Anthropic's $2T IPO, Zuck's AI Manifesto, Nvidia's $500B AI Bet, Grok's Comeback

Key Takeaway

Run a 30-minute “AI leverage audit” today: identify one repetitive knowledge task you do each week—research, drafting, coding, analysis, or customer follow-up—and test an AI tool against it. Measure time saved, quality, and where human judgment remains essential. The episode’s practical message is that advantage will increasingly accrue to people who pair their own expertise with AI, rather than waiting for a single platform or model to win.

Episode Overview

The panel examines reports of a potential Anthropic IPO, explosive AI revenue growth, and the infrastructure—power, data centers, and GPUs—required to meet demand. They debate open versus centralized AI development, Mark Zuckerberg’s AI manifesto, Nvidia’s effort to finance GPU infrastructure as an asset class, and xAI/Grok’s improving competitive position.

Full Transcript

Transcript of Anthropic's $2T IPO, Zuck's AI Manifesto, Nvidia's $500B AI Bet, Grok's Comeback from All-In Podcast. Auto-generated from episode audio; may contain minor errors.

Okay, everyone. Kind regards please return to All In podcast, the number one podcast in world. David Sachs and Gavin Baker is with us. this week. We have small team, but a long list of things to do, and we are going to play. How are you doing, Gavin? Nice to meet you. see. Fantastically. Are you still at it? SpaceX aftertaste? It the biggest victory in your career, right? Yes? You know, it was magic moment for SpaceX. There is only one SpaceX , but Yes, quoting Bill Belichick, you know, he to Cincinnati.

Got it. AND I think I have there was a lot to do SpaceX, therefore the history of SpaceX not finished. Yes. Yes, you too so good, how good are your last ones investments in our business, I think. All like: "What now?" Saxon, You are back. As How are you, brother? As Summer is ending. for you? Good. Looking forward to it I am waiting to talk to you. Gavin. Oh, fantastic. Hmm, well, I...I...I will try. stay away from you stay away and don't step on it on you.

You missed. a wonderful discussion that we had an interview- All-In show with yours a boy, Ram Emmanuel. Ram Emanuel was there. Well So what? I heard that it was actually like spicy, you know? It was very spicy. Ram came, as in fully charged, but this is really great interview, and we we'll take anyone, who wants to participate in an interview show with those who who will run in the 2028 elections. You I accidentally saw this, Gavin? Did you see this? I haven't watched it yet. Okay, logged in.

queue. Perfectly. Of course I watched. And basically the verdict regarding Ram such that he is something on sample return to National committee Democratic Party Clinton-Obama. I think that if you I was thinking about Clinton-Obama in the Democratic of the "I'm Donny, DSA" party. This is a great irony, and, I I think it's very similar. moment that happened in your Republican parties where you had so-so civil war inland. And I I hope that modified, moderate, have here is a chance; but, you know, who knows what what will happen to these crazy democratic socialists?

Hmm, in we are not special thoughts about this? Of their opinion about socialists, I would say...First of all, he has an incredible work experience at Obama, Clinton, was mayor of Chicago and ambassador to Japan. So he was great at international relationships, etc. Yes, and in your opinion, he firmly believes that democratic socialists do not have common opinion, and socialist politics —this is a road to nowhere. So he just took it. for this with all perseverance, yes as persistently as And you, Sax. Actually, I hope he...

There would be nothing. better than having moderate or more moderate candidate. I think we we can all do this agree. Hmm, so we'll see. We'll see. It's worth checking. Okay, let's go. let's move on to document that is latest news at the moment of this recording. For those who doesn't know, we're recording on Thursdays, we publish on Fridays. And Prop, for According to the Financial Times, plans to hold an IPO on the amount of 2 trillion dollars. This is obviously SpaceX's record-breaking IPO at 1.75 trillion dollars.

The FT reported on Wednesday evening, that investors expect, that the company will be released stock exchange in October, which in about six- eight weeks. It matches the data Polymarket. Specialists Polymarket say that Anthropic IPO probability this year is 80%, and Brad Gerstner, fifth best friend , also signaled, that this will happen sooner rather than later. Annual indicator, we talked a lot about This is Saks. Let's end the year with an income of 100 to 120 billions of dollars. It extraordinary income growth, which we never seen in Kremnieva valley.

Uh, at rated two trillions of dollars, obviously, you all you can count, but it's 16-20 times more sales, what is only part of it, How much did it cost? SpaceX, Palantir and some others highly evaluated companies. So, quite interesting revenue growth for last year in 10 times. Anthropic still has a top AI model, but Grok, and we today let's look at this, quickly catching up, and so does OpenAI not so good either bad. Polymarket says, that exists 67% the probability that at the end of the year Anthropic will have a top model.

