest. It's over there in China. It's actually half in China, half in Nepal. It's more complicated. But maybe it's foreshadowing.

42:04

Let's not get into that.

42:35

Maybe it's foreshadowing. You know, we're acquiring Greenland. Maybe they know something we don't. Maybe it's not going to be in China forever. Maybe we already named it Everest is named it by an American who I believe was the first person to climb it.

42:36

Yeah.

42:51

And remember land acquisitions. White House officials have talked about a $5.7 billion payment for Greenland. Right. That depending on what type of payment would be needed for a place like Nepal. Right. You can imagine it being potentially less than that. So a 16Z. They've got plenty of cash. But I'm super excited to have a whole host of folks from the fund. Jen is joining Alex Rampel, David, and then Ben will cap it off at the end for his.

42:51

And Dan Primek is pulling out the spiciest quote. VC industry. Shots fired by Ben Horowitz. Ben says as the American leader in venture capital, the fate of new technology in the United States rests partly on our shoulders. That seems reasonable. They're certainly up there and leaders obviously, by what definition, but they're certainly in the top in terms of aum. And I mean, they do have a responsibility for Josh.

43:24

Josh Wolf yesterday was saying that he expects at least one or two of these larger platform funds to go public. So we'll have to get into that.

43:59

Yeah.

44:10

I mean, that's a very interesting angle with Andreessen is because it's becoming such a large, they call it platform fund. They're almost private equity type deals. We've seen General Catalyst buy a hospital network. They're at a scale where they can keep a company private until they're a trillion dollar company, but they can also buy whole companies and roll things up and incubate stuff. There's so much that you can do at this level. And it starts to look like is the comp actually BlackRock is the comp actually Blackstone. And those two firms are worth 170 billion each. Something around there. 190 billion each. And so it feels like it's going to be interesting to start seeing these firms more as financial institutions with more traditional.

44:10

Not just a firm and institution.

44:58

Yeah. Not that it's going to happen anytime soon, but it feels like.

45:00

But this feels like it could be effectively a pre IPO round Pre IPO fundraiser, maybe. We will see. We will see. But go read Paki's piece. And Sham Sankar says we should insist that all data centers that are built are architecturally beautiful in the neoclassic style.

45:03

Yes. I wanted to do an architecture deep dive. Before we do, let me tell you about Labelbox. Delivering you the highest quality data for Frontier AI. Get in the box. The label box.

45:23

In the box.

45:35

I love that they gave us enough, just enough rope to hang ourselves. So Shyam Sankar, he wants data centers built that are architecturally beautiful in the neoclassical style. Have you seen those photos of the AWS data centers that. That back up onto Virginia housing developments. So it's just like an idyllic few houses that just look like a normal neighborhood. And then just behind massive white school.

45:37

Box, they're like, I'm not leaving.

46:07

Well, now you don't even get a box. You get a tent. Because Meta is now. They gave up on their previous architectural design, and now it's just a tent, which maybe is more aesthetic. If you're gonna do a tent, Meta, I think you should make it like a circus tent. Get some red and white stripes going. Get some constant, you know, clown music going.

46:09

Get the workers in the data center to be wearing clown makeup. IRL slop.

46:32

Yeah, exactly. But there were some interesting architectural debates that I wanted to go through. Before we go to the next one, let me tell you about 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? So imagine being as locked in as the Kyoto architecture community was in 1397. I cannot believe this was built 700 years ago. 1600 or 600 years ago. Explain. What.

46:37

So there's some lore here.

47:11

Give me the lore, Tyler.

47:12

Okay, so. So yeah, 1397, when it was built, I think in maybe 1950. So it was like a temple. Right. So there's monks that lived there.

47:13

Okay.

47:23

And I think it was 1950 for.

47:24

500 years or 600 years.

47:26

Yeah.

47:28

Okay.

47:28

One was living there, and he burned down the temple, and then he tried to, like, commit suicide right outside it.

47:29

Why?

47:35

What's wrong?

47:36

I don't know.

47:37

He just wasn't locked in.

47:37

No one knows what happened. But. Okay, but then. So this is actually. It was rebuilt, and there's a good.

47:38

But it was rebuilt in the same style.

47:42

Yeah.

47:44

The architectural style is truly from 1397.

47:45

Yes. I mean, there's. I think there's some questions about how much gold was actually used in the original design? Yeah, but yeah, there's like gold should.

47:47

Be pretty fire resistant. You know, gold probably high melt.

47:54

Sure.

47:59

But I mean, it's like a very thin. It's not.

47:59

Well, that's, that's a skill issue. They should have made it out of solid gold. Then a single ember, you're trying to light it, trying to get it started. And the monk is just, I can't get this gold build.

48:01

Yeah.

48:10

There's a good Yukio Mishima book about this.

48:11

Oh, really? There's a whole book just about this story.

48:13

It's like a fictionalized story of the burning.

48:15

Interesting.

48:17

Very cool.

48:18

How would you rate this out of 10? Jordi, would you live in this Kyoto temple?

48:19

If it was an Amon, yes.

48:24

If it was an Amman, it must have spa.

48:26

I'm kidding. I do. I like the water feature.

48:28

The water feature.

48:31

I really want to bring back moats. Right. The obvious thing that's missing from modern architecture, people talk about the material use, the form factor, but the obvious, the elephant in the room is a lack of moats in modern architecture. We need to bring back moats. Gators in the moats, potentially sharks in the moats.

48:32

I feel like sharks are a little bit safer in the moat because you can hang out. Whereas if there's gators in your moat and you're on the grass next to your moat.

48:51

Yeah, people have like koi fish ponds. But why not just go size it up a little bit, Go for the shark. Shark pond. Shark pond, shark pond. Imagine people, people go out, they like the being relaxed and kind of like feeding the koi. But imagine just throwing chicken breasts into the water for a shark. How relaxing that would be if you needed just 15 minute break from work before you get back to your email job.

49:01

Well, if you're building a fintech company, you need a moat. You need plaid. Plaid powers the app you use, the apps you use to spend, save, borrow and invest securely. Connecting bank accounts to move money, fight fraud and improve lending. Now with AI, what about the architecture at the Charles de Gaulle airport in France? It's a cross section of a wing, complete with spar box. I didn't know what that means. Need Billy Thalamer, founder of Regent Aircraft, been on the show. He says we need more buildings modeled after airplane parts. I've seen this as well. In a house, someone took an airplane wing, a physical real airplane wing, and built it into a house that sort of served as the unifying ceiling. What do you think about living in an Airplane wing. Did you do that? If it was cool.

49:26

I've seen. I feel like I saw a house at some point that was built around 747.

50:24

Yes, yes. Well, you also might be thinking of John Travolta's house, which just often had a 747 parked outside, which he flew himself, which is incredible. Imagine being so into private aviation that you learn to fly. Massive.

50:29

Okay, I found it.

50:44

And then land them at your house. With your own. With your own.

50:45

I put it in the timeline. Guys, if you can pull it up, there's something called the 747 Wing House in California.

50:48

Yes.

50:55

Let's pull up this. Let's pull up this.

50:56

I think I've seen this on a show called Grand Design on Netflix. I believe I watched a little mini documentary or an episode of reality TV about this. Let's see. Yeah, this is the exact thing.

50:57

I don't come across views like that more than once in a lifetime. It is somewhere that once seen, it.

51:09

Would never reveal it. Reveal it.

51:15

Reveal it.

51:16

Enough of the. We get cactus outside, running water or even roads.

51:17

Is an endurance test.

51:22

Only the Braves.

51:24

There it is.

51:26

Look at this thing.

51:26

Interesting.

51:28

Oh, very cool. That looks really cool.

51:29

We don't know how to make that really cool.

51:32

You know what this is? Every time you watch one of these episodes, it's always just some couple that's just maniacally focused on making this particular thing. And they interview them and it takes like years to create one of these episodes. Interesting. It's blurred. I wonder why. That is very, very odd. But it's always someone has a vision and then they think it's going to be easy. And then they spend a decade building it and they have to check in with them. We checked in with them four years later and they were still in production and still getting permits. But eventually it does get built and it looks beautiful when it does.

51:35

You got to really love the 747 to be in love with this house. I think it looks very cool. I'm glad they did it.

52:08

I flew on a 747 for maybe the first time to Europe. It was beautiful. It was amazing.

52:15

The first time.

52:20

Yeah. 747s are pretty rare. Like, except for long haul international flights.

52:21

Yeah. You're an America guy.

52:26

I'm an American. So I'm usually 737 Max. Always. Or Airbus. But the 747, it's got a special. It has a special aesthetic to it because of the smooth bump. You get the second story, but only for the first half of the plane. I mean it's the plane the that we use Air Force One for. That we use for Air Force One. Whereas the competitor which I believe is the Airbus A380 doesn't have the same aesthetic beauty. It's two stories the entire way. It's very efficient but it just doesn't have the same clean line as the 747 which is just so, so iconic. Anyway, Charles de Gaulle airport go check it out. Also check out restream one livestream 30 plus destinations. If you want a multi stream go to restream.com.

52:27

So checking in on 262-5th Ave. Yes sage Hunter Bornstein says should be raised. It's a crime. It should be raised. Shame on S L C E F.

53:17

He really dug into who built this house.

53:29

Wow. Moscow based.

53:30

He's going at him 26 residences. He calls it hideous architecture. He says it takes away Madison Square Park's views of the Empire State Building. I don't know what is that at the top? That top is.

53:32

Don't worry about the gold cube, John. Don't worry about it. Don't worry about it. Don't worry about it. It's just. It's a gold cube.

53:46

Wow.

53:55

We found a Rare post here. 0 likes 166views wow. First, except for the fact that I disagree with it, I think it's actually sort of a nice building and I think we just, we generally need more buildings.

53:55

I think it just feels harsh because of the contrast to the building next to it.

54:06

True, true. But I don't know. We'll read this in the mansion section. But there's some interesting Dynamics about how HOAs enforce aesthetics in communities and whether or not that's good or not. In other news, they 3D printed a Starbucks. Starbucks has a new drive through in Texas. The Coffee Giant's first 3D printed store in the United States.

54:09

It's funny, they really made it look like it's 3D printed.

54:33

Look at this, look at this. So I mean I've talked to one of the. There's a YC company that does this technique. I know it looks so sloppy and there's a little bit of like up at the top. It actually looks like there's a little bit of imperfection and randomness. That looks sort of aesthetic. It looks like it's sort of designed in the way that a log cabin, not every log is going to be perfect. But then you get to the middle section and it's like the tube that was pumping, it was just not working. If you scroll down over to the left and then down a little bit, just down like that is messy. So this the way it shows up. You basically get like a crane with a gantry that can move the nozzle in an X and Y axis and it just pours cement in loops, circles again and again and again.

54:36

Okay, I need to know how quickly they built this and how much it costs. Because if this came in at 80%.

55:18

They said it was two GS. Two grand. Now, can you imagine?

55:25

Starbucks was down to their last two.

55:30

Grand and they're just like, yeah, let's 3D print it. Yeah, I don't know, around 1.2 million. Okay.

55:31

That's what is a normal. What is the average Starbucks cost, standalone building cost?

55:38

Tyler asks who is the architect and Pete says, Slop GPT. People are not happy.

55:43

The U.S. jason says, you wouldn't download a Starbucks.

55:51

U.S. graphics Company says, I have the sudden urge to insult this in biblical terms. Yeah, People are not very happy with this. It does feel like a low quality print. Hopefully the 3D printing company, this is just a step in the road and they become more aesthetic, more precise, I think.

55:56

Okay, so the total investment range for Starbucks location is 760,000 to 2.2 million.

56:14

See, that's not.

56:21

It's kind of right in the range.

56:23

There's. It's the median cost, but one of the more ugly ones.

56:25

Whether you're long or you're short, Starbucks, you got to do it on public.com investing for those who take it seriously. Stocks, options, bonds, crypto, treasuries and more. With Dan. Great customer service.

56:33

Matt Stevik says they saved 10 bucks.

56:44

Good to see you, Matt. The wrath of Non says is comparing some architecture in Oslo, Norway.

56:49

Wait, going back to the 3D for a second.

56:57

Let's go back to.

56:59

Could they not have found a material to just place on that? Like, it seems like this could be a great way to build like the core structure.

57:00

Yeah.

57:07

Couldn't you just plaster over, Put some.

57:08

Yeah, plaster over. It just.

57:09

It seems like they wanted to really throw it in, prove that it was really put it in your face. The whole point of technology is not the technology itself.

57:11

Yeah.

57:19

It's what it enables. Yeah. So if you can build a Starbucks for half the cost, you know, twice as fast, that's amazing. But it doesn't mean it has to look like.

57:20

Well, yeah, it was 3D printed. You can plaster and spackle over any sort of rough material and then you can print. You can basically stamp like brick texture into it. Or some cinder block texture into it and it's fake. But it looks like. What? It looks like they didn't do any of that here. They really let the, like the loose tubes really lay out.

57:27

It looks just like. It looks like. It looks like.

57:49

Looks like a gingerbread house.

57:52

It looks like a kid's. A kid's school project.

57:53

Yes, yes, yes. Today we're using.

57:58

And you gotta be like, oh, nice.

58:00

Nice work, nice work.

58:02

There's like tongue depressors and chopsticks or popsicle sticks. Popsicle sticks and glue. And the four year old went a little crazy with the glue. The Elmer's came out out in full effect with this Starbucks.

58:03

That being said, cool that this is actually happening.

58:15

Yeah.

58:18

Because this has been promised for a while. I just hope they refine the design.

58:18

At the same time. I've talked to a number of startup founders who operate in the.

58:25

Trey says you can also build a Starbucks out of mud, but probably you probably should.

58:30

Oh, wow. Okay. I've talked to a number of startup founders working in trying to develop cheaper housing, cheaper building materials. And they've all said the last thing that you want to 3D print is just a flat wall. Like 3D printing is great when you're talking about Lucas Zinger's Hypercar. And you need some crazy structure that can't just be milled. But we're very, very efficient at making flat planes. You can just take a metal cube and slice it. You can take a bunch of wood and build a grid. Like we're pretty efficient at just building flat walls. You don't actually need 3D printing for that. You need 3D printing for building some sort of special structure. There was a rocket company. Was it Relativity? I don't want to get it wrong, but there was a rocket company that was saying, hey, we're going to 3D print rockets. And it would go and solder one piece after another in a cylindrical turb. The only problem is that we know how to make cylinders. Really, really effective. Yeah, we can just bend metal and so that's what Blue Origin and SpaceX do.

58:36

What Rocket says. Lincoln Logs Starbucks.

59:41

Lincoln Logs Starbucks.

59:44

Soft serve building.

59:45

It does look like soft serve. That's right.

59:46

It's a good Starbucks.

59:48

Yeah. Maybe it's an ad for the McFlurry. Maybe the McFlurry machine always works there.

59:49

DG says you could do a sandcastle.

59:52

Starbucks Sandcastle Starbucks might be fun.

