History’s Largest Oil Disruption, Oil & AI, Sundar's New Pay Deal | Alex Epstein, Dr. Alex Wissner-Gross, Charles Lamanna, Julien Bek, Eoghan McCabe, Michelle Volz

you're watching tvpn today is monday march ninth we are live from the tvpn ultradome temple of technology fortunes of finance ramp dot com comma time is money save odds use corporate cards bill pay accounting and a whole lot more all in one place we have a fantastic show we were able to get alex epstein to come down to the ultradome in person on short notice to get us up to speed on what's happening in the oil markets of course oil is all it's gushing all over the timeline all over the financial times all over the the wall street journal as

0:00

part of the and alex had a feeling that oil would be important in the future when he wrote the fossil

0:32

future yes which we have right here which if you haven't given it a read go pick a copy up and stay tuned for him joining us at one ten today let's pull up the linear lineup and tell you who we have coming on the show today linear of course is the system for modern software development seventy percent of enterprise work workspaces on linear using agents now we

0:38

have doctor alex wisner gross over on eon systems he went very viral over the weekend for connecting a biologically derived fruit fly connectum to a mujukou physics simulated body and we're going to be talking all about simulating fruit flies and what will come after that and we're

0:58

also going over microsoft copilot coworker julian

1:25

which we're gonna call coco coco okay

1:29

disney might like a word they have ip in that space in particular but

1:32

we'll see i see they have it

1:35

in the enterprise i don't think so julian's joining from sequoia capital owen mccabe from intercom our partner is joining because they just raised a massive clean two hundred million dollars and michelle volz is launching pax vc she left andreessen horowitz i think about a year ago and has been is now launching a new early stage fund anyway let's go through the timeline and see how people are processing the news of the strait of hormuz and the oil price spikes fox

1:36

news this morning they were saying just basically their advice to the captains was to man up and just send it

2:07

yeah send it that seems crazy effectively their advice i'll read the vessel please

2:16

stay safe i'll read the actual quote a guy brian said if you want to diminish the iranian threat if you want to make sure this ends with complete capitulation show some guts and go

2:20

through that strait that seems very very risky right now stay safe out there if you are on a shipping vessel do whatever is responsible there are some crazy twists happening so apparently ships in the gulf are declaring themselves as chinese vessels to dodge attack this is from the financial the financial times at least ten vessels have changed transponder messages in an apparent attempt to avoid becoming targets a clutch of vessels trapped in the gulf under enemy fire are adopting a tried and tested ruse to avoid attacks they're changing flags using transponders to declare themselves to be chinese there's always been a very odd like just like tug of war between how ships identify themselves because often for tax reasons they're bought in one country operated individuals from another company but they fly a different flag to be able to go from one place to another and it's all based on like the port systems i don't fully understand it but it is very interesting at least ten ships over the past week have altered their destination signal to read chinese owner all chinese crew or chinese crew on board about one thousand ships are currently shut inside the gulf and its immediate surroundings the cumulative value of twenty five billion dollars and

2:32

i don't know if you've seen some of the maps that show the strait where it looks like nothing is actually moving through yeah i think in actuality a number of the ships are actually turning off their transponders so you can't see the movement because they're basically moving a little bit going through the strait and then turning back on again well

3:48

there are plenty of outlets that are covering the iran conflict in detail we of course will be focused on the business and technology impacts i wrote about what oil prices mean for the ai buildout in data centers it's just sort of interesting to dig into the deeper supply chain of artificial intelligence but there are some posts that we should go through around the oil story so crude oil is five standard deviations above its fifty day moving average statistically speaking this occurs every nine thousand five hundred years so the last time would have been about six thousand years before moses parted the red sea imagine what that did to shipping in the air area fanciful what else is going on the kobse letter has been posting a lot about oil prices us oil prices are now expected to rise above one hundred dollars a barrel this month that already happened this was posted on march sixth with markets pricing a sixty six percent chance that it happens the last time oil prices were above one hundred was july of twenty twenty two let's pull up

4:08

this clip from landman i haven't seen

5:12

landman have you watched it is it good well you want oil to live above sixty but below ninety and don't get me wrong we're still printing money at ninety but gas gets up over three hundred fifty a gallon it starts to pinch it hits one hundred every product in america has to readjust its price dollar seventy eight a barrel that's about perfect it brings enough profit to keep exploring but it don't sting as much at the pump less course here in california i mean they tax the out of it out there it could be forty five a barrel and it's still four dollars a pump i don't know how those son bitches do it out there it's movie day twenty twenty a barrel of oil was worthless this place became a ghost town and nobody's immune kids have to quit college trucks get sold or repoed we can pause

5:14

it now we got we got the

6:07

gist of it yeah well this is

6:08

the largest how old were you when you discovered that gas is really expensive in california not just because there's a lot of demand for it i mean

6:10

i discovered it when i was filling up in montana and it was like dollar two a gallon and it's like dollar five a gallon here that was pretty wild the first major gas price shock that i noticed was hurricane katrina in my life i think it was in high school because i was too

6:18

young i was talking about just the fact realizing that so expensive because of the taxes that california puts on it

6:33

yeah because it's nice to drive around here put the top down the weather's good so they're like yeah you're going to pay you're going to pay more for the joys of driving an internal combustion engine car now the price in california is aggressive i don't know the structure of the prices though is it percentage based or fixed because that affects how much the price will move based on oil price shocks because if the price per barrel doubles but the tax is flat you don't feel it as much here as you do other places so i don't know anyway i'm pulling

6:41

it up but you can run through

7:17

this the largest supply shock by a factor of four so the hormuz blockade which is current twenty million barrels were lost in supply the iranian revolution in nineteen seventy eight was five point six the yom kippur war embargo in seventy three was four point four the iraq kuwait war was four point three in nineteen ninety the iran iraq war in nineteen eighty was four point zero and the ukraine russia invasion in twenty twenty two which is the last time that oil spiked over one hundred dollars a barrel was one to three so an absolutely huge supply shock and i'm sure it will have a lot of implications all over the economy with triple digit prices here's what's going to happen now says policytensor markets will scream when they open tomorrow vix will surge to levels beyond what we saw in april the sell off will continue for some time as intermediaries shed risk and the markets are red they have been screaming today the vix futures curve already has inverted bid up by dealers looking for insurance this predicts a massive sell off the pressure on this captured white house now the pressure on this captured white house now beings in earnest that's sort of oddly anyone can tell them that if this persists for very long it will destroy the trump presidency and gut the gop for a generation the controlling factor here as i have told you over and over again is that the united states does not have the military means to reopen hormuz there's no military solution in sight this means that not only does iran have the strategic upper hand now it means iran that iran enjoys the unambiguous strategic advantage all they need to do is keep the thing closed until he capitulates i put my neck out bar to call this in advance and someone told him yesterday i was in the minority perhaps even a minority of one no longer i was correct just calling a shot the blob heads and scribes were incorrect in their assessment of the strategic situation and now markets will price to reflect reality and we briefly touched on scott sumner's blog post on substack at the end of thursday's show maybe friday's show we didn't get a chance to read it i actually read through it earlier today and it's pretty interesting it basically makes the case you know it has a very controversial tense freak out and the thought is that like he's doomposting maybe it's about ai maybe it's about this particular conflict he is more just reflecting on this dynamic between when the market freaks out it acts as a moderating effect to policy and so he gives a number of examples around like tariffs or more

7:18

specifically the admin yeah yeah like the

9:59

tariffs caused this massive circuit breaker five percent sell off in the market and then that was internalized and very quickly adjusted and there were a whole bunch of different carve outs to sort of like create soft landings and so he's actually and what he's getting at is that after the fact a lot of people say look you didn't need to freak out because the taco happened trump always chickens out the actual proposed policy effect didn't go into effect and his point is that well it's precisely because people freaked out that it didn't go into effect so freaking out is good in scott sumner's mind at least anyway i love this yeah we have a solution we have a solution if you're feeling the pain at the pump pivot and get a horse at oil at one hundred ten the urban horse is the only option pulling up to the gas station on a horse is truly elite i do want to know what's the tco on a horse with the food and the stables versus just keeping a what is that a ford taurus in your i don't know that's something else in your garage the horse really mogs at the gas station in particular right because it's just making everyone feel so stupid yeah we got to go back one horsepower is all you need nick carter says seeing a lot of non process trusting pannikins on the tl and then he followed up by saying that's what i thought because the oil prices spiked up and then they fell down and we did not get dollar one hundred twenty a barrel oil we got exactly one hundred which was still

10:01

it's not like it's way way way too early to celebrate or anything yeah

11:43

yeah exactly there's an interesting chart here a really shocking chart from oliver grob unreal numbers about germany's nuclear power generation so if you scroll down look at this curve this is truly the bell curve meme or something like that jp morgan estimates that had germany not phased out nuclear power the country would have generated fifty percent less electricity from fossil fuels and eighty four percent less electricity from natural gas in twenty twenty four electricity prices in germany would have been around twenty five percent lower and the country would have imported half as much electricity and just complete rise and fall of nuclear power generation in germany one of the craziest graphs you don't see graphs like that very often in new technologies typically you see s curves or you see exponentials no one considers the models get better and then they get dumber that's certainly the funniest outcome very very rough goldman sachs sent a note to investors saying if oil prices increase by ten dollars and remain elevated for three months us year over year headline cpi inflation would likely rise from two point four percent in january to three percent in may those are small numbers but we're looking at an oil price increase of maybe dollar thirty dollar forty over the baseline potentially higher we don't know where oil's going to land and so you have this weird tug of war right now with the fed where if oil prices go up inflation goes up the fed has a mandate to curb inflation that means higher interest rates at a time when the labor market is shedding jobs you would expect a fed rate cut or a lot of people are optimistic about a rate cut trump certainly wants rate cuts but if inflation is climbing there's really no solution other than keeping rates higher or even raising them further so a real jam in terms of federal monetary policy fed monetary policy here's art cash in at the opening bell this is a historical video when was this this was a long time ago let's play this clout

11:47

this may be it first let me start out muratori te salutamas essay and

14:01

you know that's the gladiator salute we

14:07

who are about to die salute you so it's gonna be a tough morning

14:09

this may be it we were about to die salute you insane insane aura for you know cnbc really doesn't get enough credit for being so innovative in terms of broadcasting and entertainment really some

14:13

of the greatest clips i love one of the top comments this is an old cl yeah obviously brother four hundred

14:28

eighty p it's probably from the nineties so zerohedge says with oil at one hundred and eleven total panic and memetic sisyphus shares a clip that says half of this site for the last week has sounded like this let's producers are telling me there is breaking news the

14:36

asian financial markets have just opened to

14:53

a huge sell off and we're gonna

14:56

switch to that story right now good

14:59

i'm glad i'm here your thoughts tracy jordan on how this is gonna impact

15:01

wall street larry i'm not an expert

15:05

but i do have a strong opinion

15:08

new york as we know it will no longer exist tomorrow producers are telling

15:09

me there is

15:14

is that from thirty rock that's so good that's so funny oil came way in from its overnight highs says joe weisenthal and the qu

15:16

chris paul hits a huge three to cut down the lead to forty two

15:30

absolutely absolutely brutal so let me tell you about gusto the unified platform for payroll benefits and hr built to evolve with modern small and medium sized businesses

15:35

stop making fun of us we're somewhat of a smodern business small and modern

15:48

small and modern smodern let me also tell you about crowdstrike your business is ai their business is securing it crowdstrike secures ai and stops people so we

15:52

gotta give a shout out to george kurtz for the mercedes performance yesterday oh yeah one two one two really good george kurtz would you look at that takes a stake in the team suddenly

16:01

they're at the top yeah he's on to something well my essay this morning was titled why is no one talking about oil of course everyone is talking about oil oil discourse gushed onto the timeline this weekend crude prices spiked to nearly one hundred and twenty dollars a barrel as a broadening war in iran threatens both transportation routes and production the geopolitical and economic analyses are flowing but what does this mean for agi timelines and a lot of people in the ai world are sort of tuning all of this out because they see recursive self improvement agi asi the build out as more important and i just wanted to sort of reality check the ai supply chain to understand how does oil actually affect data center construction ai production token pricing like is there any effect my conclusion was that it's very moderate but there are some interesting effects in the financial markets that are probably the bigger takeaway but it's still interesting to hear about like yes oil actually does is used in the production of ai at least a little bit so power has been at the forefront of the ai pushback like everyone's worried about these local energy prices these electricity prices increasing near the data centers but and it's been it's become like a political issue but pain at the pump might become a bigger story as gasoline prices spike and that has been pain at the pump oh it's been the most tangible sign of inflation it moves so quickly you know one one jitter in the

16:13

economy and it's a huge component you know people on the coast people in tech don't have a good sense for this right if gas gas for a lot of people gas could quadruple and it wouldn't they wouldn't really notice it no no but if you actually look into the average american how gasoline fits into their budget it's a meaningful component of their monthly budget so they feel it super intensely and variable cost there

17:46

are there are so many different ways where a lot of americans go on driving vacations that obviously is directly impacted by gas prices and then also just psychologically there's something about filling up at the gas tank where you see the number ticking up and you're doing that on a every week basis or so that it's just so visceral it's this thing that you have to stop and then go experience and watch the money flow out of your account like in real time it's very visual it's very

18:11

interactive yeah i remember i must have been probably eighteen at this point where i would just go and i would like for a long time i just put i'd like you know prepay for

18:40

a certain amount i got twenty bucks that's how much let's see how much

18:50

i pay that i felt like really i felt like the king of the castle i just put my card down and let it run up yeah

18:53

no no totally so pain at the pump's been like a big broad economic issue it's so visceral like the actual phrase has a lot of power in politics and in like attack ads like your gas prices are going up under this person that's why you gotta vote for me this is this is common but in ai circles the discussion's been much more focused on rsi now is the new acronym that everyone's focusing on not agi agi is here we know artificial general intelligence we passed the turing test the models can do things generally intelligently but can they recursively self improve are they rsi are we in rsi now is it coming is this going to be a fast takeoff is this going to be a slow takeoff well something's taken off dylan patel said being in sf is like being in wuhan right before the pandemic something is happening it's gonna hit everywhere but so few people know it so he's sort of echoing that something big is happening mentality this idea of like recursive self improvement being agi pilled a lot of smart people agree there's definitely something happening there's still this question of like what is sticky what where will diffusion take longer what will be sort of agi resistant or asi resistant for a variety of reasons even if they're completely irrational and you could get a better thing from a computer but people like the human version

19:01

or whatever the irony is that george hot's hitting the timeline to raise money makes me more bullish on acceleration right because if not he's obviously not historically been a huge fan of venture capital

20:27

well he's not raising from traditional vc's he said he specifically said he wants to raise from basically friends and family type of round he does want twenty million we have this in the timeline somewhere where is this post from george

20:44

i'll pull it up if you want

20:57

to keep tiny corp he said that he's raising twenty million dollars to buy a building he's going to buy a building here it is so he said this is from tiny corp the makers of tiny grad and the tinybox if tiny corp was raising twenty million dollars at two hundred million dollars valuation who'd be interested business model is basically this buy an eleven point five million dollars building with five megawatts of power link in our discord wait for amd to launch the rdna five ninety six gigabyte cards mid twenty twenty seven pre order three thousand cards hopefully we can negotiate for two thousand five hundred dollars each build five hundred twenty thousand dollars tiny boxes with six of the cards in each box run all the chinese llms make six hundred thousand dollars per month revenue selling tokens on open router market depth is there this is one percent of open router improvements to tiny grad yield revenue improvements due to how power is priced in oregon it's only like fifty thousand dollars for the electric bill before the four megawatts before they price for peak not usage we get like three c kilowatt hour power three cents per kilowatt hour we can also make one hundred thousand dollars per month leasing colocation space to comma building and cards paid off in three years max and investment made back low risk of being undercut since we're using consumer gpu's and running the cheapest colo you can believe if someone chill wants in i do it i'm not going to fake hype i'm not going to hype fake tech but demand for tokens is going to skyrocket look at the open claw install numbers with crazy good optimizations we could potentially get three x more from the machines and we have electricity for three x more machines five point four revenue per month then continue to scale from there custom chips et cetera he's starting a neo cloud or he's starting yeah he's going to be serving tokens so that was incredibly bullish demand is definitely there at the same time he sort of took the other side of like this is different this is crazy he says there's no agi there's no magic threshold you guys see auto research change the random seed from forty two to one hundred thirty seven and omg it's agi it's over yet you critique the junior engineer for the same stuff the cost of development is falling overpaid engineering struggles to compete that's the story and he says humanity has been recursively self improving for centuries so everything new is

20:58

old i don't know i obviously have a huge amount of respect for george but at the same time saying there is no artificial general intelligence when you're also arguing that engineers are struggling to compete with new technology that is replacing them how is this not artificial intelligence