On Thursday it is reported that Anthropic is in talks about buying a startup AI called Dekart in 6 billions of dollars. THERE ARE- uh, their software provision reduces tuition fees and logical conclusion AI, making chips more effective. On I'll stop there. Gavin, if you could would comment, and I I don't know if you are shareholder of Anthropic or no, um, that's why it would be It's great to clarify this, but what do you think about income growth, which we see here, and What to expect from an IPO?

So, a few things. I is not currently a shareholder Anthropic. I would say that discharge Anthropic tensions regarding SpaceX, probably happened a little too late. I usually I do not invest in Elon's competitors. I I don't think that's wise. course of action. This corresponds reality. And the discharge , obviously, was adopted by Elon when he had so much unnecessary calculations with Colossus, that he was giving them away for rent Dario and Anthropic. Absolutely. And I think, what he probably is, what else more importantly, accustomed to Dario's morals and that Dario— a good person, and, like all people, be by man—means be wrong.

So, maybe he did some mistakes in that, how did he decide government issues relationships. But he reassured that his moral principles on at a good level. As Elon said, "They don't work in my evil detector, and in enough of him good detector evil, you know, he, like known, briefly interrupted his conversation with SPF. Therefore, a few things about Anthropic, you know, numbers The rumors are exceptional. We never seen Nothing of the kind. You know, I still think what are those moments in April and May were one of the the most extraordinary moments in history capitalism.

And what I'm surprised, that's what that Anthropic is probably on limits lose share in favor of OpenAI, open source and Grok , and they still are are growing rapidly. The numbers are still exceptional. So, I I think, you know, this speaks of strength that, Why is this so? It it is important that people understood. The pie becomes gigantic. So, even if in percentage, Anthropic loses in percentage, is a real number, and we talked about open source, this is dark tokens that nowhere are tracked. You know, this is important.

growth moment pie. That's what you are "you mean." Here in Why essence. I think AI generally accelerated in July. What was a little crazy about contrast with actions public stock market. Now we have a better view of what was happening. But a few things about IPOs Anthropic. Listen, this is exceptional business. They are really good. have done their job, and they just performed it for model. Their the products are amazing. I I always think about Eric Vishryu. Not I think you ever... ever participated in come on, but you know, I I think he would be a wonderful guest.

He good friend. He said within the framework thought experiment almost two years ago, what, in his opinion, OpenAI will still have great value, even if they lose on equal terms models. Because product, opportunities, acquaintance with user, um, everything this is so important, and I I think this has a grain of truth. But what Regarding IPO, first, what would I just say, that's two trillion, I a little skeptical I agree. We saw this is from SpaceX. If you... struggle for leadership from the left, um, philanthropy, and if you will lose, first, What do you want to do?

it is to expose someone leader on the left in in bad light. So you pour information to the press and you say: "Hey, we're in room, you know, IPO level bar in two trillion dollars." In the previous episode you said about three or four trillions. Absolutely, but I I think that's probably it. , where she can bargain. That is, they are now hesitate. Yes, we haven't seen S-1. You know, I think that if you output company responsibly, as you you do it, you want set a price so that you could absorb blocking because you don't want to install very high price, create a lot volatility.

It very embarrassing employees. You are not do you want someone to got distracted in this moment. Therefore, I think, that they are smart will set the price. Maybe it's two trillions. Perhaps, that's two trillion and exchange for three. We just have to remember that these always leaks coming from bankers who, probably lost leadership from the left. They want leadership looked on the left badly. So, if there will be two trillions, that's, probably means that water testing for the road show has passed incredibly good. AND people, they want to own this.

We never knew these the numbers are unprecedented . You say roadshow probably already did it happen quietly happened? No, no, definitely not. I I say that if they rate it at two, this means that water testing It went well, road- The show went well, and they are very confident in because where is it will be traded. Sachs, in history Silicon Valley you repeatedly noted here, under time of this podcast in 2026, that we never never seen anything like this income growth. Or maybe this continue? AND if this continues, What does this say about the value of AI for corporations and enterprises where Claude is indisputable leader?

Well, I think it can. continue. Although this is a legitimate question, because if you you finish the year, let's say, out of a hundred billions of dollars, which increases by ten times more than a year ago. And, by the way, they increased tenfold per year. year during the last three years. So it grows. exponentially. If such a pace growth continued, they would reach a trillion dollars per year annual profit to end of the next year. And then the question, which you should to put, is: or enough TAM for this?

But also whether enough computational resources? Or enough energy? I I think they are starting. physical limitations. I I really think that TAM big enough. I I think there will be demand. There are so many new applications of AI. Agents right now are gaining momentum. You saw this with the launch Grok bot. All these others companies will continue expand use of AI in increasing context. So, the demand for tokens will just grow exponentially. I I think the question is is whether can they physically satisfy this demand, and also, according to with a note from Gavin regarding market share, other companies see, what they do, and they are more efficient compete to get a share.