59:54

That could go pretty hard.

59:56

Well, maybe they need to do a sandcastle National Museum over in Oslo, Norway. But before we dig into this. 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. So he's comparing two images. One is from 1882. It's the National Gallery. And in 2022 they launched the National Museum. And he claims that modern architecture is meant to demoralize you. And you can zoom in on the side by side here. The 1880s building is very ornate with lots of gold and brick and details and structure and windows and all sorts of things. And the 2022 building looks like a black cube from the Borg or Star Trek.

59:57

Really rough.

1:00:45

Some cool stuff. You can do cool things with flat materials. I've always liked the design of the Walt Disney Concert hall, but it looks like this sweeping winged structure. It's very innovative and unique. This is a little boring, guys. But at the same time, Oslo, it's a gloomy place. It sort of fits in, I guess. I don't know, maybe the architects were depressed. They had seasonal affected disorder. Sad.

1:00:46

Anyway, Steve is sharing, apparently this was a proposal from a Norwegian architecture firm for the Obama Presidential Library in Hawaii, but it failed to.

1:01:11

Okay, so the Norwegians got it.

1:01:22

They still got it.

1:01:23

Yeah. Sometimes it's the flat black cube. It's interesting because, like the critique of the new Obama Presidential Library is that it's too Norwegian. And it turns out he just went with the wrong Norwegians, I suppose because they're sharing these sort of science fiction looking renders. These look beautiful with a wonderful water feature and this sweeping grass overlay. That seems very cool. Although it feels like you would be at risk of falling off the edge here. They would have to put a railing of some sort. But very, very cool architectural designs. And I would love to see more.

1:01:24

Pull up the Obama Presidential Library.

1:01:57

Does Trump get two presidential libraries, one for each term? Has he already built one? I know the plane's going there, right? Something like that.

1:02:00

I put the Obama one in the.

1:02:08

But if you're worried about a lack of gold in the presidential library architectural industry, I don't think you need to be worried for very long. I think there will be a very ornate building coming in just a few years probably. And I'm sure the.

1:02:11

Yes, this is the. This is what? Yeah, this was the winning bid.

1:02:24

Yeah. And they did that.

1:02:29

This one is in Power Source, right? This one is in Chicago. This one's in Chicago, not Hawaii.

1:02:30

Yes.

1:02:35

And again, maybe if they get a huge projector and put like a fireplace on the wall inside it could be cozy, but that doesn't look crazy.

1:02:36

How tall?

1:02:46

It looks like a great place, potentially like the. Like if the deep state felt like not super welcome in other buildings.

1:02:46

I want to see the aircraft that they're building inside there because that looks like a aircraft hangar for a ufo.

1:02:56

Yeah.

1:03:01

It seems of alien origin. Well, Wrath of Non is continuing to talk about architecture. He says traditional Korean architecture with its visually rich, harmonious patterns produces lower levels of visual stress than modern facades with repetitive patterns, hard lines and high contrast materials, which are more likely to overload the visual system and contribute to discomfort, especially in dense urban areas. There's some research that suggests that having variation in your architecture actually can reduce stress, which is fascinating.

1:03:02

Yeah.

1:03:42

Try to zoom in on this picture on the left. If you click in because you can see the one on the left. It's way for something. It just feels more organic or natural, right?

1:03:42

Yeah. And it's a pretty simple shift.

1:03:51

And then if you go over to the right a little bit, it doesn't feel like they. Okay, that looks like a prison on the one side. The left looks like they're gentrifying some area and on the right, some part of like Mexico City. And on the right it looks like a prison. And it's only a slight.

1:03:53

Or a bank building in New York City. But yeah, it doesn't seem like it costs that much to create some variation and randomness in your architectural designs. Maybe it's expensive, but I want to.

1:04:09

See the new A16. I think A16Z needs to build like a massive gold superstructure in the heart of San Francisco. Just carve out. Who knows what the fee structure is.

1:04:20

But take half of it, take half.

1:04:33

Of the fees and just build a monolithic monument to.

1:04:35

Have you ever been to the Statue of Liberty?

1:04:40

I've never actually toured it.

1:04:44

So there's a structure. You can go inside of it and walk up the stairs and whatnot. You could put offices in that. You could build a statue. And then to put your office inside the statue, that's thinking.

1:04:47

I like that.

1:04:58

What do you think?

1:04:59

I think we should kind of look back to Charlie Munger's design for the ucsb.

1:04:59

The lock in Dormzilla.

1:05:04

He just wanted. He just wanted to lock in.

1:05:07

He wanted everyone to lock in. And he was killed for it. It was ridiculous. He was.

1:05:11

Pull this up. University of California abandons plans to build windowless dorms. The Munger Hall. He just wanted you to lock in. But pull up this article. You can see the design and just how many, just how Many rooms in this place are windowless. It is very, very powerful. So on the outside it looks like. It looks. I think he just knew that people were only going to have a few by the time this was built. People would only have a few years to escape the permanent underclass. And windows would distract people. If you scroll down, you can see the actual design.

1:05:17

Yeah.

1:05:56

There'S a lot of windowless things.

1:05:58

Maybe he was just super pilled on VR at the time. He was like, everyone's going to be in the metaverse. Everyone's going to be locked into their VR headsets.

1:06:01

Maybe Aqua says Munger was cooking. Yeah, we should have let him cook.

1:06:08

We support him anyway.

1:06:12

Anyway.

1:06:14

Gusto. The unified platform for payroll, benefits and hr built to evolve the modern small and medium sized businesses.

1:06:15

Yeah, I mean, the funny thing with trying to do that at UCSB is like the most popular freshman dorms at UCSB are actually like in an old hotel. So it's these two towers that have these incredible 360 views. Not 360, but like the mountain and the ocean. And then you have like this huge like Olympic swimming pool. So it actually feels like you're just staying at like, at like a Hilton or something like that. So to go from that to windowless is just a little bit rough.

1:06:25

Yeah, well, underrated strategy. Buy a compound with your friend. Do you see this? Former casino mogul Steve Wynn and financier Thomas Peterfeed set a record when they bought a 4 1/2 acre compound in 2024. I can't seem to add this to the list, but it's in the Wall Street Journal today. It's on the COVID of the Mansion section. It's comparing Aspen to Palm beach, but it's very funny. It says five years ago, $20 million home sales in Aspen were rare, happening no more than than a handful times in a year. An influx of the of uber wealthy buyers has now upended that norm, handing the affluent ski area 34 deals above 20 million last year, up 161% from 2024. Year over year growth almost tripling in that category. The median single family home sale price hit 13.95 million during 2025's third quarter quarter, compared to 9.97 million in palm Beach, Florida. Raising the question, has Aspen eclipsed Palm beach as one of the priciest markets in the country? Long known as playgrounds for the rich and famous? One sunny, one snowy. Aspen and Palm beach are increasingly two sides of the same coin when it comes to luxury real estate. Despite 2,000 miles between them. Each is home to dozens of billionaires. Restaurants like St. Ambrose and boutiques like Brunello Cucinelli have outposts in both locales. Some of the wealthiest people have property in both places, too. People run in packs and they run to the same destinations, said Palm beach real estate agent Dana Koch. The meteoric rise of Aspen's ultra luxury market has made comparisons between the markets unavoidable. Both Palm beach and Aspen saw deal volume and prices soar in 2020. That's continued to be driven by strong financial markets. The 1 percenters are making money hand over fist, said Aspen real estate appraiser Randy Gold. Real estate, he said, is a hard asset that they can enjoy. Both are small markets that are protected geographically, adding to their exclusivity. Palm beach is an Island. Turn to M9.

1:06:54

Aspen is basically an island in the mountains surrounded by public land. Once you're here, it's very private. Each has a limited supply of homes. There are only so many beachfront properties in Palm beach and or homes on Aspen's Red Mountain, which has fueled major price appreciation in both. The numbers are mind boggling, but the reality is that is these properties are unicorns. When they come up for sale, you have buyers out there who will pay a premium.

1:09:20

Okay, so, so Steve Wynn and Thomas Peterfy, they both own homes in the same area in Palm beach, and they're known to be friends. And they're both GOP donors, they're neighbors in Florida. And apparently they became close enough to go in on $108 million Aspen Estate together. And so Wynn is the founder of the Wynn Resort and Casino in the Bellagio. Thomas Peterfee is the founder, chairman and largest shareholder of Interactive Brokers. So he's made a ton of money. Thomas is worth 35 billion. Steve Wynn is maybe worth 3.4 billion. So one order of magnitude gap between them. And you wonder if it's Wynn being like, I gotta bring in some extra firepower on here. Let's go 50 50, brother.

1:09:43

Well, here's the thing.

1:10:30

This sounds great in theory, and then it's New Year's and boy, both families want to be in Aspen, but it.

1:10:32

Is 22,000 square foot compound. You might just say, hey, we're doing Christmas together, New Year's together.

1:10:39

I know, but there's probably only one bachelor actually finding families that want to go be in a cabin.

1:10:47

Oh, it's a tall order. It's crazy to be like, oh yeah, like bro, little bro, you're only worth 3.4 billion. Hit the couch. I got the master.

1:10:53

You got the couch.

1:11:03

Oh. You only put in 50 million into this compound. But this is where it gets funny. So no one. There's no reporting on, like, why they decided to buy this together, but we have a clue because the buyer is actually an LLC that they set up. And the name of that LLC is Buddies. Aspen Buddies. We're Aspen buddies. We're a couple of Aspen buddies. And we went in on it together. And so they bought an Aspen mansion together. It's a very funny time. The other buddies who did not buy a mansion together are, of course, the Google co founders who have both bought property in Miami. Larry page just spent 173 million on two Miami homes. Some of them are very odd. We'll go into these. First, let me tell you about Fin AI, the number one AI agent for customer service. If you want AI to handle your customer support, go to Fin AI. So Google co founder Larry Page has bought two massive Miami estates for a combined $173 million, according to People familiar with the situation. The deals come as Page and other Silicon Valley moguls descend on Miami in the face of California's proposed tax on billionaires. Deleon called it just a little bit too early. But the move is finally happening. Page paid $101 million in December to buy one of the properties, a waterfront compound in Coconut Grove that had long been the home of the late restaurateur Jonathan Lewis. He then purchased a nearby Coconut grove property for 71.9 million from heiress Sloane Lindeman Barnett and her husband Roger Barnett.

1:11:04

Okay, and let's pull up. There was some interest in one of these homes that Larry bought. We can pull it up. I put it at the bottom of the timeline. Yes, but one of the homes that he purchased looks fantastic. Looks like this jungle oasis. And then you zoom in a little bit.

1:12:39

Do the csi. Enhance, enhance.

1:12:57

Enhance it a little bit.

1:12:59

And it gets a little odd.

1:13:00

A little odd, haven't.

1:13:03

It?

1:13:09

It's not. It's just a sculpture. We can hopefully.

1:13:09

Let's see. This is going to be a jump scare if it's anything but this.

1:13:11

Yes. So this. So you have to show the zoomed out picture.

1:13:13

So scroll down, show the zoomed out, out photo. Oh, wait, no, it's not there. Oh, don't, don't scroll through too much. We need to find the original photo.

1:13:17

The, the original photo is just.

1:13:25

Where did the original photo go? We'll find that and we'll tell you.

1:13:27

Put it here.

1:13:31

But if you. But, but basically there's a, there's a, you know, it's a stock. It's a real estate photo. It's a photo that was taken clearly. He was a real estate agent hiring.

1:13:32

Before it was Larry's house.

1:13:40

Before it was Larry's house, to be clear.

1:13:41

So this is the photo. This is the photo.

1:13:43

And then people started zooming in and you keep zooming in and you keep zooming in and you keep zooming in and you start seeing some very weird design and decor interior decorating decisions that tell a little bit of a weird story about whoever was here. Usually these properties are staged before they are photographed.

1:13:45

So who owned that choice? Who owned it previously?

1:14:04

Also, it's something that.

1:14:08

Tyler, can you try to find out the previous owner of this house?

1:14:09

I don't know. So, I mean, we do have some. Jonathan Lewis is one and Sloane Lindemann Barnett and her husband Roger Barnett are the others. If this is Larry Page's new house.

1:14:12

But I know Sergey also some guy named Jeffrey.

1:14:24

Oh, no, no, no, no, no, no, no, no, no. And they declined to comment. California's proposed ballot initiative would impose a one time 5% tax on the assets of billionaires who would retroactively apply to those who were California residents as of January 1, 2026. Now this proposal is not even on the ballot yet. It's still being workshopped. But many tech billionaires are not taking any chances and they're relocating as of January 1st or December 31st in many cases.

1:14:27

December 31st was really the day to send the press release.

1:14:59

It was. It was pretty much every other day I'm showing property to a client from the San Francisco Bay area, says Dina Goldentyre of Douglas Elliman. Every conversation I overhear, they're talking about the wealth tax and how it's retroactive. They're in a hurry and they're all looking at the same houses. Several agents told the Wall Street Journal they couldn't talk about the deals because they'd signed non disclosure agreements that could end their careers if broken. The former Lewis compound spans about four and a half acres on Biscayne Bay and Coconut Grove, one of the city's most coveted neighborhoods. Neighborhoods. Lewis was the son of the late Peter B. Lewis, one time CEO of Progressive Insurance Company. What a great business. The property, listed for 135 million in 2024, was most recently asking 115. The estate has two primary residences. One was designed for the Secretary of State, William Jennings Bryan in the 1920s. And Lewis built the other home for his father around 2002. The other property purchased, purchased by the Barnett's for 45.9 million in 2021 is less than a mile away. Sloan is the daughter of billionaire George L. Lindemann and Roger is the chief executive of health supplement maker Shakely. I don't know Shakley. Actually the Barnett sold the San Francisco mansion to a billionaire, Lauren Powell Jobs, wife of the late Apple visionary Steve jobs, for around 70 million in 2024, setting a record for the cash California city. The Barnett property wasn't on the market at the time of the deal, so little information is available. Real estate data website Property Shark shows it was built around 2015 and spans 17,000 square feet with seven bedrooms. Page and Sergey Brin founded Google as Stanford students in 1998 and built it into the one of the world's most valuable companies, or one of them. Step back from active management in 2019 and maybe getting back in the arena during the AI business boom. Page is worth around $270 billion, according to Bloomberg. In addition to his home in Palo Alto, he obtained New Zealand residency in 2021. Miami's ultra luxury market has skyrocketed in recent years. In 2025, there were 19 sales above $50 million in Florida, compared with just 12 in New York and 10 in California. Miami had four deals above 100 million last year. Billionaire hedge fund hedge funder ken Griffin paid 106 million for a combined state.

1:15:02

Miami really makes Southern California look cheap.

1:17:22

It does.

1:17:25

Ask a Miami guy to watch him browse. Watch him browse Zillow.

1:17:26

Yeah.

1:17:30

And they're like why they're giving the houses away here.