23:31

competing with yeah so i think he is agreeing with that but it's that there is an immense amount of desire for this binary moment this is the singular this is agi this is asi the rsi is here this thing is happening right now and there's before and after and everything has changed and he just doesn't see it that way i think he sees it much more like the internet the mobile phone like other technologies that have been rolled out electricity yes there is like a before and after but you can only really define the period by maybe a decade and you need a few decades to understand that moment it's very hard to go back and there is like the iphone moment and there is like the first launch of i don't know aol i don't even know what the iphone moment of the internet was just because it was sort of a slow rollout anyway live gpu clusters where's jordy going oh you're moving the goalposts okay jordy's moving the goalposts where are you moving the goalpost to just over here okay but why is there a reason well i

23:53

mean they have to be moved george

24:59

is moving the goalpost oh okay okay okay he's moving the goalposts so i said you know it's clear that the ai industry continues to grow and continues to need more and more power and compute as we've seen from george hotz's new project that means large data center campuses but if they're not in random office buildings that george is picking up on the cheap they're probably going to be built with construction equipment so what does this mean they don't just drop from the heavens they require building which requires oil but how much oil and is oil a serious and is a serious oil shock enough to impact the ai buildout in a meaningful way spoiler alert basically no but live gpu clusters in the united states do not use much oil directly only zero point six percent of us electricity in twenty twenty four came from petroleum we're much much more dependent on natural gas something like forty two percent of us electricity is natural gas and so america ramped up natural gas production significantly over the past two decades a lot of that was in reaction to the wars in the middle east hey we need to be less dependent on foreign oil we need to be energy independent and so you have the fracking boom the natural gas boom and that's where a lot of our fossil fuels come from today i believe data centers only consume a single digit percent of us electricity so you're looking at zero point six percent of a few percent is like the actual impact so the short term impact of high oil prices should be very limited on ai building out new capacity is

25:00

yeah that said it's worth noting i think i think it's something like qatar produces something like twenty percent of the world's liquefied natural gas right so but not for us yeah not not for us so we're in a good spot but this is going to impact the data center build outs or active data centers across the rest of the world

26:39

if this continues yeah so when you're talking about building new capacity building new data centers oil is a little bit more involved so diesel powers trucks trains boats barges generators pumps compressors excavators and tons more construction equipment petroleum is also broadly used for plastics polyurethane and solvents that all work their way into the data center supply chain the biggest problem is delaying already tight schedules because of narrowly available components going out of stock so the price of oil goes up there's one marginal factory that can't produce one ingredient that goes into the rack and that slows things down you have to wait a week while you find another supplier that stuff can add up to just a little bit of a delay this happened during COVID and the ai industry was already experiencing something similar with transformers and so you don't want products getting stuck in transit or going out of stock but the bigger problem and the one that people should be talking about and i think you were debating with dan primack at axios about this is macroeconomics so higher prices higher oil prices lead to higher inflation if the fed has to raise rates to control inflation capital formation for mega projects gets a lot harder so jll has this estimate the next one hundred gigawatts of data center capacity could require about eight hundred seventy billion dollars of new debt financing and so using this rough number every extra fifty basis points of borrowing cost on eight hundred seventy billion is four point three five billion in annual interest expense and so that's i don't know if that's a huge squeeze on cash flow but it's certainly a squeeze on cash flow if you're a hyperscaler and you're going gig along ai capex now and you're like yeah we're going to do two hundred billion this year we're going to do five hundred billion next year or whatever across a couple companies you start looking at interest expense bills that are already huge but then they're going to go up by five billion or up by a full extra billion just with a fifty basis point increase half a percent so half a percent change in the fed funds rate can echo through even a quarter percent could probably be amplified into a multi billion dollar like cost line on your on your balance sheet or on your income statement as you're servicing that debt overall higher oil prices do make the ai rollout harder but not really for existing ai capacity i wouldn't expect token pricing to change based on the price of oil but there is this more there's this bigger question of macroeconomic resilience during a time where our largest tech companies are digging through the couch cushions to find every penny to win the ai future middle eastern investors pulling back i agree with you i do think that this is a significant risk because if they pull back on big investments makes these mega rounds more difficult retail investors could fly to oil stocks or defense stocks they think that this is going to be a thing for months and months and months it could all flip if there's a quick resolution though which is what i'm hoping for but if things continue the ai industry walks an even tighter tightrope yeah i

26:59

mean the big question right now is the hundreds of billions of dollars of sovereignty ai projects in the middle east right i think a lot of those people are going to be like do we want to send billions of dollars of gpu's over there and then also

30:08

the money coming here is another thing where you might want to spend it elsewhere quickly let me tell you about turbopuffer serverless vector and full text search built from first principles and object storage fast ten x cheaper and extremely scalable and let me also tell you about public investing for those who take it seriously stocks options bonds crypto treasuries and more with great customer service whatever your

30:22

investing strategy is we have to say happy birthday happy birthday to trey happy birthday trey we love you i'm excited

30:40

to have you thank you one of

30:46

our strongest soldiers we appreciate you it's been an honor podcast have you been

30:47

noticing that it's been hotter in los angeles i have downtown los angeles is forecasted to approach one hundred degrees fahrenheit on thursday and friday in march which

30:52

is why not good we're going to do the weather new segment for you today we're doing the weather on tvpn we have our very own ben we

31:03

have ben hello guys how are you

31:13

we're doing great tell us about the

31:15

weather what's happening well i want to

31:16

start off by saying as you can see the weather today for the low today is going to be seventy five degrees fahrenheit up there high of one hundred degrees fahrenheit down there but there's something i actually wanted to point out that i saw and i thought that was quite interesting as you can see up here there's a localized low pressure area up there and a localized high pressure area right down there if you can see that right does that mean rain the issue with that's not normally an issue and not a cause for concern and it's not very common for this time of the year however today later in the afternoon these two areas are going to collide and they're going to hit each other really and what that's going to cause is a barometric pressure inversion okay it might sound a little bit scary but i guarantee there's no cause for concern all that means is that hot air rushing in from the west is going to collide with that cold air rushing in from the east and it's going to cause a bunch of turbulence in the sky moving all the airwaves around and oscillations in the sky however i want to add one more point a byproduct of this effect is that all that humidity that dropped after that hot air moved to the bottom is going to raise up because the water cycle you know evaporation and stuff it's going to raise up into the sky into those clouds going to cause big clouds in the sky and eventually all that water is going to fall down onto the ground we're going to have big rain later in

31:18

the afternoon wait it's actually going to

32:25

rain in la yeah it might sound crazy but i just want it for all you guys at home i'd definitely try to step outside with a jacket today maybe a hoodie just in case

32:28

the rain comes don't try to be

32:35

a big shot i'm fact checking you right now and the weather app says it's going to be sunny all week is this just complete fake news my

32:37

team those are the numbers my team

32:45

gave me this is complete fake news no no no no this is the fakest news i've ever heard no no

32:46

all that crap literally the transatlantic current

32:51

john john i'm i don't trust you your app i trust the weather apple

32:53

says it's not going to rain the entire month there's zero chance of rain

32:58

did you look at the transatlantic current

33:04

no the transatlantic precipitation zero inches today that's putting you in the tomorrow zero

33:05

inches on wednesday john you're really going to you're really going to trust you're really going to trust an application that was probably vibe coded yesterday over ben who's doing the weather how did he

33:10

get here what happened here

33:21

no if you look right there you can see a localized high pressure area that's a

33:25

localized area i think this is an over eager weatherman who's just looking for drama in the most boring weather market in america which is los angeles i

33:28

think this is the most important story

33:37

in the world this is ridiculous you

33:40

guys can look at the jet streams they're coming in from the west i

33:42

don't want to hear any more mumbo jumbo about jet streams get out of here you're done great work ben thank you ben the fake news weather will return maybe never who knows let me tell you about restream one livestream thirty plus destinations if you want to multi stream go to restream dot com and let me also tell you about gemini three point one pro with a more capable baseline it's great for super complex tasks like visualizing difficult concepts synthesizing data into a simple future or bringing creative

33:44

projects to life this is the future of the weather you have a weatherman who gets into a live altercation with one of the other hosts he was just constantly fact checking it's on site

34:12

next time ben it's on site if you pull up with some fake news on tvpn's weather report it's on site

34:22

ben you crushed it good job so

34:28

it's not adam smith's birthday it's the twenty fifth anniversary of the publication two hundred fiftieth two hundred fiftieth sorry two hundred fiftieth anniversary of the publication of the wealth of nations tyler you have a copy i have two of our

34:33

copies here usually i have them with me but i didn't bring mine to work because we we have a few

34:47

here yeah yeah yeah obviously for those who don't know give us the pitch for wealth of nations why should people read wealth of nations in twenty twenty

34:52

six yeah i mean this is kind of the og economics book right it

35:00

was like put it into like ai terms is this like situational awareness type

35:06

of this is like attention is all

35:11

you need it was like the chatgpt moment for capitalism yes but i would

35:13

like to you know i have a few quotes i would like to read okay all right let me open this

35:18

let's see it's a big book it's like eight hundred pages yeah it's very long and it's really really dry i mean it reads like it was written

35:24

two hundred fifty years ago okay so this is i think this is kind of one of my favorite quotes kind of the seminal every individual necessarily labors to render the annual revenue of the society as great as he can he generally indeed neither intends to promote public interest nor knows how much he is promoting it yes he intends only his own gain yes and he is in this as in many other cases led by an invisible hand to promote an end which was no part of his

35:30

intention yeah rational economic actors homo economicus thinking of the value for society is created by rational self interest sort of the spin on like greed is good it was like he was like an og of that thesis greedmaxer yeah productive labor efficient markets free markets don't hold gold don't control trade an important read in twenty twenty six somewhat heretical there are many people that disregard that book maybe they should not maybe they should give it a cracker evergreen i think it's evergreen i think it's evergreen i mean there was that debate over like tyler cowen over the weekend and somebody was taking shots at stubborn attachments for being like not like not advancing the discipline of economics and my pushback to that was that i actually think it is incredibly it's just a fun read it's like really well written and it's very entertaining and it's very thoughtful but it's particularly interesting to hear someone sort of just rehash the tenets of liberal democratic capitalism in twenty twenty i think he read it in twenty twenty three twenty twenty two something like that but there are a lot of voices out there who are anti capitalist anti democratic anti liberal and tyler cowen is just retreating to a time tested time honored economic philosophy that is deliberately not revolutionary and i thought that it was a great book for that reason and that's why i recommend stubborn attachments to so many people and so widely but if you want to crack open the real hard stuff pick up a copy of the wealth of nations why is one of those copies so much smaller is the font size just smaller yeah it's

36:00

super small it's also two columns oh

37:52

two columns wow yeah if you can get through that you're truly elite in the global economy what does the timeline have to say about the masterwork that is turning two hundred and fifty it's now needed more than ever says the washington post on march ninth seventeen seventy six four months before the american colonies broke with britain over the issue of taxation a little known scottish thinker published a long dense book with an unpromising title shots fired an inquiry into the nature and causes of the wealth of nations two hundred fifty years later

37:54

adam

38:33

smith is by any objective measure easily the most widely cited and widely quoted economist who ever lived astonishingly his work still frames the central questions we face not just about free markets trade and capitalism but about the nature of human society itself and even what it is to be human at all smith was born in seventeen twenty three in kirkcaldy scotland educated at the universities of glasgow and oxford he initially made his name not as an economist but as a moral philosopher his first published book the theory of moral sentiments offered a radical theory of how we form moral judgments radical because it derived from the creation of moral values not from scripture or divine grace but from human sympathy and mutual regard the wealth of nations as his second work second major work came to be known was an extension of that product the book is not as sometimes believed a hymn to greed a pain to market fundamentalism or red in the tooth claw capitalism it was an attempt to understand how to how a commercial society could generate prosperity without collapsing into corruption this two hundred fiftieth anniversary is not a moment for hagiography it is an opportunity to recover a way of thinking that is directly relevant indeed urgent to the economic social and political challenges we face today begin with trade in his own time smith's great target was mercantilism or what he called the mercantile system the idea that wealth consists of whole purported bullion and that trade is a zero sum contest governments granted monopolies imposed tariffs and manipulated commerce in the purported pursuit of national strength producers were widely protected consumers often ignored altogether against this smith one of these days

38:33

we should have tyler just do a speed reading of the wealth of nations live on the show as long as

40:15

speedrun yeah wealth of nations speedrun any percent glitchless you need to do glitchless though for sure a glitchless run is the only way we can do it no glitching no ai summaries i think that would take you probably five days probably five days it's so dense have

40:21

you seen how fast you can rip

40:43

a rubik's cube this edition is four hundred eighty pages around oh yeah i

40:44

guess you don't have to read all the footnotes and stuff you can kind of skip through it but it is a thick book but the way it's written is just really not fun to read it reads like it's two hundred fifty years old it's not like a

40:48

page turner well we need to talk

41:01

about the wealth but first vibe co where dtc brands b two b startups and ai companies advertise on streaming tv pick channels target audiences and measure sales just like on meta and let's also 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 we need

41:05

to talk about the wealth of sundar pitch i pitch ai sundar new pay deal worth up to six hundred ninety two million is this like ten times

41:22

what tim cook's making no it must

41:34

be overtime right cook's making around seventy a year seventy a year combined so

41:35

if he works for ten years he makes what sundar makes in three there's

41:42

going to be a conversation it is i mean but this is what we've been we've been we've been advocating for

41:46

this so this is good yeah no no we're in support but google has

41:51

increased sundar's potential pay to six hundred ninety two million over the next three

41:55

years you know that tim cook dropped this in the apple board members group chat as soon as it hit he was just like dude this is a

41:59

cool article you should read this drops it in just drops it yeah exactly the bulk of his package comes in performance units with a target value of one hundred twenty six million split evenly into two branches the psus are valued by the parent company blah blah blah it could pay out as much as twice the target or a quarter million if it outperforms significantly or nothing if it lags behind gotta beat the s

42:06

and p one hundred

42:30

pichai will receive waymo stock with a target stock in waymo with a target value of one hundred thirty million and forty five million and one wing aviation that's their drone delivery platform platform again both can pay out up to two hundred percent of

42:32

the target they got a lot of incentive pay like if he delivers and the company does well he should be richly rewarded behrens was writing about the death of the mag seven mag seven's over up and down wall street the mag seven era ends the barron's is calling it they said stick a fork in it turn out the lights hasta la vista say it any way you'd like the simple truth is the mag seven trade is over finito i love barron's writing dead the collective stock market outperformance of those seven tech icons alphabet amazon dot com comma apple meta platforms nvidia microsoft and tesla is now a thing of the past the group may still do okay and some of the individual stocks may even kill it but the slam dunk set it and forget it run circles around the market era of the mag seven is gone with the wind recall that b of a securities analyst michael harnett coined the term mag seven in twenty twenty three pretty recently referring to the nineteen seventies john sturges western gunslinger flick an adaptation of akira kurosawa's nineteen fifty four film seven samurai have you seen seven samurai no have you seen the magnificent seven

42:51

the

44:12

stocks no the movie the movie the western no it's not in the seven samurai in what world production team has anyone seen either of those movies i

44:13

think we watched them in school okay okay which ones have you seen in a while you're saying you didn't watch them you studied them is that what you're saying haven't seen them in a

44:22

while which ones have you seen tyler seven samurai seven samurai it's a great one it's a very it's a very good movie anyway barron's is bearish on the mag seven but sounds like sounds like hope it does sound like hope i think the mag seven will do

44:29

it's like the large it's over for the largest and most profitable companies in history that stand to benefit yeah the

44:42

actual way the actual argument was the growth to value narrative the loss of cash flow as they increasingly invest in capex the financials will look very different there'll be a margin compression that type of thing it's not an unreasonable take but it is just funny the way it's written anyway sundar pichai took over as ceo in august of twenty fifteen he's going on eleven years in the seat google's market cap has increased almost sevenfold from half a trillion to three point six trillion briefly topping four trillion in january this serge has made the indian born fifty three year old former mckinsey consultant a billionaire he joined in two thousand four and made his name developing the chrome browser and leading the android division he had been criticized for being too slow to adopt ai at the search giant allowing openai to release the first hit product chatgpt in late twenty twenty two but has since bounced back releasing cutting edge ai models and integrating the technology into its dominant search engine yes he's done very very well on that front pichai has also navigated a duo of antitrust cases brought against google's search and app store businesses avoiding the worst case scenario of forced breakup a third lawsuit is pending against the advertising network tyler do you think that deepmind counts as rsi capable at this point recursive self improvement so there was a take on the timeline where basically someone was making the argument that xai is cooked because while anthropic is using claude to build claude and openai is using codecs to build codecs so both of those leading private ai labs are recursively self improving xai was using claude code to build grok and so they don't have the same flywheel of the tools that build the tools build the