1:17:31

Well, let me tell you about Vibe Co, where D2C brands, B2B startups and AI companies advertise on streaming TV, pick channels, target audiences and measure sales just like like on Meta. And you know what else is surprisingly expensive? Nashville. I had no idea. But there is a Nashville condo that quote where it's James Bond meets Lenny Kravitz and it just is for sale now for 33.5 million. So looking to hang out with John Fiorentino out in Nashville? You gotta pick this one up. It's 5,000 square feet.

1:17:33

Wait, doesn't John live in the Four Seasons? I think he might not to docs.

1:18:09

But if you want to get Volta from the Source, you got to head to Nashville and hit John Fio on X with a DM In Nashville, where luxury home prices have skyrocketed in recent years, a penthouse at the Four Seasons hotel and private residences is aiming for a record 33.5 million. Asking price. The condo unit is the city's most expensive home for sale. The penthouse is also available. You can rent it. If you don't have 33 million to spare, it's gonna cost you $200,000 a month, and you gotta commit to at least six months. But if you have 1.2 mil burning a hole in your pocket, you want to hang out in Nashville with one of the greatest idea guys to ever do it. John Fiorentino, maybe.

1:18:14

It's just funny. I just realized that the guy selling this house is a buddy of mine.

1:18:57

Oh, really?

1:19:02

I actually bought my house through his office.

1:19:03

Jamie Parsons.

1:19:05

This is Cortazo.

1:19:07

Oh, okay.

1:19:08

Yeah. The newly completed Four Seasons in downtown Nashville near the Ascend Amphitheater. What? Ascend Amphitheater. They're ascending out in Nashville.

1:19:09

Whoa.

1:19:20

I know.

1:19:20

Is that where you spent winter break? Yeah. Yeah.

1:19:21

The Ascend Amphitheater just looks maxing. John has been looking good recently. He must have been spending some time there. It's a few minutes walk away from Broadway and a thoroughfare famous for for its bars and live country music. About two miles away from Taylor Swift's longtime home in the city. The penthouse spans roughly 5,000 square feet with three bedrooms and a balcony lined with floor to ceiling windows. The unit has expansive views of the city and Cumberland River. Chris Cortazo, a Malibu, California real estate agent, bought the penthouse for 12 million in 2022. When it was just a shell. There were no walls. Kitchen, he said. There was nothing there. He spent two years building out the unit in an aesthetic he calls James Bond meets Lenny Kravitz, anchored by a circular floating fireplace.

1:19:24

Wow.

1:20:11

Floating fireplace. That's cool. The residence has about a million dollars in smart home technology. He lives full time in Malibu, but began visiting Nashville around five years ago. Cortazo did. He also owns a roughly 150 acre farm outside the city. He said in his spent spent only a few weekends at the penthouse since completing it earlier this year. Cortazo is also listing the Malibu home of the late actress Shannon Doherty, a longtime friend of his. Oh, I heard about this news that that property hit the market last year for 9.45 million and has a listing price that has since been reduced to 8.75 million. Nashville's luxury real estate market has surged over the past six years, according to Parsons. While only one home traded for over 10 million in Nashville, Krueger said that she believes the figure almost hit 20 in 2025. If Cortazo. If Cortazo's home sells at or near its record asking price, it will set a record for Nashville. The current record is held by a suburban home on 50 acres that sold for 32 million in 2024. So if you're looking to head over to Nashville, pick up this penthouse. I've never lived in a. A penthouse, a condo like this. I would.

1:20:12

Then you've never lived.

1:21:23

I've never lived. But personally, I would be going for the 50 acre suburban home. Too many dogs, too many animals around. I want the steeds around. I want to be able to ride the horses. But for the right person in the right time of their life, I'm sure this is a fantastic pickup. In other news, Phantom cash fund your wallet without exchanges or middlemen and spend with the phantom cast.

1:21:24

France will delay the G7 summit to avoid conflict with mixed martial arts.

1:21:50

Really? On Donald Trump's birthday.

1:21:56

Yeah. So they're going to delay Group of Seven summit to avoid a conflict with the mixed martial arts event planned at the White House on Donald Trump's birthday.

1:21:59

Do you think they have a group chat for the G7 and it was like, hey, can we get together? We got to get together.

1:22:07

We got to get together.

1:22:12

We got to do a summit.

1:22:13

It's been too long.

1:22:14

G7 leaders, let's do it. And Trump's in there. Just being. I already told Dana I'm in. I'm watching ufc.

1:22:16

It's happening in my house.

1:22:26

It's happening in my house.

1:22:26

It's gonna be really awkward if I'm not there.

1:22:27

I gotta be there.

1:22:29

I gotta be there.

1:22:30

Plan another summit, another time. This is interesting. Usually there's some big news that comes out of G7 summits. Usually there's all the leaders get together, they're striking deals, they're taking, talking to each other. There's something going on.

1:22:30

Yeah. So definitely worth still doing. Worth delaying, but worth doing well.

1:22:44

Speaking of Trump, Lip Bhutan is in his good graces. Donald Trump post his new best friend on Truth Social. He says, I just finished a great meeting with the very successful Intel CEO Lipp Bhutan lbt, as he's called. For those who know, intel just launched the first sub 2 nanometer CPU processor designed, built and packaged right here in the US of A. The United States government is proud to be a shareholder at intel and has already made through its USA ownership position tens of billions of dollars for the American people in just four months. We made a great deal and so did Intel. Our country is.

1:22:50

You're kind of slipping into the.

1:23:27

Oh yeah. By the end of the year I'm going to have it down. Our country is determined to bring leading edge chip manufacturing back to America and that is exactly what is happening. And people are asking for particular financial advice which we won't, which we will not give. But Liputin also sat down with Howard Lutnick. Howard says just four months after the United States invested Intel, that investment is already delivering tens of billions of value to the United States people. That momentum continues with Intel's new 1.8-nanometer processor and a major step towards bringing semiconductor manufacturing back home. Let's play this clip from Lipputon talking to Howard Lutnick. I have the pleasure today of welcoming Lipp Bu and he's come to the Department of Commerce to update us on how intel has been doing since we made our historic investment in the company. Thank you so much. I'm so delighted to come over here to CEO.

1:23:28

We double the market cap.

1:24:28

It's over 200 billion and also very exciting.

1:24:29

We announced our products first time on the 18A production in Series 3 processor with multiple of our customer in us.

1:24:34

Globally and using that is the most advanced process design and also using our.

1:24:47

Most advanced music productions.

1:24:53

Right.

1:24:56

So for the United States we, we love the fact that intel is doing leading edge work in America. Right?

1:24:56

18Ameans 1.8 nanometer. 14A is 1.4 nanometer main.

1:25:03

Think of how incredibly sophisticated and tiny that is. And then packaging, you take this little.

1:25:08

Little, little, little teeny thing and how.

1:25:16

You put layers and layers of sophisticated, sophisticated circuitry on top of it and you do it with, you know, just the most amazing technology. Doing that seems like it in America. Leading in America by a U S company. Get yourself an investor that talks about your company like this. So we're proud of you. This is okay.

1:25:19

We're rooting for you and we need.

1:25:39

You to be successful for America. Thank you so much. This is the new startup launch video. You do a deal, you raise some money from a VC and you got to put out a video like that. You guys sitting down on the couch next to each other with succession music and they explain your business, shake your.

1:25:41

Hand and there's so many new formats.

1:25:55

Yeah. New format, unexplored, unexplored clone that one call video production team. Get it done. In some other news, US oil executive is commenting on Venezuela. No one wants to go in there when a random effing tweet can change the entire foreign policy of the country. And deep dish enjoyer says lmao. It is remarkable how online this administration is you saw during the, you know, people have been celebrating the death of X and yet during the invasion in one of the images of the war room in the background on the TV was just. Just axe.com search for Venezuela. Let's see what people are saying. Remarkable Times Week.

1:25:58

Next time they'll probably have Reggie's monitoring the situation dashboard.

1:26:44

They should have done that. They should also be using Railway Railway simplifies software development, web apps, servers and databases run in one place with scaling, monitoring and security built in. So I posted a piece about the Apple card and how it's changing from Goldman Sachs to Jamie Dimons JP Morgan of course and a software architect for Goldman Sachs's consumer bank. Matt Lowe chimed in and says good article. I think the main reason it didn't work with Goldman Sachs is that it lost its high priority in a double whammy. CEO David Solomon used a lot of his power to evolve the partnership culture to an exec first modern corporation. That's. That's interesting. Through that reorg, I speculate he had less political flexibility to defend Marcus from other partners who didn't want to wait for the consumer bank to scale up. Marcus even grew through leadership turnover during the pandemic and made a huge acquisition of Greensky. It seemed to fizzle out and lost the continued executive sponsorship it needed to keep going while reporting to an asset and wealth management division. Note in a couple in last couple in last couple weeks podcast with Sequoia Capital, Solomon said the main reason for the wind down was regulatory. Interesting. So just some extra context around the Apple card. Well, let me tell you about graphite and then we'll bring in our first guest of the show. Code review for the age of AI graphite helps teams on GitHub ship higher quality software faster. And without any further delay, let's bring in Jeremy from semianalysis. How are you doing? Welcome back again. Welcome back, second time on the show. Excited to have you here. Happy New Year.

1:26:49

Happy New Year.

1:28:31

Happy New Year. How are you? Are you shocked to see that the G7 that's supposed to happen in France? Aren't you in France right now?

1:28:32

I'm in France right now, yeah.

1:28:40

Well, you're gonna have to wait for the summit because UFC takes priority in America, apparently. I don't know if you saw that. Anyway, congratulations on the new article. I would love for you to set the table for us and explain sort of what were the questions that you were trying to answer? What was the overall thesis that you came into this particular article, how AI labs are solving the power crisis. And then I have a whole bunch of questions that I want to dig into.

1:28:42

Yeah, look, the question that we keep receiving every day, every hour it seems, is how are we going to power the AI race? Is the grid able to handle all of that? That and look, last time I came in, I think I said, hey, there's like over half a terawatt of data requests in all of the US grid, an insane amount of requests. And we talk about sort of prisoners dilemma where because everyone is trying to find power, then it creates sort of a vicious cycle of everyone's thoughts being more speculative and putting requests everywhere. And so basically the grid is overwhelmed, people can't find energy. And so that's why you're seeing the rise of onside gas, which is something that a lot of people have been talking about. But from our research we just haven't found sort of any good way to understand what are people actually doing, what are the challenges, what are the systems that people are actually deploying, what's the benefits, what are the trade offs and how to understand how to make sense of all of these new entrants. Because one of the key highlights is, okay, everyone talks about GE Vernova, everyone talks about Siemens Energy, but we count actually 12 manufacturers that have secured orders of over 400 megawatts for US data centers on Sagas Power. There's way more people in the pipeline. But essentially we wanted to show people how are the labs solving the power crisis? Talk about. To some extent, XAI was a bit of a pioneer because obviously it did that before everyone. And how are the other players following suit and actually solving this issue?

1:29:11

Yeah, let's stay with that point about power requests from AI companies. That's expanded significantly. Significantly. Correct, isn't it? It's over 10 terawatts now or something like that. Everyone's spamming with these requests, isn't it? It's an insane figure. Is that roughly correct or.

1:30:35

I don't know, about 10 terawatts.

1:30:53

It's over 1 though.

1:30:55

Over roughly a terawatt.

1:30:57

Roughly a terawatt.

1:30:59

It's always complicated to know exactly. Yeah, but roughly a terawatt.

1:31:00

And what are the mechanics of a request for power? Is that going to, to government run organizations? Is that a permitting process? What is the anatomy of actually making a request for power if you're an AI company?

1:31:04

Sure. So typically you send a request to a transmission provider, so say American Electric Power, the largest in the U.S. you send them a request, I want power in Ohio. You have to fill Whatever, some kind of form, you tell them what you want by when you want it it. And then if sort of that first phase moves through, you have to go through a system level study. And the reason we have to do that and that process, the reason it takes time is because the way the grid works is demand and supply have to be always perfectly synchronized. And if they fall out of sick because there's too much supply or too much demand, then there's basically blackout for everyone. That's the worst case scenario, to be clear. But it's possible. It has happened. It happened in Spain about a year ago because of an issue on the supply side. And so the implication is that if you want to interconnect a 1 gigawatt data center, you're going to have plenty of system level studies that are going to slow down the process. And this is where you get into that sort of vicious cycle, because everyone is sort of putting requests because they know it's going to take a while.

1:31:22

Isn't there some people putting in requests just from a speculation standpoint? They're just like, hey, I know that if I get access to the power, I can resell it it to somebody else and if I just kind of lock into one of these deals, grids basically giving me a contract, then I can go and flip it. Is that happening?

1:32:24

Everyone wants megawatts these days, or gigawatts, so you try what you can to get some. And if you're for example, say you're based in Ohio, you typically operate in Columbus, Ohio, then suddenly you tell your utility, hey, I want another 500 megawatts. And they respond, actually, I've had like 10 other people ask me this, so you're going to be the queue and there's 10 other requests I have to evaluate before I get to yours.

1:32:44

But this is why some of the bitcoin miners have done actually very surprisingly well in just sort of like pivoting to AI because they had the pre existing power deals in place, right?

1:33:11

Yeah. It's not in the queue. It already exists because the queue is just an evaluation. But then once you have everything approved, you have to build a substation, you have to actually interconnect your data center to the grid. And the bitcoin miners, they already have the substation, they already have the transformers on site, they already have the energy flowing, the electrons are there. So for them it's just about retrofitting and just leveraging the megawatts that they have.

1:33:22

Is there anything that, I mean, we were reviewing this meta deal today and they have someone on staff who's the director of Global Energy. Is there anything that a team or one of these hyperscalers can do to move through the queue quicker? I mean, I imagine they can't just like overpay or bribe or it would otherwise just be an auction process. But is there, is it sort of a battle of the forum? Is that actually happening? Is it just a battle of the forms? Is it the regulatory team that you have on site? Is it different like scientists that are driving the work? Or is it just the best legal team wins? Like what is the anatomy of actually winning allocation in power?

1:33:45

The first thing we have to flag is there are many different transmission providers in the US so many different parties that work differently. But there's a few ways you can do so. First of all, if you're a sophisticated end user, what you're going to do is analyze the grid and try to understand where does it make sense to have power from a grid congestion perspective? Analyzing the network, you realize, okay, this area probably should have free power. So I'm going to talk to my utility and if you want to have a better chance of being ahead of others, then at least you have identified a specific spot. But it is likely to have available power. Now to be clear, relationships obviously matter. If you've been in the business for 10 years or whatever and you're good friends with C level executives, you can perhaps move up the grid just because of trust. It's not about necessarily bribing or something like that. It's more about you have a whole lot of, you have a whole lot of new entrants that don't have a lot of experience and then you have this other guy that maybe has been doing it for 10 years and has a lot of experience and success. So obviously for the utilities, easier to trust the guy that has the track record and has the existing relationship. So that's one angle. Sure, yeah.