44:50

the tools yeah i think what about access to cloud code but then they still have access to codex okay so you can it's kind of like do you count that as rsi because it's not like really internal yeah but is gemini that's the question yeah yeah i'm sure people at gemini are using gemini i mean there is a gemini cli people don't talk about it as much yeah yeah yeah i mean the model is very good right yeah yeah i assume it is helpful it would be

46:41

interesting i feel like we don't get as many little data points from the gemini and deepmind team because it's a public company and they need to maybe break out financials at some point i

47:03

think vague posting is probably more heavily

47:15

looked at exactly exactly because it's totally possible that they're doing the same thing demis has been extremely just very clear about the path this goes and he's very agi appealed he understands the exponentials and whatnot so he certainly messaged it at a high level without actually saying definitively yes we are at a point where we're not writing any code by hand that type of phrasing that you see in the rsi crowd anyway pichai last got a stock award in december of twenty twenty two worth two hundred eighteen million which was structured in the same way his earnings are topped up by his personal security costs which rose to eight point three million in twenty twenty four earlier in the week he sold thirty two thousand five hundred class c shares an average price of three hundred and three dollars worth roughly ten million dollars the bloomberg billionaires index estimates that he has sold about six hundred fifty million dollars in stock since becoming ceo they still own along with his wife he owns one point six seven million shares of google worth half a billion at the latest stock price google's founders sergey brin and larry page still control the company through their ownership of of super voting class b stock which gives them fifty six percent of decision making power interesting i think you gotta

47:18

give miami paper hands he sold six hundred fifty million of stock no no

48:40

no he sold that but he's oh and he only owns half a billion so yeah he sold more than that

48:46

i'm just saying hopefully he put it all into the shy

48:52

but he's getting topped up and he's diamond handings the new he's diamond handing i know i'm

48:59

just saying i'm just saying looking at the stock chart since twenty fifteen when he became ceo would have done pretty well just to not do anything and trust his own process hey hey maybe

49:02

he has amazing car collection or something there's a lot of he had to rotate into metal you can't drive a google class a share you can't sleep on a bed of class c shares jordy you got to get liquid no you got to give miami tech its due because everyone's saying oh miami tech they don't have any startups there's no funding there there's no miami tech isn't panning out well google is now counts as a miami tech company because the google founders bought houses there you also got to count meta as a miami tech company because mark zuckerberg has a house there so you put meta and google that's like six trillion dollars in market cap for tech companies all of miami i count it i count it

49:12

anyway moving owner of four loko is exploring a sale of the storied alcohol brand sources say wow let's dig into

49:57

this but first let me tell you about plaid plaid powers the apps you use to spend save borrow and invest securely connecting bank accounts to move money fight fraud and improve lending now with ai and let me also tell you about cognition they're the makers of devon the ai software engineer crush your backlog with your personal ai engineering team so

50:06

the parent company of four loko the canned alcoholic beverage that became a college campus sensation in the late two thousands before being reformulated under regulatory pressure is exploring i cannot believe they nerfed four loko

50:22

it was too powerful it was

50:37

too powerful like you can just imagine the trajectory of the united states if it hadn't been nerfed and then straight

50:39

downward really yes it was so bad it was so the i thought people

50:45

were having like heart attacks yes it

50:50

was terrible oh really so four loko was a twenty four ounce can so twice the size of a normal alcoholic

50:52

beverage trey the birthday boy says it

50:58

was amazing it was amazing okay now we have buzz balls that's true so four loko the original formulation was twice as big as a normal can of bud light or miller lite or coors light your normal beer something that you would grab so it was two drinks that way and then instead of it being three point two percent alcohol or four point five percent alcohol it was like ten percent alcohol it was the strength of wine almost and so a single four loko was like four or five beers in one can and then they also added like two hundred milligrams of caffeine so you would become incredibly intoxicated and inebriated but then also incredibly stimulated from from all the caffeine and that spelled doom for many people people would be very high functioning but completely inebriated and discombobulated and so they would get into all sorts of trouble whether or not it was behind a motor vehicle or the discombobulator it was the discombobulator it was also wildly illegal from an fda perspective you can't mix alcohol and caffeine in a single product that's just a rule that they chose not to follow and they figured out how to make it and then the fda eventually gave them a warning letter and they pulled the caffeine out and that of course killed the viral sensation that was four loko but apparently the company continued to sell products that were compliant with the current fda regulation nerf that's nerfed the nerf for loco continued to

51:00

somehow they're looking to improve potentially will capture somewhere around four hundred million on

52:37

the sale the brand value it's hilarious that they're working with jp morgan on this that's wild fusion products is the owner of the brand they're working with jp morgan on the sale process the potential sale underscores how ready to drink beverages have emerged as a growth category in an otherwise sluggish alcohol market us beer wine and spirit sales declined in twenty twenty five we've talked to a number of guests about the declining alcohol sales as folks move over to more

52:44

functional here's what the cmo had to say for over a decade our sales have been the leading flavored malted beverage have been leading the flavored malted beverage market by embracing bold innovation unconventional marketing and a risk taking attitude that delivers results year after year i guess they

53:16

kept building they also own the parent

53:35

company also owns mamitas pirate water pirate water bossico tequila and just earthquake earthquake

53:38

earthquake seems like something you definitely want

53:46

from the makers of four loko earthquake

53:48

pirate water is a wild name for a brand too i mean it's a distinctive can it looks remarkable it jumps off the shelves when the four loko nerf went into effect and they pulled them from the shelves because before they reformulated without the caffeine they had to pull them from the shelves entirely they were a hot commodity and folks would go around stockpiling them before they were banned and they were one of the most guarded resources in all of college

53:53

campuses wow earthquake is just a can that's just ten percent alcohol wow do

54:19

not recommend very risky never quake anyway in twenty ten the fda sent warning letters to caffeinated alcohol beverage makers including four loki saying that the caffeine in their products constituted an unsafe food additive so the fda has specific categories for regulation around what is an alcoholic beverage what is a caffeinated beverage and they never the two shall mix so they don't let you put these two things together of course vodka red bull was the precursor to this where folks would order a red bull and then they would put vodka in it i think but at a bar which is made at the time of sale and so that was legal i believe and this might be fake news but i remember hearing a story that at one point red bull was sued because a gentleman had purchased a series of vodka red bulls become incredibly inebriated gotten behind the wheel of the car of a car got into some sort of accident and wound up up in front of a judge and the argument was that he wanted red bull to be held liable for his actions while inebriated behind the motor vehicle and his argument was that it was only because of the product red bull had been mixed with the vodka that he made the decision that he was alert enough stimulated enough from the caffeine to get behind the car wheel in the first place and this might be apocryphal but the story goes that red bull made the argument that the caffeine actually made him more capable to drive not less capable to drive and actually increase the safety of the driver and the vehicle and so they were held not responsible but i don't know it's just like an old wives tale in cpg i don't know if it's true anyway let's move on to phantom cash fund your wallet without exchanges or middlemen and spend with the phantom card and let me also tell you about label box rl environments voice robotics evals and expert human data label box is the data factory behind the world's leading ai teams so we're all preparing for the singularity what is this doing dishes just like oh if you're if you're doing dishes you're not worried about dylan patel's viral post saying that being an sf is like being in wuhan right before the pandemic something is happening it's going to hit everywhere but so few people know it and that is the question of are we in a software only singularity where the dishwasher is life as normal and every time i go to the COVID analogy i start a clock because i remember tracking covid in january and i remember the viral post biology posted in february going viral and then in march the nba shut down and and in april everyone was locked down and so every time i see one of these it's like covid i'm like okay let's see what it's like in eight weeks like eight weeks is not that long in eight weeks if i go outside and things look radically different and you talk to random people and it's wildly different then we're like okay yes that analogy was correct but if it's more like in eight weeks we're like oh yeah technology is getting more exciting and tech companies are changing and the market's moving but you know the dishwasher is unaffected the pizza place on the corner is unaffected that's always how i i just think people underrate like how crazy a global plague was like it's literally like a once in a hundred like the most insane thing ever it's the craziest change to literally everything from the sports that you watch you couldn't go to the movie theater you couldn't go to dinner like everything was changed for like full year it was like complete upheaval in our society and might happen but eight weeks eight weeks is a little fast for

54:25

me and if you live in la probably still be able to go to the beach i think so so that's

58:14

but maybe not maybe everything's i did

58:20

have a funny moment yesterday i was telling you but i'll share it again i was talking with my car detailer who's become a dear friend and he was telling me that the people in my area cancel their appointments at a much higher rate than the people in the rest of la and he was talking about running analytics imessage and i was like oh how do you do how do you do that and he's like he's fully running openclaw so he's now using openclaw to run his car detailing business he's like yeah it's just always running in the background it's some to him he was like it's some combination of having like an ea or a ceo like helping me manage the calendar helping with comms helping me make smarter decisions on routing and scheduling and

58:23

all these things so when you book a detailing meaning that this gentleman is coming to your house and cleaning your car you send an imessage to him correct and that is what is so interesting about openclaw is that there were probably one thousand companies that were like we're vertical saas for auto detailing you gotta do everything over email and they just got eaten alive because people use imessage and imessage has been this closed ecosystem that private companies could never access before but it makes so much sense for him to say hey run through all my imessages and if i'm talking to my friends about the f one race just discard that but if i'm talking to a client about potentially booking a detailing put that in a spreadsheet organize that and then run some analytics and tell me what the patterns are and that's just something that should have been saas for a decade but apple has a walled garden that has now been smashed down by openclaw and i think it's really cool and i think that that's what we're seeing the first use cases for openclaw it's like things that could not happen for business reasons for logical reasons sometimes for legal reasons if you're talking about downloading illegal films like the napster analogy but a lot of times it's just apple didn't want to do a deal to integrate imessage

59:13

into salesforce the way to break down the walled garden was just to have

1:00:35

a digital guy basically yeah who just combs over all this so is that a new type of software that should have existed that now is unlocked is it disruptive like it's not like he it's not like he stopped using his previous crm probably didn't have anything so now he has something yeah he's pretty

1:00:39

tech forward he's tried every single saas tool that there is but like at the end of the day having like a digital guy that has access to all of your tools is just a much better experience and so i want

1:00:57

to know what his bill is this

1:01:10

month he said he spent a decent amount on i want to know what

1:01:12

it is this month and then i want to know what it is in like a year because sometimes the maintenance of these systems can be really really big and then there's a question about are we going to see per token deflation or are we going to see prices are prices being artificially held low because of competitive dynamics so in a year will he be spending more or less will he still be getting the value these are all interesting questions but i'm just excited that a digital product now exists that could not or didn't exist before and that feels incredibly positive and cool so very exciting anyway we have our first guest of the show in the restream waiting room let me tell you about console dot com console built ai agents that means seventy percent of it hr and finance support giving employees instant resolution for access requests and password resets and without further ado doctor alex wisner gross welcome to the show how are you doing what's happening doing

1:01:15

really well thank you john thank you

1:02:08

jordi thank you so much for taking the time to come chat with us congratulations on all the virality for those who haven't followed along so far can you introduce yourself and the project broadly and then i have a ton of questions but i'll let you kick it off with an intro absolutely so quick

1:02:09

background on myself originally from new york undergrad mit studied physics electrical engineering computer science math phd physics harvard was there

1:02:25

anything you did you studied everything i

1:02:33

didn't study the humanities okay considered going to cambridge to study chemistry and biology for a while just started it was told physics is a game for the young so stayed in the us did my phd in physics focusing on ai and nanotech i've started invested in advised managed dozens of companies at this point i've had a number of exits and at this point most of my time is focused on smoothing out the singularity as you guys were just discussing so most recently as you were gesturing a company that i helped found eon systems public benefit corporation just announced what we've been characterizing as the first multi behavior upload if you will more on that in a second of a fruit fly which we think is a major step forward for the field and quite frankly in an era when i do another podcast the moonshots podcast with my friend peter i talk all the time about tiling the earth with computer right now all of this ai infra buildout that we're doing to the tunes of trillions of dollars of capex all of this is going to artificial mines and e on is playing i think a very important role in doing some early pioneering experiments and developments to try to level the playing field of the future not just artificial minds playing on all of this trillions of dollars of infra and data centers but enabling emulations and if you will uploads of human minds non human animal minds at the moment starting with fruit flies and aspirationally working towards mice and humans and getting the brains of a variety of human and non human animals to operate in the cloud

1:02:36

so walk us through the fruit fly experiment why it was so exciting for

1:04:16

you it's really interesting if you think back remember back with the launch of chatgpt that was just a side experiment that wasn't the main focus of openai at the time some have characterized that as an unhobbling if you will gpt three the leadership of openai thought at the time was sort of the main entree if you will similarly within aeon i don't mind saying this was more of a side project that we thought it would be somewhat interesting to the world but i think the level of response and virality probably even took most of eon by surprise so this is a video demonstration of an internal project to take a bunch of building blocks that were already sitting around for the past one to two years and put them together for the first time so one of these was a paper that senior scientist at aeon philip chu published in nature in twenty twenty four demonstrating that just from an emulation perspective that you could take the flywire which was a well funded project to capture the connectome of the fruit fly brain then you could emulate certain circuits within the fruit fly blaine fruit fruit fly brain we've had that for a year and a half or so another project neuromec fly version two it's an excellent project to be able to formulate simulations in realistic mechanical environments of fruit flies another project from ursdal et al to formulate coordinated motor actuation of fruit flies and then ai so in summary eon was really the first to put all these building blocks together that were sort of lying around this is actually tiny subset of what eon's working on eon is working on scanning the connectomes of human and non human animals at scale and achieving whole brain emulation which is the holy grail of the field there have been many projects in the past that have attempted human whole brain emulation eon is leveraging advances in ai expansion microscopy a variety of other techniques but it was really just putting the pieces together so we're quite honestly astonished by the level of interest in just this one

1:04:20

demo i love it i have a million questions let's start with io like the input and outputs like what are you simulating when i pull up the sims that is somewhat of a simulation of a human a basic for loop over a bunch of random probabilities to say are you hungry if hungry eat go over to the stove cook food that's obviously nowhere near the level of detail at which a human operates is there an abstraction layer here where the output is quantized to things that you would think a fly would do like move a like move a wing or or are you operating at the level of like choose to fly or how at what level of abstraction is the fly's output happening yeah this is what

1:06:35

i was gesturing at a moment ago so there's a paper by ursdale et al that bundles motor actuations in terms of high level representations got it so a lot of people sort of the the people who got perhaps most excited in the neuroscience community about this video some of them i think didn't bother to read ursdale's excellent paper so a lot of this is why i'm saying a lot of the pieces were sort of lying on the ground waiting for us to pick up and put together so these like these abstract motor movements were already available just no one had ever prior to this bothered to wire

1:07:28

them up that's cool then what is the do you need a system prompt like do you need an initialization function or once you synthesize the connectome synthesize the connections in the brain things just start firing do you have to trigger something to start the process of consciousness or whatever you call whatever the fly

1:08:04

is doing no system prompt needed on the other hand in the early days of chatgpt before instruction tuning there was no system prompt either in the days of early language models no one had thought of prompt engineering so it's entirely possible that maybe in the future some sort of prompt engineering equivalent for whole brain emulation will be essential but at the moment there's no direct analog of

1:08:28

that

1:08:54

help me compare some of the scales that we're talking about on complexity of a fly complexity of a mouse complexity of a human i imagine that we're on some sort of exponential here but how many orders of magnitude in complexity size number of neurons something like that whatever the metric is how far away are we from humanity we're many

1:08:57

orders of magnitude away so the canonical estimate for the human brain is depending on how you count brain cells and there are multiple types of brain cells call it order of magnitude one hundred to two hundred billion cells in the brain we're many many orders of magnitude many orders of mag many orders of magnitude away from that so this isn't going to happen immediately we're not going to get human whole brain emulation immediately on the other hand i think it's important i think back to the early days of driverless cars and autonomous vehicles and what the world is missing right now is i think sort of a scale or a framework for starting to think about what that future looks like we're at the earliest stages of that call it level one or level two uploading a lot of people even get triggered by the term uploading they prefer something else and i think we need to start having these conversations in a critical thoughtful way and start to define what is a framework for multiple levels of fidelity for uploading or emulation or maybe we come up with a new term but right now we're starting that discussion and we're trying to do it in a thoughtful responsible way so you

1:09:21

like the term uploading because you did not you did not choose a random structure or an average structure of this particular fly's brain you chose the exact structure of a particular fly's brain this

1:10:27

is a very expensive structure of a very particular fly's brain this structure was again captured through the flywire project at great expense

1:10:42

so that's like a clone digital twin type of thing right aspirationally

1:10:54

i mean again we're at the earliest stages i don't want to oversell this is just a prototype and quite frankly this is a sliver of everything that eon is doing but the fact that eon was able to put a few of these pieces together build a prototype with relatively small effort suggests that we talk in the ai community all the time about overhangs many people thought that the arrival of large language models was in some sense a technological overhang that we could have one could imagine a thought experiment could we have had large language models twenty years ago if we simply knew what we should have been building i think probably i thought we

1:10:59

were very gpu gated on that well

1:11:36

the gpu's were being spent on video games for the first few decades so we could have in principle like markov models we knew how to build markov babblers i was building markov babblers in the nineties and early two thousands yeah

1:11:39

it feels like even if you try and port back frontier models now to a ten eighty ti which was a graphics card for gaming you're going to

1:11:52

have a bad time well i mean with the rise of slms and some of the amazing deflationary properties of algorithmic progress in slms i think we will find ourselves in the world next two years where some of the earliest pcs could probably have hosted non trivial conversations with llms there's just that much algorithmic

1:12:03

progress extremely bullish for apple anyway not to put in in hardware terms what

1:12:26

do you want to do what do you want to do next like where do you where do you go from here i know it's a side more

1:12:32

flies or straight to mouse it's such

1:12:38

a profoundly interesting question so eon is beginning fundraising eon wants to tackle both mice and men and i think the timescale will be determined by both technical capability increases and also fundraising abilities so i think it's an interesting time i like to say on my other pod that the singularity looks like all sci fi tropes happening everywhere all at once and one of these sci fi tropes is most definitely mind uploading and i'm doing my best vis a vis aeon systems to push that part a little bit to the left so that the ai's don't have the dyson swarm all

1:12:42

to themselves interesting architecturally what does the neural network look like that you've built for the fly is it similar to a large language model like transformer based architecture or are you a beneficiary is

1:13:24

all you need totally unlike the transformer totally on totally unlike the transformer so transformer architecture if you subtract off the encoder and decoder layers looks comes in many variants but the most vanilla variant looks like alternating linear and attention layers so totally unlike the transformer architecture at least if you look at it it looks a little bit more like a graph neural network but really that's such an ai way of framing it it looks like a leaky integrate and fire lif model that we've had for decades from the neuroscience world so it's just a graph of nodes that have leaky and terrain fire dynamics and they're firing

1:13:41

at each other so what does that tell you about the nature of synthetic human intelligence like do you think that there is a path to human level agi that does not involve the transformer and instead falls along your path and your architecture oh sure well first of

1:14:21

all i would argue we've had agi for at least six years now five to six years at the very latest since summer of twenty twenty when openai published their language models or few shot learners paper yeah gpt three i would argue that was an agi so i think there are many ways my bet is there are many ways to achieve generality of intelligence i published a paper a number of years ago arguing that not only is intelligent behavior a general process i went further and argued that it is a general physical process that you can even formulate intelligent behavior in pure thermodynamic terms so my bet is intelligence is this very very general effect and lots of ways to implement it

1:14:42

interesting i mean you've mentioned agi six years ago singularity here and you're smoothing it do you have other sort of binary benchmarks that you're looking towards like asi recursive self improvement are any of these terms useful to you oh we're

1:15:24

already there i mean we're already in the era of recursive self improvement all of the frontier labs are pretty public about it at this point i'm looking past the singularity so i spend most of my time on bets for what the post singularity world looks like so eon is one of those bets another bet i have a company physical superintelligence that's trying to solve all of physics with ai and doing an amazing job i just wrote a book with peter called solve everything arguing that entire disciplines are going to get steamrolled by superintelligence and that what matters now is what disciplines we aim the superintelligence at interesting

1:15:44

what decisions do you make in your personal life in the way that you live that are based around your beliefs around technological progress there are a few

1:16:20

that come to mind so one is i'm hoping not to die it would be a shame if i died sometime soon and get to miss the most exciting developments

1:16:34

same that's what gets the applause that's great yeah that's a white

1:16:45

pill white pill very good brian johnson thanks me in advance but i think also i mean there are so many other angles another is not trying to bet against the collective intelligence of the market i've had a number of friends who thought they could outsmart all of the ai algo traders and day trade and when they inevitably fail they attribute it to to bad luck or something other than just fundamentally betting against progress and betting against ai intelligence so trying not to make the mistake of betting against collective intelligence of civilization and ai capabilities i also i'm trying to make bets assuming that the ai capabilities keep increasing to the point where they're ontologically shocking and trying not to duplicate effort of frontier labs and i think that's perhaps something i don't see enough of in the venture community so many new startups being formed now just aren't being ambitious enough i have startups that newly formed or recently formed that are literally trying to grow new islands and new coastlines with ai that would have been unthinkable years ago i have that's with

1:16:49

physical ai or they're just dreaming them

1:18:07

up it's with ai reaching into the physical world to steer ocean currents and grow new islands and new coastlines that would not have been possible a few years ago that's now starting to become possible i have a startup that's working on solving interspecies communication starting with dogs love that that would not have been possible a few years ago so i spend a lot of my time thinking about what the post singularity state of the world looks like and how to smooth that out and bring it here

1:18:11

sooner yeah what does the post singularity world look like where you are able to upload yourself i assume that feels like there's a copy of alex in the cloud and then you still exist and not in the cloud how do you think that that interfaces in my

1:18:37

mind if we wind up having the copy question come about that's almost a failure mode speaking for myself i don't want a copy of myself i want myself yeah and hans moravec and others have written about this what i would like to see in the space and what eon is working toward at least aspirationally is this idea of a continuous transfer of consciousness so that it really is you it's not a copy of you it's not a low fidelity facsimile of you it should be a better expanded version of you that's still you and whether that looks like moving from what eon's doing right now with fruit flies to maybe replacing a single cell in a brain at a time invasively or non invasively with a substrate independent or substrate migrated implementation some variant of that i suspect is where all of this will go so that you never have to worry is it really myself or not it will by construction be yourself and i'm not sure of the precise timetable but i think five ten years from now i think the world will see marked progress on the problem of not just whole brain emulation but also transfer of human intelligence to new

1:18:55

substrates that's very exciting this is a

1:20:07

new sci fi corner i love it

1:20:12

every sci fi trope everywhere all at

1:20:15

once time travel that's a good framing

1:20:17

is time travel on the table we

1:20:19

don't know yet so it's an interesting question we don't know whether the physics of our universe are compatible with i think what most folks would construe as time travel there are versions of temporal nonlocality that are consistent with the physics that we have right now but not in a useful way that would be worth filming a sci fi movie over

1:20:21

what about faster than light travel we

1:20:40

don't know yet same problem in some sense exactly the same problem as time travel we don't know whether the physics of our universe will ultimately allow that

1:20:44

because it seems like if you don't get ftl like there's a it's going to be a boring travel to go to another solar system i will say

1:20:52

this i think odds are pretty good that ai in the next few years will tell us whether the laws of physics of our universe are compatible with that or not and help us solve this one of the reasons why i helped found physical superintelligence to discover any

1:21:01

of such laws very cool well thank you so much for taking the time to come chat with us yeah great to meet you congratulations on the progress

1:21:14

fascinating stuff thank you luckwest we'll talk

1:21:20

to you soon cheers have a good one let me tell you about figma no matter where your idea starts figma make claude code codecs or a sketch the figma canvas is where ideas connect and products take shape build in the right direction with figma and let me also tell you about cisco critical infrastructure for the ai era unlocks seamless real time experiences and new value with cisco without further ado we have charles lamanna from microsoft in the restream ready room now he's in the tv interim charles

1:21:22

how are you doing good thanks for

1:21:54

having me on guys thanks so much for coming on the show why don't you give us a little bit of an introduction on yourself and then we can get into the news sure thing

1:21:56

i'm on the product engineering side at microsoft my team builds out things like agents a bunch of the platform work in our business applications and some low code stuff so a bunch of odds

1:22:05

and ends what's the latest and greatest in your world yes a lot of

1:22:16

big news today so today we announced copilot cowork which really takes copilot beyond chat so it can take actions and automate tasks and take delegation basically we had some new agenta capabilities in the office apps and a bunch of agent

1:22:21

management solutions yeah how are you thinking about the open claw boom people writing code locally versus microsoft has azure it's so easy to throw everything in the cloud and i feel like one of the barriers that a lot of people have is like do i need to go get a mac mini can i just delegate to the cloud some of the apps don't just by default build an app in the cloud or they don't have functionality how are you feeling about that trade off the app days

1:22:34

yeah i mean we're super excited by openclaw and all that means i think it's like the resurgence of the pc like what's the best thing you can give an agent a computer let it run the apps let it log in as you let it messages you i think like a windows pc in the cloud is probably the main way people will ultimately use something like openclaw at

1:23:04

work yeah how do you think about model selection obviously microsoft has a time honored partnership with openai at this point but you know microsoft has their own ai lab and you can get pretty much any model on azure do customers want to pick models how do you think about what's the best tool for the job in the products that you're building but then also how much do you want to give to the user because there's sometimes when i go into a chat app and i'm like i just want auto you should be figuring this out i know the details of pro versus thinking but i don't want to think at all yeah so that's

1:23:23

one of the things that we think is super important for copilot and a lot of our products the fact that it's multimodal and the reason is because the best model for a given kind of task is not always the same and it's not even the same from month to month or quarter to quarter like if i look at microsoft three hundred sixty five copilot chat is like the main way people interact with it our auto router is going to put you in openai models because that's what we think works best with our work grounding in our work context but then if you go to cowork the new announcement today and you fire off a task and you want it to run for a long period of time in the background use a bunch of tools maybe generate some code use like the terminal and stuff that uses the clot family of models and i go to copilot one place and get both and then we're going to keep adding more models models like from mai and other places where it makes sense but it's really about the experience the user wants not so so much picking exactly the

1:24:02

version or the type yeah how have conversations been with business leaders in terms of deploying a product like copilot cowork because the ramp up for software engineers was smooth and everyone was aware of it tracking everything and going from better autocomplete to agentic coding was pretty intelligible at every step of the way for most software developers but for someone who's lived their life in spreadsheets and email and maybe not really thought about building automations all that often what does actual deployment of these tools look like yeah

1:24:53

it's such a great question because i think there's like two pieces to it the first is you have it teams which are nervous with if you imagine every single employee spinning up a whole bunch of agents that can create documents send emails run workflows so they want to make sure they have the right governance and controls in place that's why another big part of the announcement today was something we call agent three hundred sixty five which makes it easy for it teams to manage and deploy agents from microsoft and other companies so that's one part and then you have the end users how do you teach a typical excel or powerpoint or outlook user this idea of agentic working even encoders it's taken time and it's probably not fully diffuse so that's the thing that we think we're in a really unique position at microsoft in terms of being able to do because we can bring things like cowork to users where they already spend their time inside of the office apps inside of windows and increasingly inside the copilot app and we can present it in a way where it's digestible to hundreds of millions of office users not just tens of millions of

1:25:34

coders is the majority of your business model still seat based at this point because i imagine that there will be a transition at some point how much of that transition is already happening yeah

1:26:38

we're still very seat based on the m three hundred sixty five side but the good news at microsoft we kind of know how to do consumption businesses because we got azure we got a bunch of other very large capacity businesses so the fact that we have user licensing and then consumption and pay as you go licensing for capacity all integrated with one commercial kind of vehicle with customers we think we're in a pretty good position to make that transition over the next couple of years there was

1:26:50

some there was some chatter on x this weekend over how different labs are pricing tokens and plans potentially to take market share from each other there's a little bit of a capital fight playing out how are you communicating more specifically competing with the application layer yes my question is how are you talking to business leaders ctos about how they should think about forecasting token budgets and spend and compute budgets if they're used to okay i have licenses and licenses maybe go up in price and my head count goes up and i'm used to that but all of a sudden it's like well everyone's paying dollar two hundred a month but they might be five thousand dollars a month in a couple months but also the models are getting cheaper and the inference is getting cheaper and the capabilities are going up and it's very hard to predict what are you talking to what are you telling to business leaders and ctos yeah so

1:27:15

the first thing is you have to embrace this technology otherwise you're going to be left behind right now because the productivity gains are just so significant and if you go talk to a coder and you told them hey i'm going to take away github copilot and the agenta coding capability capabilities they'd be like i refuse to work in this environment like it's just it's just inhumane almost so i think the same type of thing is going to happen for all information work all office work so nine months from now i think if you went to somebody and said we're going to take away your agentic tools like copilot cowork they'd be like no way i'm not going to go back to the old way of working so i think the first is there's a degree of inevitability because the benefit is so large and there's such strong pull from the end users that said everybody has budgets so what you are going to see and what you're already starting to see is a bunch of controls where it teams can start to manage exactly how many tokens or how much dollars each user can spend and they're going to allocate that in a way based on where they think they're getting return and diminishing return and it'll probably be like expense report approval hierarchy type thing so you'll see all of that show up but part of just doing work now is you need a token budget that's the future of how we think about how you budget for a team or a department

1:28:11

do you have any exciting case studies on the non software engineering agentic work side of the business because when agentic coding came to software development there was a shift in a lot of software developers workflows where they went from writing a lot of code to reading a lot of code reviewing prs they started operating at higher and higher levels of abstraction but the decision what to build how to prioritize things how to build features there's still conversations and software engineers are still doing work what does that look like in the context of a finance department or an operations department or a sales group like where when the rubber meets the road like where are you seeing like okay this is ready this is the next sub area or category that will really really benefit from the productivity gains from agentic workflows i mean a general thing

1:29:33

and then maybe a more specific thing what we're seeing is how do you measure productivity is itself a challenge i mean this predates ai right like lines of code number of pull requests number of bugs fixed i don't think i've ever heard a good engineering leader say that's how i measure the success of my team right it shows up in other ways do you have a great product people love is your cost structure good what's your install base look like so i think we have to get better at capturing that final business outcome because that's what's going to get the cfo to open up the pocketbook and continue to invest in more tokens and more ai so if i go then take that and apply to kind of other disciplines like what you don't want to say is oh our salespeople send a lot more emails or our finance professionals create one hundred more spreadsheets like that's more like of a nightmare scenario

1:30:32

for ai one billion spreadsheets per company

1:31:23

no no i mean yeah we saw someone on the timeline say that their token budget went way up but their revenue didn't go up and so they're like i don't know if it's if it's worth it like you got to

1:31:27

show results exactly so that is what the question is going to be everywhere but the good news is engineers historically have resisted clear measurement of impact because you know there's a little bit of art to it but like sales professionals or finance professionals or customer service and support reps they are so metric driven like if i'm a sales rep what's my quota what's my attainment what's my revenue per head every single sales department has all of those things instrumented and we've seen with m three hundred sixty five copilot we kind of did a bunch of experiments and we published a white paper on it like an almost ten percent increase in revenue per seller just by giving them the family of m three hundred sixty five copilot tools and things like our sales extension that's real dollars in the bank same thing for like back office finance processes we've seen people use like our agents to do like accounts reconciliation and we have like a great case study we did there where it's like thirty percent of all back office finance was automated out by this thing so those are examples of real top line and bottom line benefit and that's what everybody's going to be talking about the next couple of years it's not going to be emails

1:31:37

documents files et cetera yeah i mean you're in a high level business role yourself i find myself doing a lot of knowledge retrieval deep research reports but have you personally been using agentic this product specifically in your day to day even as something as simple as sam altman told us that he has a self he has some sort of to do list that does itself or something i still don't understand exactly what that product is but how are you using these tools these days yeah so the

1:32:49

for for like copilot cowork for example i've loved that like my favorite magic moment has been around calendar management i was telling the story to someone like i could go look out three months in the future and try to see you know how crazy is my calendar and travel look like interesting and i was able to decline seventeen meetings for

1:33:20

me okay that's right there nothing gives me more joy than declining a meeting you sold me yes asl and i'd

1:33:39

say like i went to outlook and just saw all the meetings disappear and it was like such a serene moment for me and like it did a deep analysis it went and said like who's my manager who do i work with am i optional or required what's the topic and it recommended the seventeen and i went through all seventeen said you know what decline and it was right i mean so like that's one thing that i use it for for like another thing that i like right now is i mean i'm an engineering background so i'm not going to go create the most beautiful content all the time so i use it now to pull together information design docs meeting notes and then go create either like architecture diagrams presentations or spreadsheets or docs and i mean i know it's just more doc generation but it's very much more kind of like fire and forget so

1:33:47

it works really well yeah yeah i mean even if it's just like the bones of a powerpoint presentation with the actual data pre populated that would just previously be a lot of copy paste go run this analysis copy it over make sure it's the same font there's so much grunt work that can be automated with these tools i'm super excited for that where do you see the design powerpoint generation tool use i mean people have been talking about it's hard to create an rl environment with a verifiable reward around a great pdf but it's easy to do that with coding do you do you have like an internal data set of like the best powerpoints ever created that you're like rl against or where do you see that

1:34:30

going yeah i mean i think that is something that there's a little bit of taste in all of this because like depending on the company some companies like lots of infographics and lots of colors yep other companies like super fact based no images so there's a little bit of an eye of the beholder but absolutely we have a very robust harness and gym environment where we're able to quickly evaluate models prompts tools and our own fine tuning efforts to see if it's producing better content so i think this powerpoints word documents spreadsheets it's going to be like code and you're starting to see it already but it's going to be like code for sure