1:34:29

I'm interested to know about your process to understand how much power each player is actually accumulating. I've seen the semi analysis watermark screenshots of of data centers. Is satellite imagery actually that useful for understanding like the map of total data center capacity? Are you looking at regulatory filings? Are you looking at statements from the companies themselves? Like it feels like you're at least one or two clicks ahead of what the executives of these companies are publicly saying. And the purpose of semi analysis is to provide analysis and data ahead of what's publicly available. But what is the process for actually understanding Azure's compute capacity this year before anyone else does.

1:35:36

Yeah, sure thing. So there's, there's different to some extent lead times for us the way we do the. Call it the short term forecast.

1:36:26

Yeah.

1:36:36

Short term being roughly the next six quarters.

1:36:36

Sure.

1:36:38

Extensive satellite imagery. We love satellites. We pay a lot of money for satellites. Every single blog these days you see satellite pictures and it's pretty simple. One thing is using historical imagery, figuring out when did construction of specific buildings start. Two is what's the typical time to build for an operator. So you have to get a sense of having reviewed many of their buildings. How long do they typically take or sort of understanding the design patterns, how maybe they can accelerate the buildup to what extent. Right. And then if you know how long it takes and when started, you can know when it's going to be operational. Then it's about the size. It's also about if it's a large project, how fast can you get it up to speed? Because one thing is to build the shell and the other one is to build all the electrical, the mechanical and obviously deliver sequentially data hall by data hall typically. So there's a bunch of things you can review. We use all of it. Satellites were big fans, of course, and like some of to us it's interesting because we've been able to be very successful at predicting trends with hyperscalers by analyzing data sets. Because you have to think of it, I'm saying you can just use satellite imagery, but you have to do it for hundreds of data centers. Who has the time to do that way?

1:36:39

We do.

1:37:52

We're crazy. Because if you think about it, Amazon, they accelerated growth, right, from 70% to 20% last quarter. And a lot of people were surprised because people were saying no, they're losing AI. Like Azure has accelerated, Google has accelerated and Amazon hasn't. But actually if you analyze data center construction, they've been accelerating like crazy on the construction side. In the third quarter, 2024, in the fourth quarter, 2024 takes a year for them to build. Obviously you see the acceleration flow mechanically. Q3, 25, Q4 25, that's one aspect. The other one, which is a bit more complicated is tracking the leases and so sort of understanding what are the different hyperscalers, leases, capacity from third party operators. The big guys are qts, you know, Digital Realty, Equinix and there's all the crypto miners that are basically doing the same thing. And that's also something we track very closely. And it's complicated because some of that Information isn't public, but by triangulating many data points, you can actually get to the answer using only public data, which is what we do extensively. And using permits, digging into filings and all of that is a huge part of the process as well. To get to the answer.

1:37:52

Yeah, as I understand, is the. Are you guys benefiting from like actual AI tooling on the research side? Like, are you guys running like, you know, 20 agents in parallel that are, you know, basically like looking, filing, trying to find the model, et cetera? Are you feeling an unlock?

1:38:54

Yeah. So we're using AI quite a bit. I think we can do more. I wish we could do more to some extent. Obviously it's bound by computer use. There are some complications, there's some portals for permit tracking purposes that we can automate. But the problem is that if you think about there are so many states, so many counties, so many different things, and some of them are fairly easy to use, and so today's agents can actually automate the process. Other ones are maybe more complicated, more archaic. Some of them you have a clean PDF. Other ones, it's like a scanner. It's written by hand. When you go into the weeds of the permitting process in the US So we do automate quite a bit. What we're starting to do, this is a pretty interesting project that we have on the data center side, is a vision model where you can actually, based on satellite imagery, detect real time, sort of what's the status of construction and understand sort of the inflections that you see as well on the planet.

1:39:16

Yeah, zooming out. It feels like, I think you quoted it something like half a gigawatt of, of new data center capacity was being added for a period of years, sort of linear growth. And now we're seeing a break in the graph and we're seeing exponential growth. And I'm wondering how you're thinking about. There's always this question of, like, when will AI show up in the GDP statistics? It's obviously a big business and there's a lot of revenue, but we're not seeing the 10% GDP growth just yet. Maybe, maybe that's coming.

1:40:14

When will we see productivity growth? I mean, productivity growth. CNBC yesterday was trying to figure out why US economic productivity surged almost 5% highest in six years.

1:40:48

Maybe, but I mean, my question with regard to, like, overall US power generation, are you expecting to see a meaningful acceleration, a break in the graph there this year, next year? Or is the data center overall energy picture still a small enough slice of the pie in terms of overall American energy production and consumption that we won't see it move the overall needle just yet.

1:40:59

Oh, it's already moving the needle. It's just going to accelerate. There's just one direction, at least for the next. Call it two years. One thing you can do is a lot of the utilities are publicly traded. So you can go one by one, every single one of them. The only thing they're talking about these days is the data centers is how much load growth I have and how. And then it's about capacity constraints, can I deliver on that, on that demand. But all of them are seeing tremendous load growth. And if you think about the leading indicators on the data center side. So for us we use two things on this. If you think about self built. So the data centers that are built and operated by hyperscalers, construction starts are exploding. So they're all massively accelerating. They have accelerated tremendously versus 20, 22, 23. So 24 was a big year. 25, insane. So this is a leading indicator with regards to what's going to happen in 26 for capex, for revenue and obviously for load growth. Because these days it's all correlated on the leasing side, same thing. If you think about how much data center commitments are the hyperscalers doing these days, it's also just up and to the right. 2025 has been an insane year for the leasing market. And so everything points to 26, 27 just accelerating. And then if you think about GDP growth, because I think it's an interesting topic, there's two things you can talk about. One is on the infrastructure side and the other one is on the productivity side. On the infrastructure side. You're already seeing it, but I think it's pretty easy to look at statistics. You can look at data center contribution to gdp, there's a bunch of stuff from some of the agencies. You can look at computer investments. There's a specific GDP that tells you all of that is exploding. Roughly speaking, a guesstimate is over 50% of GDP growth. Today's AI infrastructure, which is kind of insane if you think about it, right? And you can do some monkey math as well, right? You see Nvidia's revenue, what is it these days? Like 250 billion annualized or more. If you do the math right, a lot of that is going to the U.S. now there's obviously it's complicated because there's imports and things like that. You have to deduce. But you know US GDP is over 40 trillion billion, right? 1% is 300 billion easily year over year additions. AI infra investment is over 300 billion. Taking it off power plants, data centers, chips and all of that.

1:41:27

Yeah. And that should drive significant GDP growth. That's.

1:43:43

And then the second phase of this is on the productivity side. Obviously you have to see to some extent you would expect to see the people that are providing that productivity to be beneficiaries. And so this is why it's important to try to. What are the AI labs doing? What are the startups doing? Which is something we've been showing pretty closely recently at semi analysis. And so if you look at the AI labs, yes, they're all accelerating pretty fast as well. On the revenue side, like you saw OpenAI tripling revenue, anthropic 10x and so on and so forth.

1:43:46

Can you take me through the latest bottleneck in gas turbines? We had Blake Shoal from Boom on the show and it feels like he's expanding his business to build turbines. And I'm interested in that specifically because we hear about nuclear power plants coming online. That's obviously a very heavy regulatory burden. Also supersonic flight feels like incredible regulatory burden. But how difficult is it to just manufacture a new turbine or more of the same design, bring them online, actually ramp up capacity of natural gas turbines. Turbines.

1:44:14

So, so typically to develop a new turbine you're talking about 7 to 10 years R& D process. So it's pretty fascinating because you have a few, a few new entrants that are hitting the market today. A big one is Doosan doosan providing roughly 2 megawatts of turbines to Xai Do. Korean giants on the nuclear side, they've been developing their turbine for over 10 years and it's like, you know, perfect timing. Timing. They have it already.

1:44:52

Yeah.

1:45:18

What amazing timing. That's great.

1:45:18

It's pretty lucky. Pro Energy is another interesting example. Also 7 year R& D R and D program in late 23. Finally they got all the approvals. It's just very, very complex technology, a lot of very high precision materials. So ramping up the manufacturing side is complicated. Another thing to think about, and that's the difference we have with folks like Bloom is what is the. Like how long can you make your investment when you build a new factory for gas turbines or some of these systems. And this is an interesting thing to analyze because some, for some folks it's actually easier to build new capacity because their payback period is much higher and because their revenue per megawatt is higher and because their cost is also higher.

1:45:21

Just because the prices are going up because AI labs are willing to pay more for the turbines.

1:45:58

So. Yes, but I'm more thinking about comparing different technologies. Like if you think about Bloom Energy, the cost to buy fuel cells is very high.

1:46:05

High.

1:46:12

Sure is much higher than buying a turbine.

1:46:12

Sure.

1:46:14

But the flip side of this is that the payback period, because the revenue is so high per megawatt, the payback period of building a factory is actually much shorter. So for blue, from blue perspective, they take less financial risk if they expand capacity and they manufacture more.

1:46:14

Yeah, yeah.

1:46:26

Pretty good position.

1:46:27

That makes a lot of sense.

1:46:28

How did you react to the meta Oklo news from the this morning?

1:46:29

Well, I mean it's, I think all of the, all of the, all of these folks are, are looking for energy that is, you know, cheap, stable, are looking for ways to ensure their, the continuity of their supply and it just makes sense for all of the hyperscalers to, to, to work with the nuclear nuclear companies. It's not a surprise because everyone has done it already. Right. You saw Google as well with Kairos, you saw a bunch of these deals already switch with OKLO as well a about a year ago. So not too surprised about this one.

1:46:36

Yeah.

1:47:06

How much does the nuclear fission or even the fusion stuff, Google's partner with Commonwealth Fusion Systems on a few things. How much does that play into the analysis that you do when you're looking out six quarters. Because when we saw the date 2029, 2032, 2035, our eyes kind of roll back in the back of our head and say well that feels post singularity. So what's the point? Probably important at some point, but certainly less of less of a critical decision in the horse race of which lab will get power next month to train the next model that we're all going to be focused on.

1:47:06

Yeah, exactly. That's why if you read the report that we wrote is all about gas. Because if you think about the next few quarters, it's gas. There's just no other solution. Nuclear is going to take a few years. Solar and batteries are ready yet. All of these other alternatives, I think they all have good potential and I'm sure we're going to see a lot of different solutions in 5, 10 years. But today is just gas.

1:47:47

So you're modeling energy. It's mostly gas. Any plans to model or analyze or model water? Is that important at all?

1:48:07

Oh, interesting. Yeah. Big topic these days, right?

1:48:21

Yeah.

1:48:23

Overview generally speaking is that water is, is not that big of a problem because in the data center space you have this Trade off between energy and water. And so you can actually design closed loop systems that consume pretty much no water, in some cases zero water. There is some water required just to build the initial tank and the initial loop, but it's closed loop so you don't need water. So what's the issue?

1:48:25

Yeah, exactly. Yeah. It is a funny retort to anyone who's worried about the water, the AI water usage. Just if it's important, why doesn't semi analysis talk about it ever? Why are people not trading the water stocks? If it's, if it's important, I am interested in the.

1:48:50

Might talk about it soon.

1:49:07

Yeah, I mean I'm sure there's some sort of angle, but on the question of water usage, it does seem like Meta is moving from an air cooled system to a water cooled system. I think I have that right. They moved away from the H design of their data center. Can you tell me more about why the H shaped data center was not suitable for water cooling? It felt like a very modern building. Why was it impossible to retrofit that? Why did they have to go with an entirely new design?

1:49:08

Yeah, so the thing about the edge design above everything is it was really designed by Meta for, for leading cost efficiency. And so typically the ratio people use the PUE which tells you what's the energy efficiency of a given facility. Meta had the world's most efficient facilities. The energy required to cool the data center was extremely low. And that's because they had a fairly complex structure, free stories. The drawback of that is that the time to build a facility was about two years. So that doesn't work in the AI era. We're talking about month. It's 122 days for Excel that. So you need to go faster. So that's one of the main issues. Now the other one with regards to cooling. The, the way that we're cooling this specific data, sir, is like, it's. I call it an air to air system. You could, you could simplify and say they open the window. That's basically how they call the data, sir. They open the window. Obviously they have a bunch of.

1:49:39

We tried opening the window.

1:50:32

You know, I, I actually, I actually toured George Han's. He has a miniature data center, just a couple racks of GPUs that he trains for autonomous driving, for self driving cars that he builds. And his cooling really is just like a window unit that just flows air through this particular room in his office. And he's using air computer.

1:50:34

I had a car with no AC once open the Windows. Just open the window.

1:50:54

It's a time honored tradition. Sorry I cut you off.

1:50:58

It works. It doesn't work that well if your hardware is liquid cooled. Obviously if you have a cold plate that goes through the chip, then the question is how do you cool the fluid that you put into the cold plate to remove the heat from the chip? Yeah, opening the window doesn't work that well. You can do some kind of retrofit with liquid to our CDUs. It's expensive, it's not very efficient. So the best way to do is to have a dedicated fluid cooling system, which involves building a whole dedicated piping infrastructure and all of that, which is what Metadata. Their new data server can handle seamlessly liquid cooled chips, whereas the old one was much more complicated.

1:51:00

Do you have a view you said Meta's previous data center? I think it was 150 megawatts. Two years to build it. How fast are they now? Are they at six months a year? Do you have any idea of where they will be on the speed frontier since that seems so critical?

1:51:37

Yeah. So h. Two and a half years roughly speaking. Then they build some sort of rectangular design which is 12 to 15 months. And then they realized actually we need to go faster. And that's when you started seeing the tents and these tents. Yeah. The goal is to be able to put out a GPU cluster in six months. One of these tents. Wow. And it's interesting because they go back to an air to air cooling system. So what I told you earlier actually is wrong because despite hardware being increasingly liquid cooled, now they're doing air to air again, opening the window. But they have those sidecars which are liquid tour, which are expensive. Again, it's one way to go faster. That way you don't have to build a whole piping if you're in all that.

1:51:58

Yeah, yeah, that makes a lot of sense. What about how the other hyperscalers are matching up in terms of of speed, but also a lot of the other hyperscalers? A lot of the big tech companies have made commitments to maybe move away from natural gas to maybe go more net zero, more energy efficient, more carbon neutral. That feels like the water debate is maybe a moot point, but there will be some sort of climate discussion in the future as as more and more natural gas gets brought online. It is a fossil fuel after all. Or it is not a renewable energy source. So are there any big tech companies that are grappling with that or struggling to get through previous commitments that they made to be more environmental or more Net zero and now they sort of have to retool their business and messaging.