1:35:14

yeah i mean a lot of these things are like xml under the hood anyway and can be puppeteered like code anyway well congratulations on the launch jordyn

1:35:54

i hope you don't mind i've been calling copilot cowork coco coco hopefully it

1:36:05

doesn't matter unless you like it yeah

1:36:12

it's a very cute name but congrats congrats on the launch and great to

1:36:14

meet you thank you so much for taking the time to come on the show cheers we'll talk to you soon let me tell you about mongodb what's the only thing faster than the ai market your business on mongodb don't just build ai own the data platform that powers it and let me also tell you about graphite code review for the age of ai graphite helps teams on github ship higher quality software faster there's an interesting data point from apollo featured by restructuring here ai is making starting a business easier and the numbers confirm it see this weekly business formation is exploding higher so previously this matches what

1:36:18

the collison brothers said when they did the show yeah they're seeing not just more business formation but businesses growing faster

1:36:56

is this it could be that ai is actually making businesses harder to start but but stripe atlas is is making it so much easier that we're seeing an upward trend and it's fighting back against ai now of course weinstein is just these are multiplicative very clearly but stripe atlas does make it very easy to start a business us business applications jumped insanely after the COVID pandemic in twenty nineteen we were seeing about sixty thousand seventy thousand new business applications i think think a four week moving average we've been over one hundred twenty k for years now and ai seems to be driving that trend upwards by about twenty percent over the last two years which is great news more small businesses they are extremely edifying extremely valuable i

1:37:03

highly recommend we got another white pill give it to me the data center buildout has allowed loudoun county to lower property taxes by thirty eight percent since twenty ten translating to average savings of about three thousand four hundred dollars per year for homeowners this is great love

1:37:53

it what's happening here data center square footage in the millions has gone from five million square feet to forty million square feet that's a huge jump yes

1:38:11

this is like northern virginia this is like aws east one okay yeah so

1:38:22

homeowners are paying they went from one point two percent of whatever their house is so if they have a million dollar house they will be paying twelve thousand dollars a year and now they're at zero point eight percent property tax rate because they're taxing those data centers this is the easiest solution to every ai problem if the companies are profitable tax them we already tax every business we tax them by default if somebody makes a lot of money on an ai stock taxes as capital gains you get the money the money shall flow and so good news for the folks

1:38:27

over in ludon county jacob says america once named things plainly place trade surname you had business international business machines long distance telephone studebaker studebaker and bethlehem steel

1:39:03

steel that international business machines logo is goes amazing is this real or is this like a reimagining of what the ibm logo could have been because i love this logo what a very cool

1:39:19

thing that is apparently the original ibm

1:39:35

original ibm logo wow we don't know how to make logos like that in this country anymore seriously i mean that's what kicked all this off somebody was

1:39:39

upset about the granola reback america's innovation slowdown is primarily due to the amount of lord of the rings that founders have to flip through before identifying a name they can raise off of no

1:39:47

you can just start a company and name it international business machines whatever you want to do just name the company exactly that technology and business programming network it's that easy in other news two

1:39:57

chainz two chainz revealed he was an early investor in spacex this is huge he says it just felt like i got in like early bitcoin there are

1:40:07

some crazy early spacex investors i don't know what elon was doing with that early roadshow we know some of these folks very cool community get me two chains get me two chains two chains got in at some point and gabriel over at openai says two chains is about to be three chains well let me tell you about okta okta helps you assign every ai agent a trusted identity so you get the power of ai without the risk secure every agent secure any agent let me also tell you about the new york stock exchange hopefully we'll be seeing spacex there soon want to change the world raise capital at the new york stock exchange can you imagine two chains and elon ringing the bell at the new york stock exchange i can i hope it happens well we have our next guest in the lease dream raider and julian back from sequoia capital his partner there welcome to the show julian good to meet

1:40:16

you hey guys how you doing nice to see you we also hope we can ring the bell very soon holy

1:41:09

holy logomog yeah holy logo those companies

1:41:15

i do we're partnered with cisco thank you for backing them early making tvpn possible through your thirty years ago so this paved the way since it's the first time in the show why don't you give us just a brief brief back story on yourself how you wound up at sequoia what you focus on what your day to day is like

1:41:19

sure so first thank you both for having me it's an honor to be there i know you had a lot of my partners as well so we're very excited to be on the show my name is julian i'm a partner on the team i'm based in london but it's been a lot of time going back and forth to spend time with the rest of the team in california and i started as a founder a very very long time ago in high school built companies and hardware before and so that's why i've been investing mostly in software for most of my

1:41:36

career wow you were like i got to get out i was like man

1:42:08

it's hard are you thinking about going back to hardware is software dead what

1:42:12

are you thinking yeah well i mean two things can be true at once i think hardware is really interesting i'm spending a lot of time thinking about physical ai sure but at the same time i've been investing for the last ten years in application software so i want to remain true to that and this is kind of the things i've been spending time on recently yeah take

1:42:18

me through your current thesis on how software is changing in the age of ai in the age of maybe some sort of takeoff agi is here according to sequoia capital amongst many other people the labs are recursively self improving and it seems like you know sam altman made the claim that like if you want to build a startup you shouldn't build a company that is in direct in the direct path while more and more products seem to be in the path of ai and agi how are you thinking about the changing dynamic with software as we know it

1:42:41

yeah that's a really important question and i think we've been listening to sam a lot as shareholders and recently just talking to our founders a lot of them are wondering if they're just an iteration away from the models replacing what they're doing yeah and it puts us in the position of thinking what is coming next what is going to be defensible yeah and for us we've been making that prediction last week that the next trillion dollar company will be a software business that masquerades as a services firm and we shared that thesis with everyone online the reality is if you sell tools today you're really in the line of sight for the models and you're effectively competing with the next generation that they're going to launch whereas if you sell the work you're actually benefiting from what the models are doing and all the billions of dollars that are going towards ai so part of that prediction was saying for every dollar that's spent on software six dollars are spent on services until now we could only really go after the one dollars because we were building tools now that we can actually deliver outcomes companies should think about capturing those dollar six instead so i mean

1:43:21

the obvious first step is just get to outcome based pricing move away from a seat based model probably would you agree with that as sort of like the most immediate change in business structure that needs to happen or if you're starting a new company maybe don't start with seat based pricing on day one or is the like seat based pricing is dead maybe that's like an overblown characterization how do you think about just like the business model moving forward

1:44:46

yeah we think both can work they just serve different purposes we have companies in our portfolio that start as these co pilots that are helping the workers and then we call them autopilot the one who do everything and to end many of them are transitioning from copilot to autopilot to do what you're describing a company that comes to mind is a business called sierra that was started by brett taylo that basically helps companies with customer support which is pretty direct bullseye use case of ai and in this example they're actually charging the customers per outcome per resolved ticket and so if you spend dollar twenty per ticket resolved with humans they charge you say dollar five instead right and that's driving roi for the company it's driving an outcome that's very measurable and so that's the first more obvious company that's really working doing that frankly until recently outside of software engineering we haven't seen a lot of this sort of fully own end to end outcome driving but the models are really progressing quickly and for everyone who's been by coding the last couple of months i think we've all been impressed that we can code a full website just end to end without any skills frankly basically our prediction is that this is going to happen to other categories and we've been mapping out these categories and helping founders come up with frameworks to prioritize where to attack them

1:45:15

and how to do it yeah sorry to interrupt one thing i'd be curious to get your take on so if a company is just selling enterprise software that helps people accomplish a task that feels very threatened by a foundation model company that maybe makes an agent that can also do said task and then you don't even need the software so if a company transitions to just selling the work selling the service like you're saying you would agree that they're still going to be under threat from the labs who will also presumably be selling intelligence or agents that can go and do the work but you're saying it will maybe be more closer to a fair fight is that the right way to think of it because i still think let's say if somebody is like spinning up a instead of making a legal ai tool they make a ai native law firm and they're selling the work i could still go to a chatgpt a claude a gemini and a future version of the model i can just say like hey i need to do this thing can you help me do it so there's still going to be that competition but you're saying that's a much better sort of competitive environment to be in than just selling like the software that enables somebody to do the task is that roughly right or is there something i'm missing yeah i'll

1:46:52

just add one framework that i think is helpful for people to clarify what that means basically everyone is conflating the model's intelligence with what humans are really good at which is judgment and i think if you strip those two concepts that you really understand what the future will look like and so right now the models are becoming really good at everything intelligence related so think of tasks that are rules based where just logic gets you to the goal whereas the judgment is everything the humans really spike at which is more and harder to define its instinct people call it taste call it experience and those things are things we think the humans will remain really good at for a very long time until intelligence is able to absorb these skills into it but it's just really hard right because we give the models a lot of credit for their intelligence but they're just trained on set amounts of data and so we actually think that this hybrid motion of coupling the intelligence from the model with the judgment from the humans is going to help you build the next generation of companies which we call these autopilots right to your example in the law firm you will automate a lot of the intelligence work that's very much codified which is what the model will do if you put the employee at the center someone who's really experienced who can talk to a jury who can you know mark a pause when at the right time those are all the things that you learn only from real world experience that a top lawyer will be able to experiment but that the models might will emit today does that make sense

1:48:14

yeah i still i still like you know maybe i'm too agi pilled but i but i just assume that a sufficiently advanced agent would be able to to capture and accumulate the ability to

1:50:01

agi pills just talk to doctor alex wisner gross for thirty minutes and he's like i'm uploading my brain right now

1:50:15

no it's not that but it's just the idea of like you put a smart person into an organization where they don't have any experience and they start doing things and they start getting feedback from other humans in the organization and over over time they accumulate enough knowledge and domain expertise that they're able to excel at something that even using the customer service example you've always been able to outsource customer service and you can go work with a firm that is excellent at customer service the alternative has always been to effectively hire a really great leader for that part of the organization have them build out a team and a structure and an approach and execute something that would be a similar result there's like a cost trade off

1:50:24

but it does feel like overall there's maybe some pressure on the b two b two b model or the b two b two c model if instead of to use your example of customer support there was a time when software companies would sell phone tree software not to companies directly to do customer support but to customer support outsourcing agencies that would then deploy that and now you're seeing more companies go directly to the firm that makes the customer support experience end to end go directly to the company that needs customer support as opposed to having this like middle level and it almost makes me feel like we're maybe going to rotate even a little bit further into d to c when i think about law there's a lot of stuff where the software would go to the lawyer the law firm goes to the person and if the person can go straight to the law firm or the legal model you wind up with a little bit more b to c activity a little bit more company directly to the service provider or the technology provider but again this feels a little bit dependent on specific market dynamics i'm interested to know your reaction to that but also what does hold in your mind in the near term agi scenario because there's been a lot of debate over how will marketplaces perform form or are network effects still super sticky in a world of agi or what what are what what are the powers the seven powers yeah i guess i

1:51:15

guess part of what kind of closed out my thoughts like i i agree i agree with the entire thesis and like the relative short term like the one one to two years but then beyond that there's still some uncertainty around around like i don't think it i don't think it's solves the sort of discount that even traditional saas companies are getting

1:52:57

on that point i think you're right then basically what we realize is no one really knows right i've been in many partner meetings recently where everyone just is astonished at how quickly things are moving i've been doing that for ten years and it's just by far the fastest rate of change i've noticed but some of my partners you know alfred lynn pat grady they've been doing that for a lot longer and they also acknowledge that everything is moving so quickly and so what we're trying to figure out is also not what is going to change what are the things that will remain true no matter what right customers will always want things faster better cheaper right and so that's kind of the thesis that we're describing there are just categories that require more intelligence than judgment and so we think that in the immediate the ones that require more intelligence will be actually really interesting for these autopilots because you can have you know instead of having ten humans in one ai you have one human and ten a's right and just that that ratio is just going to shift as the the models get better and are able to absorb what we used to call judgment from the humans that are that is turning to intelligence and just to put like a practical framework to that you mentioned outsourcing it's actually the best wedge right now is just to focus on outsourcing if you want to capture this services spend because there's three reasons the first one is that outsourcing has already been agreed to a third party so it's very easy to sell right because it's already a vendor that's external to the company the second is that there's an existing budget that's assigned to it which you can peg yourself against and the third is that it's actually very straightforward to swap a supplier it's not so straightforward to reorganize your company right we all talk about headcount reduction you know in terms of roi but actually putting that in practice takes time whereas swapping a vendor is very quick so i actually think that that's the prediction is that the companies that will succeed very well with this autopilot model will start with where there's a lot of labor spend that's being outsourced today we mentioned customer support we'll mention also things like accounting tax insurance and so on and so forth

1:53:23

is

1:55:48

there a world where in the next five years or something like venture capitalists truly shift away from software and go back into rare metal mining or something crazy like that i just keep coming back to daniel gross now at meta venture capitalist i think one of his first agi trades the first bet was like in the world of agi is copper mispriced and it was it doubled in price over two years and i'm just reflecting on this idea that if the job of venture capitalist is to back software companies that's one thing but if it's just to allocate capital in a way that can have exponential upside a single investment return of fund would

1:55:48

it ever make sense to stop investing in software i would hate that because we obviously love software software but it has been funny a lot of people have been been pushing like you know the models are so good we just need better harnesses yeah and to me this is giving me flashbacks to early gpt days where people were like gpt is really powerful you just need the right harness to get all these like wrappers that had explosive growth very momentarily until and so i do most agi

1:56:34

pilled person ever over here i love it but what do you think about the more physical aspects of the job rare metal mining for example does that seem like something people will be working

1:57:08

on a couple of years yeah i

1:57:24

mean everyone's talking about the fact that energy will be the biggest bottleneck right to train these models so it's very natural that you would go there personally that's not really where my skill set is so i really hope you're wrong in that we'll find other areas in

1:57:27

between i think the real path will be somewhere in the middle hopefully right

1:57:43

my way of doing things has always been to focus on talent and so right now i'm just always asking the smartest people i know who do you know who's the smartest person what are they working on and you're right more and more people are going more into the physical world but right now that probably means things like robotics and things like that people are making a lot of money trading commodities but that's a

1:57:50

different world yeah who knows maybe that's the job of the venture capitalist in twenty thirty just trading copper all day long you mentioned using ai to build a website despite maybe skills that have languished since your time in software or hardware but how are you using ai tools these days are you still doing mostly like knowledge retrieval market research for your job in venture capital or have you actually built tools that have accelerated certain things have you automated any of your workflows i'm just kind of curious how people are using ai these days

1:58:15

yeah well i don't know i hope my competitors are not listening

1:58:52

they are so you can say i haven't touched it actually it's useless yeah exactly you

1:58:57

should uninstall organic organic capital pen and

1:59:04

paper actually is underrated exactly back to

1:59:07

the basics pen and paper and really don't try cold code it's really really dry now jokes aside it's been really a transformation and particularly the first quarter of this year i mean i use it for all kinds the work is basically sourcing picking winning company building and harvesting right then i really have workflows that fit all these boxes so it goes from things as you know at the end of the week it will just like scan my entire calendar look at all the meetings i took look at the granola notes which is a transcript do an analysis and help me you know try and find the key questions and the due diligence and you know it'll help me just prioritize right because this job is just so much about prioritizing it's a power law right there's only three meetings in the year that will really that could change your the course of your career in so making sure that you're attuned in those meetings is really important so so that's really what i would say claude has helped me do right now and i'm sure you know openai will release something next month that will be amazing and so on you just have to experiment right and i think that's also like what we've seen with things like open cloud being so top of mind is like this culture of experimentation is not just in software engineering now but it's transpired across the entire knowledge workforce and venture capitalists are not an exception i

1:59:10

love it well thank you so much for taking the time to come shout

2:00:40

with us great to meet you have

2:00:42

a great rest of your day thanks

2:00:43

guys we'll talk to you some more

2:00:44

goodbye great let me tell you about eleven labs build intelligent real time conversational agents reimagine human technology interaction with eleven labs you know i if i was a venture capitalist open claw every morning at five am text me the name and market cap of the company that i passed on three years ago i need inspiration to grind harder that's the secret to success as is lambda lambda is the super intelligence cloud building ai supercomputers for training and inference that scale from one gpu to hundreds of thousands and without further ado we have owen mccabe the ceo a huge supporter of tvpn as well welcome to the show

2:00:46

how are you doing great to see

2:01:23

you again great to see you both

2:01:24

also it's been too long man you have you just have a voice for radio live streams podcasts all the it's incredibly calming i love it i love

2:01:26

it okay what would you like me to say i can record a meditation for us we could kind of start up the hour that way do you

2:01:37

have time do you have asmr funding announcement do you have time to meditate

2:01:44

these days it's kind of the only thing that keeps me sane that's good

2:01:47

well you have some new funding for new project break us down what happened what's the plan how much did you raise take us through it yeah so

2:01:52

we raised two hundred fifty million dollars

2:02:00

let me hit the gong congratulations

2:02:02

did that gong get bigger since i was

2:02:08

here it did get big it did

2:02:10

get bigger it will be cool if it got bigger every every single season

2:02:12

right fast takeoff just wait there's a

2:02:16

fast takeoff in gong size over here

2:02:19

it's growing exponentially except this was actually i think the largest we could find in the us so we got to

2:02:21

go yeah no there's always bigger i kind of have gong envy right now i wish i had a gun we'll send one over let's talk about that

2:02:26

later we'll talk we'll talk so two

2:02:32

hundred fifty million dollars what's the plan

2:02:34

yeah yeah so i mean we've been building this thing called fin for three years now service was just this obvious place where ai was going to blow up and all of a sudden everyone else started to chase it we were about a year ahead of everyone else and we started doing customer queries and then more complex problems originally it was answering questions for customers and it's fantastic the typical response rate with a human was one to two days you'd have to deal with the beautifully perfections of humans unfortunately these bots are just far more perfect than humans and they answer in seconds and of course they're cheaper too so we started to say better faster cheaper these things are really better faster cheaper so that was a tremendous success fin is the biggest service agent in our category nearly one hundred million arr eight thousand paying customers some of the biggest lows in the game thank you thank you from anthropic to i don't know snowflake the cool guys like polymarket and vent and everyone else are on finn so wonderful but where does it go next clearly it's just so obvious as each of these categories develop that there's just more to explore none of these categories are in any way fully baked and that's one of the interesting things about all of these ai spaces that we're still kind of pulling the thread and seeing what's going to show up next and the next big play for fin is not just answering service queries but really getting deep into every stage in the customer life cycle so we call that the customer agent and this money will primarily bring the

2:02:36

customer agent to market okay so when i want great customer service for a company that i'm running i want yeah you know search the knowledge base answer the obvious questions handle the refunds track that's part of the customer journey are there other touch points that are unlocked by ai along that journey absolutely and

2:04:06

completely yes from day zero on the website when the customer has very basic questions when they get more interesting when the business wants to start to qualify and understand who that customer is when the customer actually then wants their handheld when they sign up or they want to actually engage and learn and meet salespeople they want to book a demo or they want to talk to a virtual salesperson i mean that's just the very top of the funnel alone to get into the app if it's b two b there's all the onboarding all the detailed help required to get something successful if it's e commerce it's actually finding products and getting it into the cart and bringing them back for more shopping so there is really literally the full customer lifecycle of work to be done by agents and customer service was just the little sliver i want to know what's happening off camera though oh

2:04:28

we just have our next guest walking in into the studio got it got it got it so i'm just hopefully they're less interesting to me yes i want to talk about some of the best customer service agents that i've hired have sort of wound up real people real people have wound up in sales roles almost like they have a fish on the line they're reeling them in saying you don't want to cancel we actually this product does exactly what you want and the fact that it was delivered once day late we're happy to refund you but it seems like the underlying issue is that you actually want this product instead of this product why don't we get you over here and that's this sort of different skill set that comes out in the best customer service agents how are you thinking about actually driving incremental sales through customer service

2:05:17

agents yeah i think one of the interesting things is that a lot of the work in these functions isn't entirely separate yeah we just needed to draw lines between them when we were running human organizations because you can't possibly have one person who's outstanding at service sales marketing success et cetera that's just things

2:06:07

going to go away with the saas market maps too complicated totally a pain

2:06:25

in the ass but unfortunately that pain in the ass is coming with ai because i just don't believe you're going to have these separate sales agents and marketing agents you're not going to have these things fighting with each other having different context and information different memory in different parts of the customer life cycle different messengers not going to happen there's going to be vendors that will provide one seamless solution across the customer life cycle and that's a big thing we're passionate about and one of the very fascinating things we've seen we've been building this for quite a number of months now and we're going to launch the first part of it next month one of the fascinating things we've seen is that when there's an underlying relationship with the agent and then there's actual connection in the moment and then there's clear intent the context is primed for selling and encouraging the user to do something next if you're on a website and you get a super annoying pop up that says hey john or jordy in fact they're not even going to use your name they're going to say do you want to learn more about our whatever and you're like get out of here it's like not the right time or place whereas if you're actively discussing it can ask you questions back and it's so much more effective and so we've seen just these things the service agent when it starts to do sales for example and vice versa they're far more effective and so the real magic from ai across the board and of course this agent world is going to be when the agents can do things

2:06:31

that humans can't yeah one thing that stands out is like right now you have this spikiness in dialogue as a customer of a business right you might have a question coming in and it creates a moment and then it goes away and then you get maybe transition to someone else and you're talking to a real human you talk with them a little bit and you get passed to someone else and there's something about creating i think the customer experience of having this continuous conversation conversation with the organization that you're a customer of that i think will be super powerful because it just means that you know a split second away the entire time i was talking with some guys that own a number of like big car dealerships recently and i was talking to them about ai and they weren't fully sold yet because they use like chatgpt but they haven't had anything spread across their whole organization and i was explaining to them how there's so many different moments where like a customer lands on their website and they want to know if a car is available but somebody gets back to them too slowly and like the light was like finally going off to them do you think that there's a number how many industries out there have been kind of resistant to even using software from silicon valley to date that are going to kind of take a second look at all the different tools and things like fin now that they maybe have had a cool experience with ai in their personal life and they're basically saying they're willing to kind of look because maybe the historical saas platforms they tried them but it wasn't quite magical yet yeah there'll be so

2:08:02

many spaces so many verticals that some smart entrepreneurs will benefit from from bringing sexy technology to that aren't immediately obvious to those of us in silicon valley you know we tend to sell to ourselves it's really insular so all the names i just mentioned there like anthropic and snowflake they're all the cool brands in our world however there's like car dealerships that are you know i don't know maybe it's not as big as anthropic but they're pretty damn big depending

2:09:42

on the car dealership i would put car dealership above anthropic and snowflake personally if it's like a ferrari dealership lamborghini dealership like just on the cool side

2:10:10

we gotta give it on the aura factor sure for sure maybe not on the economic factor but there are so many industries that just don't have our attention and so there will be you know there'll be companies like ours that sell into many verticals but really we'll focus on all of the verticals that are adjacent to ours and those you know tens of billions of dollars of opportunity just in that alone but there'll be people who will say we're going to make agents for dentists car dealerships or certain government bodies maybe even for tax bodies like there'll be people who will do this because so much of the effort is not just in building the technology it's actually creating the trust as you say jordi with the actual people who will have to deploy this technology and make it work for them with their legacy system so there's a lot of very unsexy work that will be rich and rewarding for some people who bring it to these places talk

2:10:18

about the decision to raise debt instead of equity not something you see yeah

2:11:12

i know i just realized i had this kind of blind spot for this as an option for capital and there's just so much equity opportunities out there we raised a bunch last year for is secondary but when you're in our dynamic where you're accelerating as quickly as we are every year we're doubling our rate of growth pretty much there's no way we're going to be able to achieve the valuation that we're pretty confident we'll get next year and so any equity raise is going to just be massively dilutive we calculated that raising this in debt is going to cost us about a tenth the kind of price to shareholders that it would cost in equity so this is you know the incumbents have disadvantages they tend to move a little slower they have older businesses they need to wrangle with but they have access to phenomenal amounts of debt like two hundred fifty million is a relatively small amount of debt for the size of our business and i think that you're going to start to see a lot of these older saas companies as they pivot to ai and they actually find new growth start to lean on debt as an opportunity to invest rather than super dilutive equity and honestly i just think it takes a little bit of maturity to take that route

2:11:18

there's so much it's not just maturity it's confidence to know your business well

2:12:37

i like to think so but the reason i say maturity is that there is so much temptation to go and attain that latest hottest valuation your employees want it your previous investors want it media wants it but if you're able to tell them we don't need it actually it's way cheaper for us to get a lot more capital with far less effort then you can actually thread the needle in a way that i think is going to work much better

2:12:45

for big companies like ours the gong over here rings just as loud for debt as it does for equity so don't let them yeah i didn't actually

2:13:11

you guys are hiring a ton you said you're bringing out six hundred and fifty new people this year yes yeah

2:13:17

so that's gross i think net it'll be a little less than that it's a phenomenal amount of hiring i fear it i celebrate it with great caution there are just countless stories of the hubris of ceo's who boasted about how many people they've hired and it's all ended in tears so it's actually not a point of pride frankly i think small is beautiful but we are growing incredibly fast and we need all the

2:13:26

help we can get and you've been at this long enough to know the impact of what one hundred people six hundred people what impact that will have on your organization so i have the full faith in you thank you for that thank you so much for taking the time to come chat with us

2:13:53

great to see you congratulations good to

2:14:11

see you both well take care keep doing your thing all the best goodbye

2:14:13

let me tell you our railway railway is the all in one intelligent cloud provider use your favorite agent to deploy apps servers databases and more while railway automatically takes care of scaling monitoring and security and let me also tell you about sentry sentry shows developers what's broken and helps them fix it fast that's why one hundred fifty thousand organizations use it to keep their apps working and without further ado we have alex epstein he's the author of fossil future live in the tvpn ultradome welcome to the show good to see you thank you so much for coming down on short

2:14:17

notice what are your text been like the last forty eight hours is it over because you're john and i were thinking who's our oil guy we had one we had one yeah and i'm

2:14:46

local too yeah if i'm in town it's great i always like coming in

2:15:00

person versus thank you so much so

2:15:03

much versus the other stuff yeah appreciate it i mean there's so much in energy but i'm happy to talk about

2:15:05

this i'm happy that you lead the conversation however you think it should go what's going on how should we even be thinking about oil prices right now the impact of the war in iran how should we be processing this news

2:15:10

okay so we did this so there's a million things to comment on so i'm going to try to segment what i comment on but we did get involved militarily in iran which i would just say in my non area of expertise in general that is a good idea i think it's a better idea to have congress involved if you can get them involved and i think it's a good idea to have a lot of expertise on international oil markets being brought to bear when you do this and i think this administration does a lot of good stuff on energy but i think it's pretty clear at this point that that there was not maximum expertise brought in on this and so

2:15:27

the standard issue when you're talking indicators there would be the strategic petroleum reserve hadn't been refilled we can talk about

2:16:05

that so let's make sure to get back to that one but the main thing is just rough numbers strait of hormuz which iran has control over is twenty percent of the world's oil production is flowing through that every day and so whether you're looking backward or looking forward there is no replacement whatsoever for opening that straight and keeping it open and one of the things we've heard from different people is oh we have a lot of options on the table there's a lot of things that we can do there are definite things that you can do but none of them is twenty million right so it's just i mean that's more than us oil production just flowing through that one place so we can talk about the other things but i think the first thing people need to recognize is if you do not get that thing open then you have dramatically higher oil prices and then the other thing to get is

2:16:13

oil prices really matter straight from hughes dumb question can't you just go around

2:17:05

i mean you can try to some extent but there's a reason why twenty million a day are being routed through

2:17:11

because it actually comes out of there

2:17:18

as well right yeah there's just i mean all of these things are very very optimized one of the thing about just understanding the energy industry is like things are just very optimized in terms of where the infrastructure you're doing all of these things so there are other and we'll talk about this i mean there are certain pipelines you can pipe more oil through there are alternate routes but we're talking in the millions a day maybe most of our other options are sort of in the one to two million barrels a day and just so people know barrels forty two gallons so you're talking about almost one hundred million gallons a day for the stuff but yeah so you don't have a lot of great options and i think that's very important for i'm going to give some options but the number one option is keep that thing open and there are basically two ways to do that one is you just win the war if you want to call it that a lot more quickly like if you get some form of surrender and you have friendly people controlling it then guess what it can open that's one thing so i'm not an expert on how to do that but schematically doing that is very effective the other thing

2:17:19

which is a little bit game theory for iran is they actually benefit from global markets being in turmoil because it gives them some leverage over any type

2:18:20

of negotiation this is the biggest damage that they can do i mean clearly we've seen they're not gonna do they're not gonna send missiles to israel and get rid of israel let alone do anything to the us sure but they have this incredible control of the one of the centers of the world economy and they have a bunch of stuff going on there so let's talk about that so option one is you get a surrender and they don't use all of their options but then let's talk about their options so they have mines in this thing so they can blow up mines they have missiles maybe the worst thing they have and this is a change in recent years and you've had palmer on here and ethan on here the drone thing is just a total game changer because you can have a small vehicle you can launch one of these shah heads out of the back of a truck they can go really far and that's a lot harder to deal with than a fixed installation

2:18:34

yeah they have something like five hundred miles of range on a typical yeah

2:19:24

depending on yeah right so truck drives

2:19:28

up to a beach drone launches as soon as it finds that cargo tanker that has a bunch of oil on

2:19:31

it it hits it yeah and then there's the you're worried about that as you're going through so you have to think about those are the kinds of threats you face so how do you deal with that i mean basically you need the us to lead some sort of convoy where people have sufficient security and economic assurances that will go through so what's involved in that well one thing is if possible you want to get allies involved and it's even possible you get unconventional allies like china who has well they depend on this now they have a bunch of unlike us you want to go back to spr they've been filling up their reserves so they have more reserves relative to their import ports than we do relative to ours if you think about ours we have about four hundred in the spr and we can release about four million a day so we don't have as much but still they really care about this and a protracted thing is bad but for sure japan south korea india so one thing is just at a high level if you're doing this convoy can we get these countries that are aligned economically maybe china but definitely these others can we get them involved because then it's an addition they have military presence there that's an additional threat to iran if they attack one of their ships but that's a macro thing is just can you do this alone or can you do it with allies i think ideally you would do it with allies then there's a question of what you do about these various things and i think the biggest i mean the other macro thing you can do is just in some way credibly threaten iran and say hey if you and i can't give the details of how to do it but if you attack us if you attack in the strait of hormuz it's gonna be very very bad for you so just at a high level can you make that threat so they understand this is gonna be very very no we've wiped out a couple layers of leadership so they might be taking these things seriously after that i think the biggest thing from what i've heard from military experts is just these drones like what do you do about these drones how do you give any kinds of assurances and as far as i can tell we don't have any one perfect solution but you can do a certain amount of stuff from the air you can do a certain amount of stuff from the ground although it's very very expensive and then most cost effective where you can do it is you can take out certain stockpiles and facilities with the proviso these things are decentralized and of course i have no specific knowledge about where they are but

2:19:35

in theory if there's the suicide drone the shahid that there's a factory and you take out that factory you have

2:21:55

reduced that right so that's gonna be it and you know i know you guys had on ethan thornton who i'm friends with and he talks about this a lot as just this cost asymmetry issue is you just want to be doing things even if you're way wealthier you do not want to be spending three million dollars to take out thirty

2:22:01

thousand dollars which might be what's happening with a patriot missile battery yeah that's

2:22:17

the kind of thing that happens so you need a combination of those kinds of things and then the other thing so that's all the security stuff so you get the allies involved you take these actions specifically against drones i think what was i going to say about the other thing you can do i'm just going to make sure there's still

2:22:21

just well the other big thing oh

2:22:38

the insurance okay insurance yeah the insurance and this has been floated by the president and it's a reasonable idea but my understanding is the insurance vehicle we have the development corporation does not have enough funding so you might need to go to congress but basically you want to be able to say hey we're going to lead this as the united states we're going to lead it with allies if iran does anything it's big trouble for them and we can counter their specific attacks and if something goes wrong you're insured you're trying to create that confidence to get enough people going through and then you get a certain amount of stability and then you start to get closer to your twenty million barrels a day either that or you just defeat them really quickly or both

2:22:40

so the other factor is just the loss of production disruption to production plants shutting down plants being damaged how does that factor in do those do refineries get brought back and i'm talking about an allied country you mean just the

2:23:20

trend over time or what yeah generally

2:23:40

i mean just seeing the videos coming out you've seen places i think like qatar kuwait places like yeah basically just saying like yeah we're going to pause production because we don't even have a place to store this and or they're suffering damages and it doesn't make sense to keep refineries online but this is

2:23:42

all just downstream of do you have the route open got it if you have the route open then you're good it solves everything if you don't the way to think of the other things we do have some other interesting options we have to think of them as these are temporary stop gaps if anyone says i mean you hear some crazy things people are like oh venezuela we talked about venezuela is less than a million i mean it's beyond crazy it's just impossible it's not like venezuela has the ability to just increase their oil production by any significant amount in the near future i mean we're talking about companies considering going in harm's way to get some incremental boosts so venezuela's essentially

2:24:02

useless so best case scenario in the next year venezuela goes from under a million to two million no i mean

2:24:38

two million would be i mean if we made it our entire goal in life i don't know but why would

2:24:45

you for that venezuela was at like three million like years ago or something

2:24:50

yeah some years ago yeah yeah right

2:24:54

but but there's a lot there it's

2:24:56

not gonna happen it's just so yeah so venezuela i would not even that's just in the category of it's not even a bad idea it's just not an idea there's nothing there's nothing there's nothing there no no that's right so the bad ideas by the way are the worst idea imaginable is ban oil exports from the united states which i have heard floating in conversation seems somewhat