1:52:42

Yeah, you've already started seeing this in 2025, in 2024 where they're all saying, said for the time being we keep our commitment for whatever, 2030 or something, net zero. But short term we're not going to be able to meet our goals and to I guess clean our fleet as fast as we expected. But yeah, there's just no way. So there are a few interesting things. So the first thing I would say is you're seeing some projects that are natural gas based but they have ways to become more sustainable. For example, it could be a site where you have great geology to do carbon capture. So for Crusoe has a one like that in Wyoming. So that's. Those types of projects obviously have a long term potential as well because then you can meet your commitment if you're still committed to that. It's clear they're already that committed. But other topic, the other thing is so Google, from our analysis, Google seems to be still the most committed hyperscaler and they're doing some pretty interesting stuff in Texas. You saw the acquisition of Intersect Power. They're building some of these, some massive campuses in Texas where they actually have on site solar energy and battery. But to be clear, it's not behind the meter. The reason we don't talk about it in the article is because it's not fully off grid. There's still a grid connection, so it's not really comparable to the off grid deployments of folks with turbines.

1:53:43

Interesting. Yeah. On the Google topic, is there any sort of durable advantage in multi data center training that you're seeing from Google? Are you seeing evidence that they're leaning into that more? Are they building more smaller data centers or are they also, I don't know off the top of my head, are they competing with like the Colossus 2 these. You know Mark Zuckerberg comes out with the picture of the cube in Manhattan and it's very clear that Meta is in the one big massive data center race. At least they're trying to visualize it that way. I haven't seen that from Google. Is that intentional? Is there anything we can read into that about their actual training and deployment strategy?

1:55:08

No, no, you're right, that's a good point. It is interesting to analyze the different, I guess frontier AI training architectures from different players. If you look at what Meta is trying to build in Louisiana, 2.1 gigawatt campus for the first phase and they have individual buildings that are 400 megawatts each per building. It's pretty insane.

1:55:49

But it's still split up. It's still split up. But then putting it all on the same campus, there's some sort of economy of scale about power generation. Or is it just. Or is the latency with the fiber connections actually relevant? And you wouldn't want to have it across town, so you put it on the same campus. Like when would a big tech player choose to split across state lines, across the country, across the world versus centralize everything in one campus?

1:56:13

So I would say for now, Google is really the only one that has been that's had adopted this strategy and is very unique about it to some extent. You could argue it's because of, first of all, they're the most sophisticated on the networking side for a while. That's one of the advantages with Google is for the last 15 years or whatever, they've been the best at infrastructure on every single part of infrastructure and so networking. They've been building their own fiber networks for a while and so they have much higher bandwidth inside a Metro and between Metros and other hyperscalers because they've been doing that longer. They've been planning ahead for all of that. So Google is sort of ahead of the others. And on the tech side, they've figured out multi data center training from a model perspective, way ahead of everyone else. They've been very open about this. You had a podcast from Jeff Dean, for example, a few months ago where he said openly, yes, we're doing a multi data. So it works pretty well. No one else has really done it at that scale. For now we have a bunch of startups, but they're not doing it at scale. So yeah, just Google is better at doing it and what it enables them to do is it enables them to to some extent have more options with regards to site selection. They're not really limited to finding the 12 gigawatt site or the 1 1.5 gig sites. They can just go on a Metro, maybe a 50 mile or even 100 mile area and find a bunch of sites that are each 200, 300, 400, 500, connect them and bam, they have a 2 gigawatt campus. So that's the strategy they're pursuing.

1:56:41

That makes sense. Last question for me and we'll let you get back to your day. I know it's late there have AI workloads have this. A shift in AI workloads had any effect on power decision making? It feels like you go back a couple of years ago and the vast majority of power that was being used by AI was for training. Now we're moving more towards inference. Does that change the landscape of power acquisition for AI companies broadly or is it sort of an irrelevant point?

1:58:11

So first of all, I was actually disagreeing with the premise based on our analysis. Training is still the majority and it's growing equally as fast as inference. Inference is surging, but training is surging as well. And that's normal because you have an incentive to do it. Yeah, everyone wants to, you know, invest in the model that is going to unlock revenue growth next year.

1:58:42

Sure, sure.

1:59:00

We haven't seen the limit. So there's. The incentives are aligned to invest today and that's what everyone is doing. Interesting. Anyways, to answer your question, yes, inference does have different requirements. There is different types of inferences. Where? Well, if you think about OpenAI for example, they have two main businesses. ChatGPT is a vertical application. It's fully controlled by themselves. So for them they can plan larger campuses, can be gigawatt scale, can be a few hundred megawatts. It's going to be big campus anyways. They can make use of smaller ones. But if you think about it from an infrastructure perspective, what is their biggest pain point as a company is they want to maximize GPU utilization rate. This is the single largest expense by far far is gpu. So they need to maximize it and that's easier to do when you have large campuses. So that's why you still see campuses are fairly large even for inference. They also in terms of latency, that's a common sort of topic of discussion. Do you need to be very close to the end users? Well, if you think about it, what is actually consuming power for OpenAI it's, you know, deep research, it's GPT 5.2 Pro. It's these thinking models and they take many and so you don't really need to be near the metro, you can just be sort of far away and still find large buckets of power.

1:59:01

Yeah, makes sense. Well, thank you so much for coming on the show. Congratulations on the progress. Really appreciate the episode 78 analysis is hiring correct. Can you take us on a brief summary of roles or how to apply?

2:00:12

Oh yeah, we're hiring a lot of people so we have a careers section in our website so you can all check out. We're looking for folks, folks in the AI space. So if you're interested in digging into what we call tokenomics, which is analyzing the economics of AI, analyzing the latest trends in terms of LLMs, different types of model architectures. Reach out. If you're an engineer, you have experience with GPU clusters, reach out as well. We're hiring to increase our technical team Inference Max. Really cool project as well where we benchmark all of the different models. If you want to work on Inference Max and work on TPUs and Trainium and GPUs and AMD and Nvidia and all of that, reach out as well. We have a lot of pretty cool offers, so you should all check out the website and yeah, it's a good adventure.

2:00:24

Amazing. Well, thank you so much for taking the time to come for us. Great to see you and happy New Year. We'll talk to you soon.

2:01:06

Cheers.

2:01:12

Goodbye. Turbo Puffer, Serverless vector and full text search. Built from first principles on object storage. Fast 10x cheaper and extremely scalable. Next, we are entering our Lambda Lightning round with Andreessen Horowitz celebrating their 15 billion dollar fundraise. We have Jen Ka in the restream waiting room. Let's bring her in for the TVPN ultradome. How are you doing, Jen? Good to see you, Jen.

2:01:12

What's happening?

2:01:39

Hey, hey.

2:01:39

What's happening, brothers?

2:01:40

Not too much. You're off to a banger start of the year. Congratulations. Break it down for us. What happened?

2:01:42

We knew we'd need a bigger gong for you guys in particular this year.

2:01:48

And distribute the gong hits across the the four Andreessen Horowitz partners that we're talking with.

2:01:52

We'll see you know what I have to say.

2:01:58

There we go.

2:02:02

There you go. Wait, what does it say? What does it say?

2:02:03

On says it's time to build, baby.

2:02:06

I love it. I love it. That's fantastic.

2:02:09

There we go. You've probably been hitting that a lot. Do you. Do you. Do you just do like one hit every time you get off an lp, you get a new commit. Just a light tap. You probably had to do a few of them to get up to 15.

2:02:10

Exactly. It's not quite your 80 inch one, but next time I'll get a horse and we'll call it even.

2:02:22

There we go. How are the LP pitches going into this fundraise? Different than in previous years. The market's evolved, technology has evolved. Andreessen Horowitz has evolved. What were you saying that you felt like it was the first time you were saying to LPs this time around?

2:02:26

Yeah, so. So let me first break down. So $15 billion, huge headline. The number is huge. But you know, I first I should foremost say I normally don't respond to online rumors, but I feel need to do so at this moment. The billion dollars is not for the Nepal or Himalayan or Greenland.

2:02:47

Right.

2:03:04

Let's spell that right now.

2:03:04

People tease things all the time. Little, little, little breadcrumbs in the releases. I'd like to see an American Everest. I feel like I've heard for Greenland.

2:03:07

The moon.

2:03:15

The moon might be American at some point. It already is in many ways. Maybe Everest. But you know, you're putting it in startups, you're putting in technology.

2:03:16

Exactly. So hope springs eternal. So the prelim number, you guys covered this yesterday, but the prelim numbers for NVCA came out to 66 billion. So that actually ended up being what would have been equivalent to 22% of what was raised in 2025 with the $15 billion.

2:03:23

Wait, so does the 15 billion get included? Are they going to update the 2025 numbers or do they count it towards 2026? Because it's actually being announced now.

2:03:40

Yeah, it's going to get counted for 2026. So that's forward. So we're closing our fund today. Okay, so that would be 2026 numbers. Yep.

2:03:51

Cool.

2:03:59

And so calling now VC winter maybe potentially over for some folks, but fantastic. But the sentiment, to answer your question, so the sentiment from LPs is different insofar as that we are now, this is the next, the second set of funds that we've raised in this super cycle. And so we were oversubscribed in three months. You know, it was, I think, very clear for most people that AI is obviously taking over the world. And particularly when LPs have conviction and also the right information, they will close quickly. So most funds, you know, the average VC fund takes probably close to over a year to fundraise. And the reality is it's, it's a tale of two cities. If you have great companies, great performance, great dpi. Yeah, it's very, very easy to raise capital. We are luckily in that camp. If you don't, it's just a lot tougher. And by the way, let me tell you also a story because it's related to, on this liquidity topic that you guys, guys oftentimes talk about and hear about from LPs. So, you know, there's a lot of bellyaching from LPs about liquidity, but the reality is it's in select companies. So we actually went through this whole exercise last year. So this was in the middle of liquidity concerns. And this is early days of, you know, the administration stance on on endowment tax cuts. And, you know, we internally had this conversation, said, gosh, you know, should we offer some liquidity, particularly for some of our older funds, funds to our lp? So, so we went around, we called all of our LPs in those older vintages, and specifically we had a strike position, seed position, and fund one. And then we had data bricks at the series eight and fund three. And we said, hey, you know, we know you're in a liquidity crisis. Would you be interested if we got you some liquidity in those names? And I'll tell you, 30 out of 30 of those early LPs that absolutely not.

2:04:01

We want liquidity. You're telling me what, I like to not ride my winners. Interesting.

2:05:51

Exactly.

2:05:55

Exactly.

2:05:56

So there's like, subtlety in that conversation, which is like, you know, they want liquidity, but they want liquidity. Not out of those names. They want to ride those winners. Right. They want to leverage.

2:05:56

Yeah. It's unfortunate you want liquidity the most from the assets or the companies that.

2:06:05

You'Re least excited about, which is this paradox. Interesting, interesting. Yeah. Yeah, that makes sense. Talk about the split of strategies going on at Andreessen Horowitz today. I think a lot of people were curious.

2:06:10

You guys have the dough you guys.

2:06:24

Have being included in this suite of funds. Is that just a different cycle? Is that a mechanics thing, what's going on there? Or is it truly like a completely separate thing? And we'll be hearing about that later.

2:06:25

Yeah, so the latter. So the funds that we raised and announced today, it's five out of seven of our funds. So crypto is off cycle, and then so is our games fund. Okay, so more to come from that.

2:06:34

Got it. Okay, that makes sense.

2:06:44

Yeah. How are LPs up to speed on this kind of structure? Now, you obviously don't have to go into a conversation, explain Andreessen Horowitz. But most funds are not at the scale where there's this multi fund approach. How much of the conversation is about, okay, I'm giving you capital, where is it going to be actually allocated and how is it going to be split across the funds? Is it just straight pro rata across the different strategies and funds, or how does that work?

2:06:46

Interesting.

2:07:16

Yeah, yeah, so that's a great question. And in fact, I think we're one of maybe the first who actually split our funds. Most firms just have a generalist fund. Everything's out of one vehicle. And we very early on presciently realized we needed to decentralize as a firm. And then also our Funds as well to match that to the teams. And so if you look at any of the individual teams, the deal teams are no bigger than, you know, four to six people. And so we're kind of similar to smaller funds and smaller firms, but with the breadth of course, of the injuries and horrors umbrella. And so for LPs, when they think about allocating to us, most of them just say, hey, I want to follow you into all the different funds. And they allocate pro rata. And in fact we actually set up a vehicle to allow them to do that. Some folks pick and choose and our view is, hey, every single fund needs to stand on its own and it needs to earn zone keep from its LPs. And sometimes those LPs might be different. Right. Some LPs for example, internationally, can invest into certain strategies, like crypto, for example, maybe American dynamism. So there's some nuance there of which we do account for.

2:07:16

Interesting. Yeah. How many questions are you getting from LPs about trying to predict the next Andreessen strategy that might take place in this fund? Like Andreessen started as. I mean, you look back at the early, like the fund won and it's basically a seed fund, but by today's standards, growth was obviously added on, then bio, then crypto, and then liquids, tokens. There's so many different strategies that if you went back to the dawn of Andreessen, you would say, well, that doesn't fit in this fund mandate. And we've seen firms buy hospital networks and do more private equity style deals, do more secondary deals. Are LPs looking for you to lock a strategy or are they leaving you with a lot of flavors, flexibility? Are they looking for guidance on what might happen in the next 10 years in terms of creative financial plays that you might be able to make?

2:08:17

Yeah. So it's funny, our first fund is. Funny you should say it's the size of a seed fund. The first deal we actually did was the buyout of Skype, which is, there's a, there's a good story around that we made for X our return in 18 months and the rest is history. But, but you know that, that first.

2:09:10

Yeah, I always remember seeing that on Andreessen's website. And it was like it wasn't a seed bed. It was this weird deal, but it still panned out really well. And it was like a great logo to have on the, on the portfolio page, but for peculiar reasons.

2:09:27

Yeah. And there was a bunch of risks in it because everyone was like, oh, you won't get the IP because of eBay and blah, blah. And then, you know, Microsoft ended up buying it. But, you know, and then there's a whole story there because a lot of our LPs have, suffice to say, after raising a venture fund, we're like, what are you doing? Yep, it ended up working out. Nonetheless, you know, we always talk about internally how, you know, the way the individual funds are set up now is almost in the incarnation of the original Andreessen Horowitz, from the size of the team, from the capabilities and resources on the operating front. And so you've got these, you know, kind of seven different funds and teams that are effectively the incarnation of that first Andreessen Horizon, now replicating. And that's actually how you scale. We have 600 plus people at the firm now. That's the only way you could actually nimbly move without getting mired in the morass of bureaucracy and oftentimes what big organizations end up being. And so in some respects we don't necessarily, we're not motivated by innovating into, you know, there's a lot of VC firms out trying their hand at private equity, as you said, buying hospitals, raising private credit, strategies like that. That's not really of interest to us. I think you mentioned at one point, like going public, like, as long as Ben is CEO, we are not a public company, so we don't try to innovate on fund structure. Right. We like boring vanilla VC returns and you give us money and we'll send you back. And where we'll take risk and innovation is on the companies we invest into.

2:09:40

How much do you involve individual gps in the fundraising process? Sounds like, I mean, a three month process really not that long. Are you tapping them in at key moments? Because certain LPs want to understand like, like really get to know the individual investors. Or are you like aggressively trying to protect their time? Because their time is really best spent with founders and actually evaluating and doing deals.