2:24:59

reasonable i want the oil yeah of

2:25:19

course because we have oil so we don't want to sell so i mean so many different things but one obvious one not obvious one but it's obvious if you know how these things work is the refiners in the us do not match up well to the oil produced in the united states the refineries in the united states are based on heavier crude like crude is rated in terms of weight and sourness and so our refineries are mostly for heavier crude it's actually one reason why venezuela has some appeal canada has a lot of

2:25:20

appeal because they have heavier crude yeah

2:25:48

and we're going to talk about canada

2:25:50

and we're good at refining it but we produce a lighter crude yes because

2:25:51

the shale revolution was just this dramatic shift very quickly in oil production so our refineries are not primarily equipped that's why when we ended the crude export oil ban i forget it was twenty fourteen or something like that it was just this huge unlock because it allows the global market it allows us to produce for a global market so you

2:25:54

can't look at american oil production as a monolith yeah it's not fun to import heavy crude and export light crude

2:26:11

is that roughly correct that's roughly correct so that's one thing it's just insane to do that but in general think about it what is the value of higher prices is it stimulates production we're going to talk about all these different ways where we want to unlock oil so you want to tell american oil producers hey we're going to actually totally screw you over it's actually going to be worse for you maybe than it was before because you don't even have a market so you're just going to strand all this so that's terrible there's also this financial manipulation idea which has been floated which i think is terrible of hey let's manipulate the financial markets basically in a way that where you're selling futures short on a certain time period and then you're buying them what you're trying to do is lower the

2:26:18

near term price oh like fed raise

2:26:55

the long policy or treasury yeah so

2:26:57

a very american way to try to

2:26:59

solve the real world problem people like

2:27:01

that the only solutions involve unlocking oil in one way or another you do not want to screw up the markets so that people are less inclined to unlock oil nor do you want to screw up the markets to destroy information so again the main unlock is reopen that strait there's no way around that but then we can talk about other things so one category is what you call spare or emergency capacity and there's some uncertainty here but quickly this is

2:27:04

separate from the strategic reserve yeah well

2:27:32

emergency capacity is the strategic reserve oh it is so spare means you could pretty easily be producing more per day but you're not because you're not happy with the price and this is mainly saudi arabia is considered exhibit a in terms of amount of this now there's debate among experts about how much they have but but some people estimate they have one two three million barrels a day of spare capacity that could ramp up on some kind of and do

2:27:34

they know how much capacity they have and they're just not sharing it or does no one know i mean that's

2:27:59

always been the story that's been the story of the gulf right they're not going to if they just flooded the market with every single barrel that they could possibly produce they would be getting a worse much worse price yeah and

2:28:04

there's this cartel arrangement that does i mean in a sense everyone has this on different timescales you think about us shale producers if prices were sustained at one hundred dollars guess what there's a whole bunch of shale deposits that they could produce at seventy five dollars a barrel and make a fortune on that they're not going to do four weeks ago when it was around dollar sixty a barrel but in terms of the thing about the saudi oil is it's produced at a much lower cost so this is them saying hey we have some of this on the table and then the question is you need to be able to produce it and then you need to be able to transport it so you have estimates around maybe you could get one to two million barrels a day so that's one of them in terms of emergency if you look around the world you don't have a lot of immediate spare capacity besides that so we have what's called emergency capacity or strategic petroleum reserve geordi mentioned that we didn't fill up our strategic petroleum reserve i think this is unequivocally a mistake i mean if you look at the whole thing of the strategic petroleum reserve is you want it for when you need it and the best possible time to fill it up is

2:28:14

when prices are low and is that physically a place with a bunch of barrels that we there's a bunch of different knocks of oil well there's multiple

2:29:16

different places but but you can think

2:29:24

of we basically cash the physical oil in america or somewhere where we own it or can access it yes and it's sitting there going unused until we're

2:29:27

ready yeah now there's issues of biden biden definitely misused it so they they used it to basically have lower gasoline prices during midterm elections even though parenthetically their entire policy goal was get rid of fossil fuels which means you want the price to go up so people can't afford it it's a very cynical kind of move they did it in a way that degraded the facilities so let's say we're at about four million barrels a day we can get out of that thing we have four hundred million in there we should have over seven hundred million we could have easily filled it up i mean it was a perfect time when you're talking about dollar fifty barrel oil sixty dollars a

2:29:36

barrel oil can i do the basic arithmetic of four million a day four hundred million in the reserve one hundred days of oil but that's not actually if we're talking about relieving price pressure you could trickle it out over a year and have one quarter of the

2:30:09

effect but just think about one hundred million barrels a day is roughly the global market okay okay so four million is just that's its maximum throughput capacity got it it's kind of an analogy to batteries we're talking about batteries so there's a certain amount like with a battery if you hear there's a one gigawatt battery installation that means that's the it usually means there's one gigawatt for four hours but if you wanted to do half a gigawatt then you could do eight hours right it's the same deal with this but keep in mind this is not one hundred days of us oil demand and by the way for the refinery reasons we can't just supply it all with our hpr it's just the maximum output is but four million barrels a day is something we're talking that's one fifth of what's flowing through the strait of hormuz yeah it's a thing but yeah you can think of yeah maybe we'd be willing to do one to two million barrels a day from ours now interestingly the international energy agency we are part of the international energy agency reserve program so we have about four hundred but they have about one point four hundred million they have one point four billion so we have allies around the world that can also release this oil so maybe we could get one to two million from them now at least last i checked the news they hadn't agreed to do this because they don't think it's an emergency but that's what it's there for so notice the pattern is at what

2:30:28

point would they think it's an emergency

2:31:43

i don't know i mean and we can put pressure on them i think some of these determinations have to be made in conjunction with the military determinations you can't make them in isolation because if you think about what's the military objective how close are we to that or how close are we at least to opening this up via high security and financially backed convoys then you can think about the timetable here if you think about well the strait of hormu is going to be closed for five years let's just get used to being poor black bill no i mean this is people don't get oil like oil is there's a reason why i focus a lot of my life on oil there is nothing more there's no material in the world of energy that is more valuable than oil because oil is just so unique in terms of it has this very high energy density and portability and there's nothing like i mean nuclear has plutonium yeah yeah five years have the same portability we're going to

2:31:46

be rich we're going to have energy too cheap to meter in five years i talked to a lot of nuclear

2:32:43

founders well look i love nuclear as much as anyway they don't even have a portability solution though for the near future unless you're talking about aircraft carrier or an icebreaker or something like that so i'm just saying the world runs on mobility and oil is ideal for

2:32:47

mobility so portability in the sense of using nuclear to actually power trade and commerce and moving ships around yeah yeah i mean even if you can power the grid you will be poor because you will not be able to trade

2:33:00

yeah trade is really important trade is really important trade is trade and well and also it's within the country too we don't have the nuclear trucks to move things around sure sure yes obviously you want super cheap electricity and you want super cheap transport fuel and we want to see if we can electrify some of this stuff if you can do it cost effectively there's just some

2:33:12

people working on like electric bulldozers for example it's very hard and by the

2:33:31

way guess what you need a lot of oil to even get that whole supply chain started because you need the mining anyway bottom line yes i meant that thing about we can all be poor very seriously very seriously so we have if you look at the spare capacity i think that's all of them so we have the rest of the iea we have ours and then to my knowledge we have saudi slash uae them using spare capacity and them piping it where they can so maybe combined we're talking about six million a day if you use all of those i mean somebody could imagine ten million and all of these have time particularly the emergency reserve things these all like at least the saudi one is you're sort of you have a deposit a big deposit here it's just you have a small reserve so you can't do this forever and of course there's a risk of if you deplete the spr more what if you really really need it in the future and then there's a bunch of nearer term stuff that you can do and some of which i like just because it could be done and anyway so number one to piss a whole bunch of people off is let's stop using this jones act do you know what this jones act thing

2:33:36

is i've heard of it but explain

2:34:37

so the jones act is a set of restrictions that say that you can only transport things among ports in the united states if you have everything is basically american so it's owned by an american it's run i don't know all the it has all the yeah it's

2:34:39

basically american ship american crew yes so

2:34:54

it leads to all these crazy crazy things like we end up importing things like instead of importing it from one part of the country to another you end up importing it from some really far away place it leads to that kind of thing but in this case in particular as californians we should all recognize this we don't have oil being piped into california so we want very efficient maritime transportation and if you suspend the jones act then you can get more efficient transportation which means we can at least have lower prices here and we can have less of a price shock but in general i want people to get used to having no jones act because i want there to be no jones act so that's a good thing to do my favorite idea here

2:34:58

and i imagine there's a bunch of powerful lobbies that will fight and kill for the jones act because there's an entire industry built around yeah it's the

2:35:40

idea that we can only have a good shipbuilding industry i don't know what people think is happening with our shipping like our shipbuilding industry is not going

2:35:51

as well as other people didn't exactly it didn't exactly work yeah i mean

2:35:58

it worked to protect certain people's i'm

2:36:02

just saying if you look at american shipbuilding it certainly hasn't allowed us to be competitive on a global stage the

2:36:05

shipbuilders would say we're early but i

2:36:12

take your point early but you've seen but i'm just saying like how many decades of declines yeah okay well you

2:36:13

need i think john's like still early

2:36:21

it's like a smiling curve like churn it goes down and then goes back

2:36:24

up so that's the bull case that's the all right well it's definitely not going to help us in your turn so here's the most interesting thing which is a pet issue of mine because we're like this administration i should say does a lot of good stuff on energy i say that i i'm in a fortunate position where i get a lot of people ask for advice and stuff like that and i feel like one of my jobs is to tell people things they're not doing a lot of things they'll just do right on their own but canada is not one of those things right now we are just absolutely sleeping on canada as an opportunity so canada has no people and infinite resources and is really friendly so they have these oil sands you know

2:36:27

what these things are you know explain

2:37:08

so they're just like these deposits of oil it's like nature basically committed an oil spill is the way to think of it okay so there's all these oil sands for years they were not

2:37:09

very useful they weren't viable because it's not just you drill down black gold comes up it's like buried in the sands need to be repaired there are

2:37:18

different kinds of things so you can mine it out directly or you can do what's called in situ which is you heat it up underground oh and it liquefies in any case they have just this unbelievable oil seep they have way more oil deposits than we do way more oil reserves but they have on the order of a third of the production and in part because they have absolutely terrible policies which are their fault but in part we have had terrible policies including getting rid of the keystone xl pipeline and this kind of thing so but if you just think of canada it's just we should be so they have water i mean they get no people one tenth the population infinite raw materials they got everything they got timber they got natural gas they got oil like and we're talking about venezuela they're a friendly country right they got smart people there they haven't been driven out of the country because they're afraid of getting killed by chavez or maduro or whatever so this is an opportunity to say hey canada let's have a task force you guys are at a low in terms of we don't have pipelines but we can do rail at least we can do rail maybe trucks but at least rail rail transports from canada are really low right now so let's have an initiative where maybe you bring a couple hundred thousand depending on your expectations maybe we can add a couple hundred thousand barrels a day so that's one that excites me because i want to be like the us canada superpower thing is this huge opportunity that we're sleeping on because of the idea of oh we should only be doing it in quote america but this is just by the way unbelievable i think they have the best uranium in

2:37:25

the world yeah it's maple syrup's up

2:38:55

there yeah i don't have a canadian personality but i love that place no strategically that makes me feel strategically and they're very friendly yes they're very friendly

2:38:58

maybe we should be friendlier yes you

2:39:05

know we should and we have trade agreements coming up the other thing is we have some you know we can increase capacity here as i mentioned depending on prices going up so there's that but that might be a couple hundred thousand barrels a day but so high level open the strait of hormuz that's the only thing that really works then you can talk about saudi mideast spare capacity our emergency capacity and iea emergency capacity and then below that you have canada jones act and us production that's

2:39:07

a really really thorough list thank you

2:39:38

i appreciate that anything else jordy what are you gonna be paying attention to most closely do you watch the price

2:39:40

chart is that important to you well

2:39:47

it's important i mean it's weird i work with i run like a kind of a network of energy executives but i'm focused on the policy side of things so i'm often laughably out of i mean i know in general within ten dollars usually what it is but as a policy i don't invest in

2:39:49

because you're looking at longer term five

2:40:08

years but also like i'm deliberately avoiding exposure to the commercial part of it

2:40:10

of course of course you want to

2:40:14

be independent yeah i want to be independent so yeah i'm not but at the moment yes i am looking i am looking at the prices but i don't do predictions but i'm just saying i think there's a service right now in just laying out the options for people because even in the government i mean not everyone has a lot of expertise in oil and this is the kind of thing where yeah it's kind of why i think you texted me this morning i mean sometimes you need to know sometimes you need to know about oil today and sometimes you don't but when you need to know you need to you really need to know

2:40:15

yeah well it seems like whether you're in the business community or just humanitarian everyone should be rooting for a swift and peaceful resolution yeah do you have

2:40:45

any insight this might be totally out of your wheelhouse but insight into how people running businesses that are most impacted by fluctuations in oil prices have they been actively hedging over the last couple weeks are they doing how do you run your business when your input costs can go up dramatically potentially at a much higher rate than you can adjust

2:40:56

your own prices just in general these guys are sophisticated i think within the bounds of doing it i don't think you can hedge indefinitely for this big a price kind of differential like somebody is going to say but it's also you think about there's on the consumer side but on the production side is even more dramatic if you just look at the margins of these guys i mean the oil industry is just fascinating i just remember anecdotally when i came out with my first major book the moral case for fossil fuels in twenty fourteen even i could see speaking engagements correlated to these prices because you had a big price crash around them and then you think wow we don't need alex yeah variable yeah it might be nice to keep the troops motivated but they'd probably rather be employed but it's a hard business and one of these things is the people who survive have a very strong constitution and they can manage the risk very very well i was talking with a bunch of these guys and actually people we know in common and some of the better ones some of the two of the young superstars in the industry they're just saying we love the chaos because we can just no we can handle it when the prices go down because you think about what they've been dealing with over the last year they've increased production prices have gone down and you look at this this administration is very focused i would say too much focused on having prices go low this is one of the variables here although the administration i don't think was not fully prepared for this situation in general they are maniacal about prices going low like trump in his mind has a very strong focus on dollar fifty a barrel which i think he feels like that's always the perfect price for oil and the people in the oil industry say wait a second like we've had inflation of everything else why does our one product have to stay at fifty dollars and that's part of the appeal of venezuela rightly or wrongly because it's not going to make a big difference soon but it's like how do we keep oil prices low but that's that's hard to be in that industry as a consumer as a producer rather and it's hard as a consumer so they have financial instruments but the main thing is you just you need to you need to just have the flexibility in your business model and oil is again it's the most valuable material in the world of energy which is why everyone uses it but it's also why it's so inelastic and you have these price fluctuations so you just need to come to terms with that or somebody needs to figure out something better but it's really hard and of course policy wise we can have better policy to make it more stable but it's still that thing where it's so valuable that guess what you get a little extra demand a little less supply price goes up a lot it's different than other things which you can substitute for much more rapidly you're out

2:41:21

of a job when we build the

2:44:17

dyson sphere oh man i'm working on this ai thought leader thing we talked about last time so i got plenty

2:44:17

to do we'll cross that bridge when

2:44:25

we come back i got plenty to do with my time thank you as

2:44:26

long as we need energy you'll be doing just fine all right guys thank

2:44:29

you so much for coming on the show 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 on marketplaces and now with ai agents and let me also tell you about applovin profitable advertising made easy with axon ai get access to over one billion daily active users and grow your business today and without further ado we have our next in person guest live in the ultradome michelle volz how are you doing good good to see you thank you so much for coming down to the tvpn ultradome for those who aren't familiar with your venture capital career and your marathon running career introduce

2:44:33

yourself yeah so i am michelle i was previously at andreessen horowitz i just launched my own fund pax which is a fifty million dollars early stage venture fund focused on foundational categories so i was on the american dynamism team andreessen horowitz similar themes to that just earlier

2:45:07

stage biggest lessons from can we hit

2:45:26

the gong first oh yes we gotta hit the gong i think michelle should

2:45:28

here you go hit the gong here

2:45:33

in person go for it now that's already sort of broke you won't break

2:45:34

it smash that was good anyway let's start with like lessons from andreessen horowitz like what did you learn what are your biggest takeaways what are what's the playbook that you think you will be continuing and then we can go into what you think might be different yeah

2:45:38

i mean i think in the categories that i look at it is common for founders to need to raise a lot of capital okay and the ability to be able to fundraise for that capital ahead of maybe what our traditional milestones in software investing is very important you need to be able to be a magnet for capital for talent and sell a vision because what you're doing is like very hard with fewer obvious proof points maybe at the early days and i think identifying that in founders early is something that andreessen horowitz is