2:11:06

Yeah, so yes, we try to protect their time, but also at the same time, like, this is just like a company, like a fundraiser is a very important exercise. And in fact, you know, a few years ago when people were asking whether we would go away from the traditional fund structure, you know, Mark and Ben kind of like the concept of pressure testing our thesis every couple of years, right. You got to go out to your LPs, you got to prove to them that your thesis is worthy of their capital and then it pressures you also. Sorry for my voice. It's a little Hoarse because I've been shouting too often.

2:11:30

Like, why are the wire deadlines now?

2:12:04

Exactly, exactly. But that kind of sentiment of pressure testing your thinking is incredibly. And you guys know from fundraising with companies, like, you learn a lot through the course of that process as well. So all of our gps get involved. All of them are in the meetings, they're talking to the LPs, they're shaking the hands, kissing the babies. They are front and center of it.

2:12:07

How do you realign the LPs from just endless AI questions and. And actually keep them interested in bio, healthcare, American dynamism? Do you try and have the AI narrative cut across everything, or is it drowning out the rest of the stories that you're trying to tell?

2:12:30

I think AI is similar to any platform shift where it's just going to infiltrate everything and it's obviously in the zeitgeist, but eventually it's just going to go in the background, just like cloud or.

2:12:47

Yeah, there was that quote from Mark in the launch video that was like, Someday in 10 years, we won't talk about the Internet because it'll just be everywhere and it feels like we're already. I like when companies. I think this is the year where companies, like, maybe stop pitching AI as aggressively in taglines because you should just assume that a company is leveraging it to the fullest extent.

2:12:55

Yeah.

2:13:18

100%. Yeah. What else are you doing now? 100%. But I do think it's great too, because I also think, you know, with this platform shift, LPs can actually have a feel for how transformative this is themselves. So our entire, for example, fundraise process, we tried to take a native first approach. So we had an AI chat bot that was replicated. I was about to ask in my. In our data room, there was an AI gen in there. There's an AI chatbot that was answering any question at any hour of the day.

2:13:19

Yeah.

2:13:51

And then our LPs also are playing with these tools themselves as well in their underwriting and their diligence. But also even individually, I won't name the LP, but I was talking to an LP earlier this morning who was playing around with repl.it and I was like, you got to try it. Just code something that you wish you had access to. And she's like, well, I really want to be able to code an app that can look at peloton classes and let me know when this instructor sees. I'm like, try to prompt it. And she was able to do it literally in the course of the Morning. So it's one of those things where I think these worlds are converging so quickly. It's also. So it's great that we're doing it's testing and trying and iterating and for the first time LPs as users can actually see the real world visceral impact to how they run their day to day life.

2:13:51

What is general LP sentiment specifically around 2026? What are expectations? Obviously we're expecting a slate of IPOs and that's very exciting if you've been in, in these, in these companies, these names for, you know, a decade or more at this point.

2:14:36

Yeah, I feel like Elon dropped like an early Christmas present when he was, it was like rumored to say that Space X might be going public in 2026. And it was like, oh, maybe we'll go public in 2026. So I do think sentiment and people are generally positive. Obviously you know, we'll see where the IPO markets kind of turn out to be. But generally speaking it's, it's early, it's off to a good start, we'll see what happens. But I do think people are expecting more capital this year in a way that once one breaks through, it's going to be a watershed moment that might even top 2021 in terms of the amount of liquidity coming back to folks.

2:14:55

That's going to be exciting. That's going to be exciting.

2:15:29

Was 15 always the target or did you go out thinking that you do less and then you kind of upsized it based on demand?

2:15:31

We had a range for our target and we try to kind of keep in that range just to avoid.

2:15:41

But you knew it was possible because if maso can do 100, it's like, come on, can we do 15?

2:15:47

I won't make that comparison. It is funny in retrospect, that vision.

2:15:57

Fund, the whole thesis was AI.

2:16:02

Well, yeah, I completely blocked Vision Fund out of my head. And then I was like, I was researching before this, I was like, okay, 15 billion has to be the biggest venture fund ever. And it was like, oh no, of course, of course.

2:16:05

Anyway, well thank you so much for coming on the show. Congratulations on massive news and we will talk to you soon.

2:16:18

Jen, incredible work.

2:16:24

Awesome.

2:16:25

Great to see you guys.

2:16:25

Catch you later.

2:16:26

Cheers.

2:16:27

See ya.

2:16:27

Let me tell you about Vanta Automate Compliance and Security. Vanta is the leading AI trust management platform.

2:16:28

And thank you to the chat. There has been some updates to Claude's policies. Tyler is researching the story now and we will get to it as soon as we can.

2:16:35

Well, we have Alex Rampel from Andreessen Horowitz coming in to be again. Ultradome. Welcome to the show. Alex, how you doing?

2:16:45

Thank you.

2:16:51

Good.

2:16:51

How are you?

2:16:51

Beautiful background, fantastic American flag. What a day. Congratulations. How are you doing?

2:16:52

Thank you. Good, good.

2:16:57

Fantastic.

2:16:59

Super excited to have you on. I've enjoyed you're somebody who's. I've read your writing and listened to your podcast appearances for a decade now and always, always appreciated your point of view on a bunch of different things. So welcome to the show.

2:17:00

Yeah, thank you. I'm here to prove that I'm real.

2:17:18

Fantastic.

2:17:20

It's like proof of life is very important increasingly, right?

2:17:22

Yeah, it is.

2:17:24

I mean, first time on the show. Joe, can you give us a little bit of the backstory, the journey to Andreessen and how long you've been there?

2:17:26

Sure. So I've been here for 10 years. Previously been a long time success. Previously. Longtime entrepreneur. So kind of started by writing software when I was a kid in high school, actually even before that. And then out of college. I graduated in 2003. I was probably the only person from my class that just kind of became an entrepreneur right away. And it wasn't because I was smart or dumb. Probably more dumb than smart.

2:17:33

Start.

2:17:58

It's like I had a little business that I was running in college, so kind of kept doing that. Met this guy named Chris Dixon who was at Harvard Business School when I was at Harvard College. And you have to remember like 2002 when we met, the Internet 1.0 had just died. Everybody lost their jobs. September 11th it happened. And what do you do if you're a dried up entrepreneur? You go to business school. I remember there's a company called Cosmo.com, a huge hit that kind of went to zero. What did that guy do? He went to HBS.

2:17:58

No way.

2:18:26

So Chris Dixon was there and he and I were like the only two people I swear in like the entire state of Massachusetts that thought that the Internet was still kind of cool. We got introduced by a mutual friend. Had had coffee at Aubonpan, this like little east coast chain, and then cooked up a product called Did They Read it? Which is still around today. It's an email tracking tool. It embeds a tracking pixel in every email that you send out. That did pretty well. Then we started the venture backed company together that became Site Advisor or that company got acquired. And then I started another company called Trialpay to like the thing that we learned at Site Advisory is that nobody likes Paying for software like you're willing to pay for an intangible good, like a glass of wine. $20 for that seems totally reasonable, right? But paying $20 for one song on iTunes, there would be riots in the streets. So the idea was, I'll give you this digital good for free if you buy something else. And if you know how affiliate marketing works, it kind of plugged into that. So it's like, hey, we'll give you this product for free if you sign up up for Netflix or if you switch to Geico, or if you shop at the Gap, or if you get a Discover card using the affiliate commission to go pay for the product. That did pretty well. Like it was like half of the revenue of Site Advisor. It was like from my, from a little shareware business that I used to have back in the day. It doubled our revenue. So I turned that into a company called TrialPay. That did great for a while. Then it did terribly that I kind of resumed to, okay, sold it to Visa. And then along the way met this guy, Max Levchin after he had sold Slide to Google. And we cooked up a company called Affirmative Firm. So I co founded a firm with Max in 2012 and actually brought a firm to Andreessen Horowitz as a funding opportunity, which they did. And Chris Dixon kind of ended up talking me into joining here in 2015. So I've been here ever since.

2:18:26

How quickly did you realize that Chris and Max were special? Because I imagine during those two periods you were meeting hundreds of different people. I'm sure people wanted to build stuff with you, other entrepreneurs and you picked well back to back. And that's probably the hardest. Hardest, you know, one of the hardest things to actually clock at times.

2:20:11

Yeah, I mean, I think a lot of the greatest people, they have two things in common. They have this term that's going around a lot like high agency. Like they don't just like follow the rules, they just like take matters into their own hands and do something and then they just kind of know the history of everything before. Like they're just like students of history, philosophy. Like, you know, Chris was a philosophy major. People don't know about him. He went to grad, like he went to, you know, he got his bachelor's degree in philosophy, was going to do a PhD, kind of realized that was a bad idea and then eventually went to business school which was, he will probably say, the worst idea. But it was kind of self evident. I mean, Chris and I, the history.

2:20:34

Thing'S a real thing. Like if you're talking to an entrepreneur that has been building their business for one to two years. And you can tell them companies that in somewhat recent history, in the last decade even that have attempted that, or companies that are adjacent and they're like, oh, I'm not familiar with that. It's like immediately such a bear signal.

2:21:10

It's a red flag. Well, the opposite. That's the red flag. Whatever. The opposite.

2:21:27

That's what I'm saying. But yeah, the green flag is never a green flag.

2:21:32

What's the green flag?

2:21:37

The green flag is because not only do I know everything, I mean, I'll give you one example. I think the Collison brothers went out to Dee Hawk's ranch. Like Dee Hawk started Visa. He's kind of like a weird quasi communist, even though he started one of the biggest companies in the world. Because Visa was meant to be this, like it was a non profit. Visa was a nonprofit until 20, I don't know, 2008, maybe 2009. It was the biggest IPOs.

2:21:39

The biggest IPOs. Yeah, that's right.

2:22:01

Yeah. But it was a nonprofit until then. A nonprofit, like the NFL is a nonprofit. It like looks like it makes a lot of money, but it's owned by the constituents. And the constituents that owned Visa were the banks. And it's like, okay, I'm starting a payment company. There were a lot of payment companies that came before. But it's like, who will. Let's find this guy who's 90 years old, who's moved outside of capitalism, is working as a farmer just to learn from him. And like I have this mental model that I now use for entrepreneurs. And it's a memo that I've written that we use internally a lot. I gotta say, like the best entrepreneurs, they have five things that I look for. They can materialize. Labor, capital and customers. And hopefully those are self evidence. Like you can get people to quit their high paying job for like certain failure. It's like the Ernest Shackleton thing. It's like you wanted men for dangerous journey, almost certain failure and death. But if it worked, you might be famous, right? It's like you want that very, very hard to do. You have to find people that can materialize capital. It's like get people to give you money. And the best, the best sign of future fundraising success, like if we do round N, we want to make sure there's going to be a round N plus one or you're going to be profitable on round N, which is unlikely. So are you good at fundraising? Can you get customers. Like, imagine it's like, I have two weeks of cash left. Please be my first customer. I have none. It's very hard to pull that off. Then you want to know the history of the space, which is super important to your point. You want that green flag version, not just the intermediate. What's the combination of green and red flag? You don't want the turquoise or the whatever, you want the green flag. This person knows everything that's tried before and they have a new angle of attack they're not going to learn on the job. They've actually learned through history. And then the last thing that I care about a lot, everybody on my team knows this. My favorite book is the Count of Monte Cristo by Alexandre Dumas. And it documents the story of this guy, Edmund Dantes, who's like wrongfully accused, is like in prison for 17 years, but then becomes the richest person in the world, but doesn't give a shit, right? It's like all the riches in the world do not matter. He wants revenge. So you could either. Revenge kind of sounds bad.

2:22:03

But you.

2:23:58

Either want revenge or redemption. And some of the best entrepreneurs have this in common. And the reason why it's so important from a venture lens is imagine that you're a 20 year old kid, you start a company and somebody offers you, I don't know, half a billion dollars to buy your company and you own 25% of it. You're going to make over $100 million. You'd have to be insane to turn that down, right? And we need people that are insane, that actually are going for. It's not that we don't want people that aren't capitalists, that don't care about money, money, but it's like they care about. If you've seen the movie Spaceballs, it's like, we're not doing this for the money, we're doing it for a shitload of money. A little different here. It's. I'm doing it for another reason. And like a great example of this is there's this guy, Reneau Laplanche, who started a company called Lending Club. Very famous company at the time. There was like a dearth of IPOs. Like Lending Club kind of gets to scale, goes public, he gets fired from his board, ousted from the company. He's probably made hundreds of millions of dollars. He's the Count of Monte Cristo. He's like, you know, fuck those guys. I'm going to start a new company. I'm going to start an upgrade. And you know, what he called his new company Upgrade. It does the exact same thing as Lending club. It's probably 10 times the size of Lending Club now. And what's motivating him is not just the hopefully, you know, shit ton of money Spaceballs quote, but he wants revenge, he wants redemption. And you see this with some of the best entrepreneurs. Like, what is the driving motivation? Because when times get tough, you need something because there is no money if your company's going to zero. If you're Ernest Shackleton in the winter of Antarctica, your voyage is not successful. You need something else driving you at that point. And that's why that metal is something that I find extremely valuable.

2:23:58

Have you seen Spaceballs, Jordy?

2:25:37

I have not.

2:25:38

You got to.

2:25:39

I famously have seen no movies. He's not a scientist.

2:25:41

Last question.

2:25:45

One of my probably like a really. I was going to say maybe it's hard to exactly place it top 10. Like, I loved your episode on Invest, like the best on operating systems. How is AI kind of like updating your thinking on moats and operating systems and how somebody can create a lot of value with a startup?

2:25:46

Yeah, well, I think. Well, maybe I can rewind a little bit. And. And just because we announced this new fundraise, I can tell you exactly what we told our LPs in terms of like, what we want to invest in at the application layer. Because I do application layer stuff and it's really three things. You know, category one is I call it Greenfield Bingo. And kind of maybe another way of answering your question is there's something that there's a quote that I use a lot. The best companies have hostages, not customers.

2:26:08

Right?

2:26:35

Like, that's why nobody likes using Salesforce.

2:26:35

You gotta be taking revenge, you gotta be taking hostages.