2:46:00

very good at yeah in american dynamism companies hard tech companies it does feel like there's like high capex requirements high humanity capital requirements you're not going to automate everyone on your supply chain on day one and so there is this like how you talk about growth how you talk about about scaling and success and finding product market fit it's it almost feels back to like the okay we're scaling dau and we're let's focus on that metric instead of like profit in software what does it look like in hard tech is that the okay we have a program of record or we have these sbrs coming in or we have this long contract that is very solid but we're not going to draw down on it for a few years so we have a capital need but how do you think founders in hard tech should be talking about what progress looks like at top line or whatever top line means yeah i mean

2:46:33

i think that's the hardest part like in software there's such a structured set of metrics that everybody has been trained around like growth rate like weekly active users revenue it is so different company by company in these sectors and that's why the storytelling is important because if you're talking to later stage investors they might be trying to make a pattern recognition off of a pattern that doesn't quite exist yet and so in defense companies like yes there are government contracts you should be trying to get you should be working towards programs of record but there are like different intermediate proof points before that i usually explain it to founders as like the three things you're trying to make progress on are people product and then like revenue or traction and so it's like are you hiring the best people and the people that are experts in the different categories that you need so maybe you need a mechanical engineering hire maybe you need a government sales hire maybe you need software as well are you making progress on people are you making progress on product that's tangible like can you show that there's like developments maybe that's like you've built something and it's like test like gotten to a test milestone maybe it's in the hands of users like are you making product progress and then are you getting some sort of signs from customers that like they are willing to pay for this like are you getting contracts are you getting lois like are you getting something that shows that there's a commercial use case for this i think sometimes in like hard tech there's a million names hard tech deep tech frontier tech american dynamism i think sometimes people skew more towards thinking it's just deep tech which is like i would science project exactly like science risk like nuclear fusion is probably like the most extreme example that's hard like it's like of course there's a customer if you can do this but like we'll see what the timeline is hopefully soon but i think in a lot of the categories i look at there are commercial proof points you can make like there's usually a latent demand or like or an immediate demand it's like how do you show that you can get those customers and they can like signal that they will buy from you yeah

2:47:31

there was this viral citrini article that said software is dead sell everything how are lp's thinking about hard tech as a category with that backdrop of like the saas apocalypse i imagine it makes things easier for you but has it actually rippled through the lp community to the degree that your investors are more excited than ever to have a hedge against chaos in automatable software all of

2:49:30

that yes i will say this with love to all of my lp's i think founders are typically like leading indicators categories like they they are at the cutting edge like they are seeing around corners often like great founders will see something before vc's like vc's are lagging indicators like once something is hypey it's like usually too late to find a good early stage investment lp's are lagging lagging indicators that once something is very mainstream and there starts to be some signs of returns then they're like we should pay attention to this category got it and so i would say thing that was more surprising partly because i live in my own echo chamber and constantly hear about all of these great companies building in defense tech hard tech and industrial tech lp's weren't as familiar as i expected so there was a level of education i think there's still some wariness there haven't been a ton of exits yet i think that might change if spacex ipos would be one of the biggest ipos ever in history but i would say like the the appetite for it grew over time as

2:49:59

i was fundraising on space x yeah i mean great point it's it could be one point seven five trillion is the number that's being floated it's staggering amount it's going to return like not just like one venture capital fund it's going to return like twenty venture capital funds like everyone who invests at early stages going to see their fund return from that but all the lp's had to go on a twenty year journey with like a million continuation vehicles how are you thinking about timeline messaging to lp's because ten to twelve years is what i've heard recently is kind of standard but i've talked to some vc's and say yeah like most of my lp's want their money back in like six to seven years and yet in hard tech it feels like it might be longer are you communicating a specific particular timeline or is that even a

2:51:06

relevant conversation yeah i think every every fund roughly pitches ten to twelve years that's sort of the expectation i think it's actually an interesting catch twenty two because if you had ownership in spacex your lp's actually hold on to that yeah totally like keep that i think it's going to keep going up but maybe you would have an option to take some off the table and return some money to lp's if that's what they wanted but often if it's like a really really good deal lp's are like i think it's gonna keep going up but i think it's like case by case i actually just was talking to lp's about this and they were like sometimes it makes sense and sometimes like hold on and i think if you build a lot of trust with your lp's like they're usually along the

2:51:52

ride with you what what geos are you spending the most time right now obviously the it feels like we have so many many so many founders that are just like in our kind of neck of the woods here but are you spending a bunch of time in texas bay area where do you expect to deploy most of your dollars yeah

2:52:35

i think there's five geos it's san francisco louisiana austin new york and dc i'm actually seeing a significant amount of companies in dc dc austin i would say though is the biggest surprise for me i know people have been trying to make austin happen for a while and i was skeptical and now i think like it's sort of been an explosion of great companies it was kind

2:52:53

of the same thing with la like there was this long la movement a while ago snapchat was sort of like the first big like you know public tech company or one of them to come out of la and then the el segundo thing just happened like completely organically without any of those people that were set cheering for la and it just sort of popped up and it seems like austin's got going through a similar thing what's unique about the companies that are building in dc are they thinking more demand generation first more more work on capitol hill first and then maybe they'll set up a satellite manufacturing office or work with like a contract manufacturer to deliver their capability yeah i

2:53:14

think like you're seeing founders who are working at companies like like palantir for example that have a big office in dc they've been relocated to dc they're close to their customers and then they spin out start a company usually these companies will end up opening an office in new york as well it's just a deeper pool for engineers but there's a surprising amount of really strong engineering talent in dc that is like pretty fixed like they want to stay in dc they have a life they have a house they have a family so it can be a very good place to start a company what's the view

2:53:49

on shipbuilding jordy says it's impossible to build a ship in this country you're saying it can't be done is it

2:54:22

possible we need to figure it out otherwise you're going to be behind i think i mean like the optimistic take would be like we can make it possible we should make it possible the pessimistic take is if we don't make any changes into the regulatory environment or the cost of building these things it might not be and that would be

2:54:27

a shame yeah that makes sense strategy

2:54:47

with to fund these like two million dollars lead three seed type checks what's

2:54:50

the deal yeah on average one to two million dollar checks like early stage try to be in the pre seed seed occasionally a little bit later rounds like to partner really closely with founders and i think the special thing about the early stage which is just my favorite stage of company is you get to be very aligned it's like a significant check for me if i put one to two million dollars into your company and i'm very motivated to help you be successful to get to the next round and then for that next round i can be much more strategic in helping you get there because i can't lead your next round whereas i think sometimes at the multi stage funds it can be a little bit tricky to give advice for future fundraisers because you might be able to participate and so you have to be more delicate about it at a big fund and as a small fund you just get to be totally aligned with the founder

2:54:56

one last question about sort of the category targeting american dynamism made a lot of sense as like a theme around defense tech and hard tech and you put a manufacturing company in there one of the interesting subcategories of ad that always popped out to me was something around education like re educating workforce education also just like education technology broadly do you think that that's within your purview or is that something that you're sort of carving out and focusing more on the deep tech hard tech physical products

2:55:49

yeah i would say i like to look at things that sit at the infrastructure layer of society which could include education and housing sometimes even healthcare often thinks at the intersection of hardware and software or tech and government or just like doing something fundamentally important for like the broad majority of americans i think american dynamism is this beautiful phrase that can be an umbrella phrase as well i sort of embrace that as well with pax but at the early stage like what you're really betting on is people like there is so little to understand except like the team and a dream like are you building something in a category that's very exciting or you're taking the big swing do you sort of have a right to win in that category and some sort of unfair advantage i like founders that have like worked in the industries that they are building in before and do i believe you can be that magnet for capital and talent and everything yeah where where

2:56:23

are the big talent pools these days there's the spacex mafia the palantir mafia it feels like there's an anduril mafia now where else probably going to end

2:57:16

up base power basepower maybe in austin

2:57:23

austin eventually yeah i think saronic is starting to see a few things basically any company that's that's like reaching escape velocity or like escape velocity and product market fit has grown really quickly has hired exceptional people there are some people that really love that zero to one stage and the company gets big like i think we're seeing it at anduril we saw it at palantir we continue to see it at palantir people learn how to grow quickly and then want to apply that to their own company

2:57:26

amazing well thank you so much congratulations thank you for the rest of your

2:57:55

day we'll have a lot of your

2:57:59

companies on let me tell you about vanta automate compliance and security vanta is the leading ai trust management platform and let's go over to some hard tech that's happening with stargate by the way

2:58:00

john apparently trump said something to the effect of i think the iran war is very complete pretty much oh okay well so we may be on the verge of new era crude is back

2:58:16

down to like pre like friday pre

2:58:31

weekend what does that mean eighty ninety

2:58:33

eighty three okay eighty three eighty three

2:58:35

okay well maybe he was watching maybe he was watching alex epstein and alex said hey we got we got to wrap this up we got to stop the price is too too darn high says alex epstein oracle and openai allegedly dropped their stargate expansion lease meta swooped in the same day ben pouladin says that's not weakness that's a hot real estate market and then he goes on to comment that the real story is openai passed on six additional buildings because power won't be ready for a year the power is the gating factor here not the chips interestingly we are in the power bottleneck the energy bottleneck and by then rubin replaces blackwell so why are you going to build a blackwell data center when you could build a rubin data center so why expand with last gen chips the eight building four hundred thousand gpu base build continues this is smart capital allocation he says and so this is based on reporting from the information about oracle and openai sort of adjusting the plan for stargate i am particularly interested in digging into the stargate of china they have a project that's the chinese stargate funded to the tune of thirty six billion not not up at half a trillion but still

2:58:38

very big number it's a big number

3:00:03

and it's growing and i've been i've been digging into it a little bit trying to learn as much as i can we'll do a deep dive probably with someone at semi analysis who knows

3:00:04

a lot more about chat gbt is back at the top no way the app store leaderboard you know we do have to adam gpt says gpt five

3:00:13

four sense it's for gods you know what that meme is from big chungus so you get one anyway yeah so

3:00:22

so i'm kind of surprised by this for the the one reason being yeah i expected anthropic's been spending a ton of money on marketing they started ramping

3:00:30

it up yeah yeah i see that

3:00:39

super bowl i expected them to basically be willing to spend whatever it took to stay on top yeah but i

3:00:42

mean it's capital war so both sides are going to spend a of time lot of money right and you know

3:00:49

there was what wasn't the rise in the claude like ranking just because of the dow news and now that's like

3:00:54

not on the full page so yeah that's what i was getting i was like when the super bowl ad dropped and we had some harsh words for the nature of the characterization around ads our favorite thing in claude led us to launching claude with ads but our question was will this super bowl ad move the needle for the consumer app like a super bowl ad is a massive campaign millions tens of millions of people will see this it went viral like it's will it work on people will people see this ad that's like sort of deep in the weeds around how ads work in llms like will this be enough for people to go and hit the download button there's a hurdle there will they do it and they did it a little bit i think claude went from like like twenty six to eighteenth or it started climbing then and then you're right the department of war narrative really really shot up but as with any story in tech it has to go back and forth and back and forth and the memo that came out on last wednesday probably sort of reset the narrative there and changed how people were thinking about that and also these charts are based on momentum and so it's very very hard

3:01:00

to stay long if bought it had gotten to something like a million downloads a day according to mike krieger oh yeah and at the same time chatgpt last month i think did like fifty five million downloads so still not outpacing

3:02:10

paramount plus is number four in the free apps although how can it be free when you have to subscribe sort

3:02:24

of odd yeah that's the whole thing

3:02:30

but bullish for the ellison team karish

3:02:32

says umfs ruined agency so bad that i'm going to have to start using

3:02:35

saying gumption gumption gumption is really good i like gumption we hire for gumption ai gumption ai agent we only hire for gumption what's the noun version of gumption i don't know friend of the

3:02:39

show amjad said not having coding experience is becoming an advantage and shuriya says not knowing how to cook is becoming an advantage ceo of mcdonald's i love

3:02:55

it amjad has some great takes though this is just really funny easy riff it's all in good fun what else is going on lots of debate over the singularity pricing as will depew put it because barely ai which i believe is trung fan says cursors internal analysis shows how hard anthropic is subsidizing claude code last year a two hundred dollars monthly subscription could use two thousand dollars in compute now the same two hundred dollars a monthly monthly plan can consume five thousand in compute it does feel like there's a capital war i'm very interested to dig into the history of capital wars in winner take all markets but also in oligopoly markets you should

3:03:07

write the book on capital wars capital

3:03:55

wars are fun coogan i lived through it during the uber and lyft era there was there was a flippening for a little bit where lyft was was at the top of the charts and

3:03:57

the friendly rideshare company it was very

3:04:07

friendly the cars would pull up with a pink mustache on them and it had a very very fun whimsical branding and you would sit in the front seat of the car because you were

3:04:09

i could see claude did you not know this is a real thing yeah

3:04:18

so when lift supposed to do so when lyft so when lyft launched people see lyft and uber as like commodities at the time but lyft was like revolutionary and people were super black belt on uber because uber at the time was only black cars so they didn't allow anyone to just hop on the network like it was not a three sided marketplace or whatever two sided marketplace like you had to have a like a livery to tcv denver i forget what it's called but like you had to be registered as a limo driver and you had an escalade and normally someone would call you and say hey there's a wedding and we want you to come shuttle people around at this wedding and they would pay you for that and uber went to those folks and they said hey you are registered to legally drive this car as a professional driver we want to book you through our app and so the black black car became uber's main product their only product and then lyft figured out through some interesting strategy how to allow people to just show up with a prius or a toyota camry or corolla or whatever and so drivers on launch day i remember like the launch week in san francisco would just show up in just some random car but lyft gave them all pink mustaches huge pink mustache the team can pull up a picture of the huge pink mustaches on the front of the lyft cars that were very whimsical and it was remarkably cheap because uber had been somewhat subsidizing the black cars they hadn't been taking a huge margin they were losing money but the lyft cars were way cheaper because someone was just showing up with like their toyota corolla or whatever with their pink mustache on it and so you would hop in the front seat because this driver was your friend there you go the pink mustache on top of the what is that a crv or something and so and so you would hop in the front seat you they would ask you to pound the driver give him give him the knuckles give him knocks this is the thing i'm almost sure so you'd be like what's up and then you'd get in the front seat and they would drive you of course if you had a group you'd be in the back seat

3:04:22

but then everyone had to give knocks

3:06:28

or i think the knucks might have been optional but encourage recommended because it was like hey we're buddies and we're and we're fighting the war on the borg we're going up against the but

3:06:30

you were encouraged to get in the

3:06:38

front seat front seat front seat in the lift in the lift when you're taking a lift your friendly lyft driver shows up you sit in the front seat you give them nugs this is the thing pink mustache chungus big chungus

3:06:39

sends his regards all right so you got to hire a car but you have to be absolute boys you do

3:06:51

but but truth be told like a lot of the early lyft drivers were like super cool tech forward san francisco people and so you'd have a lot to talk about because people were it wasn't professionalized at all yet it was very much like two sided marketplace you'd make a lot of money as a driver there was no margin compression yet it was like free money everywhere for everyone and it was a capital war and lyft raised a ton of money and came into the market with a disruptively cheap product and wound up delivering a really fun experience that was very very cheap and then uber sort of fast followed and launched uber what's it called the normal one not black uber x uber x yes uber x and uber x was bring your own car and then and then the rest of the game played out and it became a capital fight and for a while like the meta was oh you're you're you're you need a lit you need a you need a ride you open lyft and you check the price and then you open uber and you check the price and people would check both prices constantly and then and then over time like the liquidity and power loss stuff came into john do you remember uber helicopter yes that was was it a real thing with blade i think it was a partnership with blade which was the uber for helicopters and it was a thing but in in like you know for a few weeks or something did you ever do it no no because uber uber's had a whole bunch of of different like stunty products like uber you know uber yachts or uber they did uber ice cream where you could get ice cream delivered that was like the precursor to uber eats they did an ice cream delivery truck where every uber driver checked in with the field office got a cooler in the back with a bunch of dry ice and a bunch of ice cream and you could press a button and for like dollar two they would bring you an ice cream they even did i think uber kittens and you could ask yeah uber dogs or something you could like dogs or something i remember that yeah it was very funny anyway lots of stunts lots of ways to get attention go viral and that was sort of the launch strategy they would do that for whenever they launch new markets they would come in and they'd be like we need to go viral on the local news we need to let the town know that we've arrived and so lots of lots of back and forth and the capital wars of twenty it was fun times i lived i fought through them what else is

3:06:56

going on i think that's a good place to call it well thank you

3:09:16

so much for listening we will see you tomorrow at eleven am sharp leave us five stars on apple podcasts and spotify sign up for our newsletter tvpn dot com and have a great rest

3:09:18

of your day it has been an honor and a privilege and i hope by the time the show goes live tomorrow the war is over me too

3:09:30

me too i would love that that'd

3:09:37

be quite nice thank you thank you goodbye cheers

3:09:39