2:26:37

You know, it's very Old Testament stuff. The best companies have hostages, not customers. Those are great companies to invest in. Right? And that kind of goes to the point that I was making like, you know, netsuite, Workday, Salesforce, like they're all hated by their customers, but none of those customers can leave. However, if you build a better version, like kind of a more AI first version of all of these companies and you're selling into the green field, you've got a shot. Right? Because I was lucky enough to be the first investor in Mercury. And Mercury worked not by stealing people from svb, they just worked. It's like, oh, you're a brand new company, you can use shitty SVB or you can use really good Mercury. And that worked. Whereas they never got customers from svb. Until the weekend that SVB failed. So that greenfield opportunity for software that is AI enabled in the same way that that was true for cloud, that was true for mobile, it's like, here is the new thing, the incumbent will eventually build it. Another expression that I use a lot is the battle between every startup and incumbent is whether the startup gets the distribution before the incumbent gets the innovation. My default assumption is that the incumbent normally wins because they have the distribution. They will get their act together three years later, and with AI and cursor and everything else, they'll get their act together maybe three weeks later. So the kind of of the might of distribution is very, very powerful. So one option is you just go into the greenfield. So that's kind of category one is we call it greenfield bingo. It's just like build. We have a bet that's just like NetSuite. It's better, it's AI enabled. But they're not going to steal customers from NetSuite, they're just going to get greenfield. Category two is this kind of new, super exciting category of software does labor. And there is no incumbent software product for trial attorneys. Like it's called Microsoft Office, but like Eve does that and does that really well. There's no incumbent software product for dental office receptionists, but Tenor does that and does that very well. So that's a really. These are all industries that I wouldn't say they've been untouched by software, they've been untouched by specialty software. And the reason why is because the market was perceived to be too small. And this is exactly what happened to SaaS. Like, like fintech really changed SaaS significantly. Because take. I'm sure you've heard of Toast. Toast is one of my favorite businesses. It's like Square, but it's only for restaurants. It's this whole operating system for restaurants. How many restaurants on their IBM PC Junior in 1984 use software like Zero. And how many of them would pay tens of thousands of dollars a year for software? Zero. But they all need payment processing, they all need payroll, they need these other services. You kind of bundle them in with software. And this is the really exciting thing about AI is you go say, hey, trial attorney, I want you to pay $50,000 a year for software. You said this 10 years ago, like, no way. Like, we'll pay for Microsoft Word because we use it to write demand letters like, that's it. And now you can say, hey, we'll handle all these cases for you that you could not handle profitably. And that's AI plus software. Now they are software buyers. So that's category two and then category three, I call the walled garden. And I wrote a post about this a little while ago. But walled garden businesses are amazing because if you assume that in the world that we live in today, and this is another way of thinking about kind of defensibility and AI OpenAI has their sights kind of on everything, right? Like anthropic probably has their sights on everything. It's so easy to build everything. So I don't know. Have you guys heard of open evidence?

2:26:41

Yeah.

2:30:09

Yeah.

2:30:09

Okay, so I tore my Achilles in February. So almost a year ago. Sucks, right? It's all better now, doing something fun. It was, I was skiing in Japan.

2:30:10

There we go. That's a good reason.

2:30:20

What do I do? So I go to ChatGPT. I'm in like the clinic in Niseko, Japan, talking to this Scottish doctor and he tells me, oh yeah, you only have surgery in the U.S. nobody does it outside the U.S. i was like, this guy's on crack. Of course you have surgery to fix an Achilles. I go to ChatGPT, it tells me everything. Then I find this thing open evidence and it's like ChatGPT, but it has every single medical document in the world. And imagine that tomorrow ChatGPT5.3 comes out. It's AGI, everybody agrees it's AGI, human race is over, but it has no medical data. And then on the other side you have GPT 3.5 and it has every single piece of medical knowledge ever known to mankind. What would you rather use? And the answer, at least for me, and I did use this, is open evidence. There are so many businesses that look like this where they find some proprietary piece of data. They're the only ones that have. Because before you would have to sell data that was like your only. That was your only hope as a business. And like another example that I mentioned this post, there's a company called Velix. And Velix is this like 25 year old European data business that bought up, you know, legal records in Spain to start and they would sell it to firms like Wilson Sonsini that needed it for case law. Now they sell an outcome because they're the only ones that have all the records. So you can chart, you can build a really interesting business if you're the only source of some unique form of data. And I love businesses like that because that's the other. Sorry for being so long winded here on the answer to your question, but the businesses that can be Very, very large. In AI that can grow very, very quickly, you still need to make sure that they're fundamentally defensible. And that's the really hard thing to disambiguate today, which is you can have things grow so quickly, but they can also go to zero so quickly because. Because anybody can build software in like a weekend, which is both great and terrifying at the same time.

2:30:25

Yeah, indeed. Well, thank you so much for hopping on the show and breaking it down.

2:32:12

Incredible overview.

2:32:16

Yeah, thanks guys.

2:32:17

Yeah, great to meet you. Thanks for being here.

2:32:19

We'll talk to you soon. Alex, congrats. Have a great rest of your day. Let me tell you about console.com. console builds AI agents that automate 70% of it. HR and finance support, giving employees instant resolution for access requests and pass password presets. And up next we have David George. We are running late. David, good to see you again. How are you doing? Congratulations.

2:32:20

Great to see you.

2:32:42

All the progress, how does this change growth, the growth practice, Is this just more of the same strategy or can you do new things with this new fund?

2:32:42

It's more the same, honestly. First of all, I'm still thrown off because I've worked with Rampel for seven years and we're close friends and, and I just realized that his favorite book is Count of Monte Cristo and that's my favorite book.

2:32:52

No.

2:33:03

1.

2:33:04

Yeah, yeah, it's a great one.

2:33:05

Yeah.

2:33:08

No more of the same for the growth fund. I mean, look, the last seven years since we started it, it's been a pretty simple mandate. It's like best companies in the world, best founders in the world with huge ambitions. Typically, the bet that we're making is that it can be bigger than other people believe it can be. Paki wrote a whole loan, 16,000 word essay about this today. But if I were to summarize our views and what defines like, you know, quintessential growth investment for us? It's, it's a belief that it can be bigger than people would realize. And we got a ton of success stories on that. So more of the same in terms of our strategy. And it just so happens that we're on the back of like the best trend of my lifetime.

2:33:08

Do you think there's something changing in the psychology of startup founders where the new crop of mega corns, the SpaceX, is. It sort of gives the next generation permission to stay private longer maybe, hey, you know, previously it was maybe $100 billion IPO would be crazy big. Now we're maybe having a few trillion dollar IPOs happening soon all of a sudden, logically, you know, even if you're not in that category, you're like, I'll stay private until 100 billion. Is the psychology changing for founders?

2:33:49

Yeah, it is changing. I'm going to talk about the psychology of founders in two different ways. So one that I'm more excited about. Like we can talk about capital markets and public markets and private markets. The thing I'm more excited about is the psychology of the founders that has changed post Covid. Like this generation of founders is just way more hardcore. Like I for one am all for, you know, the being back in the office, the working really hard, you know, unabashed pursuit of success. And I think it's a big change. Like in the last five or six years that's happened. I think it's part of what's propelling these companies to be so good so fast. Obviously, AI is the big technological driver, but I think the vibe shift is a huge part of it. You know, we were looking last night at some stats and going back and forth on some of our best companies, and we have a bunch of the AI native companies that are application companies that, that are kind of 100 million plus 100 million to a billion dollars of revenue.

2:34:22

Yeah.

2:35:14

And we were looking at do we think they're running themselves differently? So to this point of like, you know, different vibes, what they, what they, what they care about. And it turns out like the old rule of thumb for companies was, you know, if you look at like revenue divided by all their employees, it's like $400,000 of revenue per employee. Like you look at like public software companies. That's kind of where it shakes out. All these AI native companies, they were basically like 500,000, 900,000, 2 million, 2.

2:35:15

Million, 2 million, 5 million.

2:35:44

They're totally different. I think they're being run totally differently. We're really excited about that. And the biggest thing is just there's a tsunami of demand coming their way, which I think enables them to run much, much faster. So I'm super excited about that. As it relates to the capital markets thing, the staying private longer, it's totally rational. Look, there's, there's a bunch of it that is our own sort of government doing as it relates to being a public company and how difficult and expensive that has become. We have a robust private market. It's been a benefit to us. It's allowed us to invest in a bunch of companies that otherwise would have gone public sooner when we can still invest in them. And they're growing really fast. The value proposition for founders is pretty good to be a private company. You know, they can, they can stay private. It's probably a little bit of a higher cost of capital for them. But the trade off for them is, you know, they can sort of avoid the daily kind of volatility of the stock price and what that means for.

2:35:48

More M and A options as well. There's a lot.

2:36:50

What is, what is your day like? What are your weeks? What are your weeks like? Because I imagine like the balancing and like prioritizing what you're doing.

2:36:52

Most of it's adding zeros to spreadsheets.

2:37:01

Yeah, yeah, yeah.

2:37:03

Okay.

2:37:04

We gotta add.

2:37:04

The model's breaking. The model's breaking. No, but I imagine it's hard to prioritize when you can write everything from like $100 million check up to, you know, at times like, you know, multi.

2:37:05

Billion dollar rounds are coming together.

2:37:17

Multi billion dollar rounds are coming together. Like how, like what, what, how do you prioritize your day and your week?

2:37:18

Yeah, I mean, the reason we raised this fund that's a little bit larger, you know, one, it's a reflection of the opportunity set. And two, we want to be able to say, hey, we can write a billion dollar check into a company directly out of the fund and we have resources to be able to do more than that, beyond it. But that was a big part. If we're high conviction in a company, we want to be able to do that. There's not very many of those. And so we have to use our time wisely. Fortunately, given our brand and given our coverage, we're able to see most things. We're able to meet a lot of entrepreneurs. And so we see probably as a firm, hundreds of deals or investment opportunities or companies a week. And then, you know, we have the, you know, we have the luxury of getting to kind of matchmake with the ones that want to work with us that we think are really special. You know, after this, I, I just, you know, was texting with a founder and. And he was like, he kind of cracked open the door that he might do a fundraise. And. And I'm like, great, all right, I'll meet tomorrow.

2:37:24

Yeah, in a couple minutes.

2:38:21

Let's meet our office at 4:00'.

2:38:22

Clock.

2:38:23

Like, you know, we got to do it. So the way I try to spend my time is like, you know, the thing that gives me the most energy is to meet those founders and make investments.

2:38:23

Yeah, well, we'll let you get back to chasing deals Good luck with that particular founder. Hopefully that founder will be on the show ringing the gong in just a few weeks.

2:38:31

I'm excited to see what you do with this, with this new fund preparation, timing, opportunity. Yeah, this is going to be an amazing chapter.

2:38:39

Well, have a great rest of your day. David, good to catch you.

2:38:49

Update.

2:38:52

We'll see you.

2:38:52

Cheers.

2:38:53

And up next we have Ben Horowitz, the founder of Andreessen Horowitz. The Horowitz in Andreessen Horowitz. Ben, how are you doing? Welcome to the show.

2:38:53

Good.

2:39:02

How are you guys?

2:39:02

We're fantastic. Massive news today. Congratulations. Obviously, we'll get into the fun structure. I'm sure we'll have a bunch of questions there. I wanted to kick it off with, with a reflection on your book. The Hard Thing about Hard things. What is the one piece of advice that you think has aged particularly well from that? What has never changed? And then maybe you could take me through some things that might have changed in this era. Bigger companies, AI. What do you go back to and what do you maybe think needs an update?

2:39:03

Yeah, well, I think it's still really hard to be an entrepreneur. And one of my favorite quotes in the book is something Mark said to me. You know, when things were extremely bad. He said, you know, one day we'll look back on this, chuckle nervously and change the subject.

2:39:38

I think some of us have seen this.

2:40:00

He would always say is things get darkest before they go completely black.

2:40:05

Yeah, I mean, it's underrated. How long you two have been in partnership beyond just this firm? You've worked together for so long.

2:40:09

30 years.

2:40:20

30 years. What a run. An overnight success. A true overnight success if there ever was one. How do you two like to work together now? How is the day to day working at the firm?

2:40:20

Yeah, I mean, I think that it works pretty well. I mean, we have pretty different roles. So I, I run the firm. And then, you know, Mark is kind of, in a lot of ways the face of the firm. And he also, you know, he gets very deep on specific things. So policy AI are kind of the two things that he's like super focused on right now. And he has many, many ideas about running the firm, and I have many ideas about things he does. And so it's very collaborative, I would just say. And we argue all the time about everything.

2:40:34

As any good person.

2:41:15

Sometimes he's right, sometimes I'm right.

2:41:16

Yeah. Well, how is the structure of running the firm? How is the structure of the firm changing in this era? Obviously the numbers are bigger, but on the fundraising side, but maybe not on the team side. What's changing? Is there anything that you've felt like this technology shift requires different management of the firm?

2:41:18

Yeah, no, for sure. I think that what's happened is we have such a powerful new technology platform that the number of really important companies that will be created out of it has just multiplied.

2:41:41

Interesting.

2:41:56

And you know, like the, the tech industry itself used to just not be that big and now the tech industry is all industry.

2:41:56

Yeah.

2:42:05

And that change is kind of what really changed the architecture of the firm. So originally, you know, we look like every other venture firm. We were, you know, a team of venture investors. You know, we were a little different in that we had a more elaborate platform. But now what we've done is we've kind of subdivided the technology market into all of its submarkets. So, you know, infrastructure, applications, you know, crypto, early stage stuff, bio, these kinds of things. American dynamism. And each of those teams is basically looks like the original Andreessen Horowitz, but they're independent of each other. And that enables us to both kind of COVID the whole market in a very, very serious way, but also be nimble and not have. I mean, you don't want 20 people in a room talking about a deal. Yeah, like you're not going to get to the truth like that. You know, just in my management experience, it turns out you can't have a conversation with 20 people. You can have a presentation.

2:42:05

How do you think about empowering the firm or the sub teams to become subject matter experts and actually investigate and prosecute deal theses in entirely new markets where no one in the firm might have ever done an oil and gas deal or a solar deal or some bio thing that's entirely new and you have a team. But there's new markets forming and new markets coming online as potential transformation targets for technology. How are you keeping the firm sharp on every corner of the global economy?

2:43:08

Yeah, so a lot of times there are super experimental things that we'll look at, but we don't necessarily kind of build the organization around yet. But once we commit the flag, then our big commitment would be, okay, we'll create a fund around it. So we did that with crypto. We made the coinbase investment before we had the crypto fund. But then as we got into it, we said, well, this is going to be a larger market and it's super different than everything we're doing, so we need to commit around that. More recently with AI, AI, like the way you build AI Companies. The nature of the AI founder is just so different than everything that we've seen before that we ended up bringing in a lot of expertise from the outside. We kind of reoriented everybody on the inside. Like we, we actually had, you know, a huge amount of training materials and like, you know, basically exams to make sure that, you know, everybody who has worked on that was what we call AI native and understood like all aspects of it before getting into it, just because, you know, these things do tend to be different. And this is why you, you see a lot of, of people age out of venture capital and then a lot of kind of firms be what they once were. They were very important in 2015, but they didn't necessarily make the transition. They didn't bring in the right kind of talent.

2:43:42

Yeah. When you're managing the firm, how do you think about the dividing lines and the walls between different funds? We've seen just with the Neo cloud, a lot of those folks started as crypto companies, then they became AI companies. But they're building things at such massive scale. I wouldn't be as surprised to see them in an American dynamism portfolio because they're sort of re industrializing. So are you the person that, the firm, that one of the subdivision leaders comes to to say, hey, I want this in my fund. How does that work?

2:45:16

Yeah, so there's not that much conflict in that. You know, the categories are pretty clear. There are, you know, it happens occasionally where, where they bump into each other, but, you know, for the most part it's like, what are you really trying to do? And then the entrepreneur will gravitate towards one of the funds based on what they're trying to do. Like, we want to sell things to the government. Yeah, okay, that's likely going to go into American dynamism. Whereas, like, like, okay, we've got eight PhDs in AI. That's almost certainly going to end up in infrastructure, kind of model world and that kind of thing. And so it's really matching the funds are you think about markets of entrepreneurs, and the funds are designed to address that market of entrepreneurs. And those tend to be fairly distinct. Now, sometimes people will try to game us and they'll get rejected by one part and then they'll go to the other. We have very, very, very good, comprehensive data on everything we've seen. We've got extremely good systems, so we catch those people.

2:45:49

When did you realize a $15 billion fund was possible? Did you imagine this kind of scale was possible from inception, or was it. Did you Build?

2:47:02

Yeah, Our first fund was $300 million. So we definitely weren't thinking about it then. We thought 300 million was a lot. And people thought we were raising too big a fund in 2009. But now what we've done is we've looked at the markets and said, okay, how big is this market? And then what kind of fund do we need to win in that market and generate large return? And we tend to have a relatively optimistic view of the future. I think there are some cynical VCs out there there. And, like, when I was a boy.

2:47:13

Valuations weren't this high.

2:47:52

She's like, play the game on the field.

2:47:54

Yeah. We like to look forward and not look backwards. And so as a result, like, something, you know, I think we have done a good job of getting ahead of the game. Like, when we raised Fund three, which was a billion dollars, we got a lot of criticism from other funds going, like, that's crazy. You know, no billion $40 fund has ever returned money yardy, yardy y. And we're like, well, okay, but, like, the world didn't look like this and software is eating the world and things are getting bigger. And we think that, like, we can deploy a billion dollars. And, you know, that fund, you know, had Coinbase and Databricks and Lyft and DigitalOcean and GitHub, and, like, a lot of big outcomes. And if we had that much money, be a problem.

2:47:57

Yeah. On that note of optimism and understanding the scale of the Internet as it eats the entire world, how did you process the bubble talk that took place in the back half of 2025?

2:48:43

Well, you know, I was CEO during another bubble. Yeah. So I know a lot about bubbles. Look, I think that. So there's a couple things that I learned from the bubble that we were in. One was.

2:48:58

Sorry, sorry, we keep a bubble gun handy.

2:49:12

Yeah.

2:49:16

Look, you know, one of the things, if you look back at that bubble, there were. There were a lot of things that were present then that are definitely not present now. So, like, probably the biggest being the Internet. Everybody knew the Internet was going to be giant, but at the time that everybody was investing all the money, the Internet was very, very small. If you go back to 1996 at Netscape, we had 90% browser share, and we had $50 million in revenue. We had 50 million users. Sorry, 50 million users. So the entire number of people on the Internet was 55 million. So you're funding these companies and giving them a $10 billion valuation, selling into a market of 55 million people, and then half of those were on dial up. So it was limited in what you could do. And so those valuations were running way, way, way ahead of the technology is kind of what caused the bubble. If you look at AI, the technology is like, working and getting to the world right now. Like, how many people are on ChatGPT and how is that business going? It had, I think, zero revenue in November of 2022. And I don't know what the current number is, but it's probably between 15 and 20 billion. We've never seen that before. So the things are working. The things that were bubblicious in 99 aren't quite the same. But to me, the biggest thing that I learned was right before the bubble burst, nobody thought it was a bubble. Warren Buffett, who had never invested in any tech, in early 2000, thousands started investing in tech. So everybody capitulated and agreed prices would never go down. Like that's what you need to get to a bubble. It's a psychological phenomenon, you know, not a. Not a financial phenomenon. And so, you know, right now, with everybody talking about a bubble, I was like, oh, great, we're not in a bubble. Because it's when nobody believes it's a bubble, that it becomes a bubble. Same with the financial crisis, by the way. If you look at, at the price, you know, the kind of interest you pay on, like, home loan debt in, in, in 2007, it was the lowest in history.

2:49:17

Yeah.

2:51:40

Right before it all came crashing.

2:51:41

It should have been the highest.

2:51:43

Right before everybody defaulted.

2:51:43

Yeah.

2:51:45

You know, it was the lowest in history. And that's because it was a bubble, because everybody believed, hey, they're not building any more land anymore, you know, like that. That's what's going on. And so once you get into that kind of psychological convergence, that's when you really get into, like, a really crazy bubble. Now look, in venture capital, everything is always priced at either half or double what it's worth. Like, that's the steady state. And so are there going to be things that are priced way too high? Yeah, of course.

2:51:45

Speaking of land, how are you processing the movement out of California? The news in California of the wealth taxes? A lot of folks are saying California might shoot themselves in the foot, kill the golden goose. How have you been processing the news?

2:52:16

Yeah, so it's very kind of an interesting kind of view of the world. I think that the groups in California have been pushing this idea. So, you know, we. I go all over the world. I've met, like, in the last year, you know, the president of Mexico, the president of El Salvador, you know, the crown prince of Saudi Arabia. So, like, I'm always with world leaders or I've spent a lot of time with them, and they always want to know, like, how do we create Silicon Valley here? And when you look, we want a golden goose.

2:52:33

We like your golden goose. We want one.

2:53:10

It's pretty remarkable that we've repeatedly created companies with larger kind of GDP than most countries. Routinely, we've done that. And so rather than asking, how did we do that? It's like, well, how can we rearrange it and run an experiment and see if it destroys it or not? And so I think that's probably the weirdest part of it for me, that people would think about it that way. If you start confiscating wealth and taxing unrealized capital gains for people who aren't liquid. Actually, we saw this in Norway. So Norway has an unrealized capital gains tax. Norway's got a lot of extremely smart people, great entrepreneurs, but they all left. And when you talk to entrepreneurs in Norway, they're like, well, I literally can't pay the tax because the company got marked up to whatever, a billion, $2 billion, and I own a lot of it, and I can't get that money out. It's a private company, and so I'm stuck. So I have to leave the country. And there are no entrepreneurs. There is basically no tech entrepreneurs at all in Norway now. And if you wanted to get. It's been so hard to break the Silicon Valley network effect, but this is the best strategy I've seen. If you wanted to wreck the California tech ecosystem, how are you processing?

2:53:13

It feels like today we have this incredible optimism within the technology industry, this incredible excitement. And then outside of the technology, your neighbor, somebody nearby has like this. There's. It feels like this real tension and kind of fear from broader society about the work that is being done within the technology industry. And you see interviews that, you know, AI leaders will give where they'll say, we're. We're summoning the demon. Or they'll say, you know, you know.

2:54:47

Not the most optimistic storyteller.

2:55:14

We're going to end. The world will end, but we're going to create some great companies. So I think people like this, these interviews and these quotes spread so quickly. A lot of people have heard them. And the question from the broader populace is like, hey, do we need to.

2:55:16

Do this optimistic vision.

2:55:29

Yeah. Or can we stop?

2:55:30

Right.

2:55:32

Obviously, technology is proven to be somewhat inevitable. Relentless.

2:55:34

Yeah. Yeah, yeah. So I think the good news is it speaks to the importance of the moment. So this is on the order of the microprocessors or the steam engine or something, or electricity or something like that. So. And those things all turned out to be like really good for humanity.

2:55:40

Was there that much with electricity, was there like the level of fear? Because there was people that would like go, and obviously I know the stories, people that would like, their job was to light the lamps, right? But.

2:55:59

Oh yeah, like if you go back and read about the beginning of electricity, it's why. Well, they made a law. When automobiles first came out, there was a law in the United States that said if you're driving your car and you see a horse, you have to stop the car, disassemble it and wait for the horse to pass. Like it was that level, that was the regulatory idea. So yeah, I mean, I think, by the way, watches were the same. You know, when watches came in, there was like huge fear that like people would never be able to have a conversation again because they'd be just checking the time always. Yeah, these technologies like generate a lot of fear. But I think that, you know, the good news on it is, you know, this one is really important. I think that the impact into the well being of humanity is going to be bigger than certainly anything in my lifetime. And you know, one of our bigger problems, I think is there are people in the industry going for regulatory capture who kind of feed into the fear, to the fear. And then like there are also people who have just, you know, it's moving so fast that it's actually freaked them out who are working on it. And that's.

2:56:11

How do you advise, how do you advise portfolio founders or even people at the firm around processing noise? I think historically, you know, there wasn't like this constant chatter, right. We have like X now, which is like a constant, you know, stream of consciousness from millions of people that are sharing their opinion. And it's, you know, I know a lot of entrepreneurs that, you know, one day everybody's saying that they're the greatest thing ever, and then the next day, you know, people start to criticize. And how do you kind of like, what's your, what guidance do you give there?

2:57:24

Yeah, well, I think that like the world of media changed and I think it's tricky for people and companies to process because if you grew up in marketing or in old media, your whole concept of the laws of physics is different. So in old world you were always thinking defensively because there were very Few channels to get your message out. The format was very tight. You know, you could get a quote in here, you could get a few sentences before the host cut you off or whatever.

2:58:01

Yeah.

2:58:36

You know, you guys watch CNN from time to time. And so like, in that world, the way you would think about media is just like, let's make sure we don't say the wrong thing. Let's spend hours and hours crafting the message and so forth. You know, in the new world, it's like wide open. There's. There's media everywhere. The formats are whatever you want it to be. And so the right kind of way to think about it is you have to be interesting and don't worry about making mistake because you can just come back tomorrow and flood the zone. You know, like, just keep going and that. I think it's. I found it very, very difficult to reorient somebody who has spent a career in old media world kind of thinking in a new media way. And so the biggest thing that, like I really talk to our CEOs about is like, you've got to approach the. You have to approach new media with new media thinking, new media people, that kind of thing. And it really. It's a remarkably opposite world. It's like, you know, it's like you're landed on Mars and you're like, well, what the fuck happened to gravity?

2:58:36

You know, different.

2:59:47

And you can't even say, well, no, gravity is different here because it's like. No, no, gravity just is like. I can't deal with the fact that that's just. That's just the truth.

2:59:49

Yeah.

2:59:59

Well, we would love to keep talking about media.

3:00:00

Yeah.

3:00:03

Very few things that we enjoy.

3:00:03

We should bring the gong.

3:00:06

But we know you have a late. You got late fees if you're late to meeting. So this gong is for the whole A16C team.

3:00:07

Congratulations.

3:00:14

And we won't keep you any longer, but come back on again soon and. And congratulations.

3:00:16

Thank you.

3:00:21

Great to see you guys.

3:00:22

We'll talk to you soon. Goodbye. And with that, we need to check in on the Claude codes.

3:00:23

Eric just said that. That they're good for the 10 bucks. Twerenberg, he says he's gotta, you know, he's one minute late to his next meeting.

3:00:31

Oh, oh, yeah, yeah, that's right. That's right. They have a late fee.

3:00:38

Yeah. For those that don't know and treason partners. If they're late to. If they're late.

3:00:43

If they have phones too. If they have phones out, they get fined.

3:00:49

Yeah, no, but it's Late. If you're late to a meeting with a founder that you're looking to invest in or you're just meeting, you got to pay if you're late.

3:00:51

Well, we should check in with Tyler. There were some rumblings about a change to Claude code.

3:01:00

Can you break it down?

3:01:05

What happened?

3:01:06

Basically what happened was, okay, so when you get a Claude subscription, there's Pro and Max, you get like Claude code credits, basically. And so what was going on was there's like third party harnesses. So there's one open code. There's like a bunch of these and they would basically use the credits that you get from Claude code from your Claude subscription, like use. And it's like they're like open source agent, whatever.

3:01:06

Sure.

3:01:34

So anthropic stopped that. So you can't use your base, you can't use your subscription as like the credits. You have to use the actual API.

3:01:36

The actual API. Okay.

3:01:43

That's like the main thing. It's like, not to me, it doesn't seem that crazy of a thing because you can still use it with the API. It's more expensive.

3:01:45

Is it an exchange rate thing?

3:01:52

Maybe?

3:01:54

Like, if I'm on Claude Pro or Claude Max, am I getting on a per dollar basis more tokens than I would if I. Okay, yes.

3:01:54

That's why there's like a lot of arbitrage.

3:02:04

Because they were getting arbed, they said no more Arbing us.

3:02:06

You can think of like, cloud code is going to get much better if more people use it. Right. Because it's like an RL environment.

3:02:09

Sure, sure, sure.

3:02:14

So the value of the data, so.

3:02:15

They shouldn't be incentivizing people to go elsewhere, but they are still allowing people to go elsewhere just at the consistent actual API rate.

3:02:16

So you can still like, it's like bring your own key. You can do that.

3:02:25

Okay, that doesn't seem like too.

3:02:27

It's not that crazy. But people are very mad. They're canceling their cloud.

3:02:28

People are mad, mad at Claude. So sad. Oh, well, I'm sure that they will figure it out and the fun will continue. We do have to cover another story in the AI world. Logan Kilpatrick, friend of the show over at Google, he said, I'm happy to share that we, the Google AI Studio team, are now sponsoring Tailwind CSS, the project that had to lay off three people. And it was very dramatic because it was 75% of the team. But their business model was not really working because they were selling templates which of course could be assembled by AI agents in the modern era.

3:02:31

Yeah, I love this.

3:03:09

Logan said honored to support and find ways to do more together to help the ecosystem of builders. And I said, you drop this king.

3:03:10

And I gave him the crown.

3:03:17

Mario holding the crown. If you scroll down, you should see it.

3:03:18

Yeah, this is great.

3:03:21

That's me.

3:03:21

I expected this. I expected this to happen pretty quickly. I'm glad Logan made this move and I think a handful of others did as well. So hopefully Tailwind can hire back the handful of engineers that they were forced to let go.

3:03:22

Yeah.

3:03:37

And yeah, this should give them some more predictability while they figure out the next chapter.

3:03:38

Yeah. Well, in other news, OpenAI is reserving $50 billion for a stock grant pool. Jack Rain says 500 billion company doing 13 billion in revenue. Projecting 50 billion in equity comp is so good.

3:03:44

So price of price of SF real.

3:03:58

Estate is going up, price of the AI research are going up but they have the money to distribute. That is a big equity pool but it's an opportunity to join pre ipo, get some shares and hopefully do well for yourself. Well thank you so much for watching the show and tuning in in today. We will be back first week in the books pacific on Monday.

3:04:00

Only three more sleep podcasts until we're.

3:04:21

Back Spotify and subscribe to our newsletter tbpn.com and with that we'll say goodbye and have a great weekend.

3:04:24

Have an amazing weekend.

3:04:30

Thank you for watching.

3:04:31

We love you.

3:04:32

See you Monday. Goodbye.

3:04:33