Remember, Anthropic made a huge bet on coding, and it turned out very far-sighted rate. I don't know whether it was like that. Through ideology they thought this was the way to recursive self-improvement. I I don't know if it was because they saw Cursor. That's the reason. As far as I understand. Yes, they are watched using Cursor and said: "Well, we "Let's take this business." "And this is very important a note for you. Vertical. Yes, they started to move into verticals, where they see, as their users successfully create products on their platform.

So, you know, I think that's there was a signal for them to do coding . Nevertheless, they were the first to take action they took up this this is very successful. It led to a boom. We we hear that now, when Open AI actually reoriented to coding. They are better are dealing with this now, with GPT 5.6. We heard, that the pace of their growth are accelerating. They used to grow three or four times a year. I heard that for the last two months growth rate now exceed 20% monthly, what if extrapolate this to 12 months, will be 10- at a multiple pace growth.

Hmm. So now you see, that Open AI is growing so as fast as Anthropic, making this turn. We can talk. about whether we think that SpaceX AI will also be able to achieve this. But in in any case, Look, I think that regardless of , will Anthropic be able to to grow 10 times next year, they will grow very quickly, and then the question is because what will it be like market share compared to theirs competitors. And I think, Gavin, it would be worth twice click two at a time points.

First point , will there be a market for these tokens and whether will be able to grow to stay the same equal? In other words , is there demand? I think that all three agree, yes, Of course, there is demand. But the next two points, which Sachs mentioned , are processing centers data in energy. AND This is the question I am asking. I want you to answered. Can they they reach one trillion in revenue? And let's go here let's slow down and let's try, you know, to do some, uh, calculations.

And this is yours. cutting. What is needed from points of view infrastructure, so that they increased their income out of a hundred billion dollars per year to 10- multiple? And is this maybe next time year? So, a few thoughts. I I think that's it. the right questions. On -first, internally Anthropic is very confident in itself. Several people, whom I trust, they told me that Dario said that Anthropic could to become one private company in the world at a certain time moment. Think about it. It would be, and you know , in this vision, maximalist Anthropic vision, is Anthropic, and then there are governments.

And this all. So, there are many confidence. I sure they have more advanced checkpoints points than Fable, in sleeve. They really are, very well done your work. I would, probably underestimated that they are the only private company in the world. Emm, yes, I think so A lot of time will pass, before they they will land a rocket, or, umm, you know, and will deliver your lunch. I don't know. Perhaps there is also in the world other companies. Zipline may want bring you a burrito. Uber may want to to take you somewhere.

Waymo may want to to take you home. But of course, Dario. I can interpret this as negative signal, because that's how it is... arrogance, you know. Yes. This goes into a bit of the SBF sphere. Yes. I would definitely refuse. Dario from any repetition of this to someone. So, don't you think... So AI Jesus is not must claim that he is Jesus. Good . Pride precedes fall. Yes. Yes. The fall will be long, if...if they...I I mean... But no...but Listen, I heard from other people in companies...I heard that when they had approximately 60 billions per year profitability, they thought they could reach 600 billion in 1 year.

That is, they thought that from the inside this year they will increase 10 times. Now they exceeded 80. It seems they don't are slowing down. So, I think, listen, I I think this is the main thing. question, you know, how they can quickly grow next year? Will it go from 100 or 120 to the trillion dollars? Listen, if they just reached 4-500 billion , that would make them the largest software company . I mean, if they...Get to 200 billions–this amazing result . So where does that leave them?

compared with Microsoft Office franchise? Well, where or with the search engine a Google franchise? We we are starting to become equal to these two companies. Absolutely 100. So, in order to to sort out the issue, this good question. It supply and demand. So, demand, you know, depending on how to count it, there is, you know, from 25 to, you know, 65 trillion in work related to knowledge. Hmm, 25 is the number that I used. Me said, um, the manager Institute of Artificial intelligence in one of the three largest investment banks that this is much too little.

Hmm. So, this is the same one total number jobs, which they strive get today, before we have will appear robotics. And then the question is whether it is replacement of the worker force or acceleration growth? Hmm. So far It really looks like this. accelerates growth. Today there is, you know, more vacancies for developers software provision than It was a year ago. ago. There is a greater demand on programmers, and this approximately in epicenter of the zone explosion. There are some evidence that for very young people, which immediately graduating from college, Yes.

Uh, for, you know, maybe there is a negative influence. Hmm, but in general saying, it has to proceed from one of these two. And I just I think this is, you know, we...hmm, it's better for everyone if this happens thanks to acceleration growth. And these companies now have achieved this scale, that if this will take place thanks to acceleration growth, we will start see it in national accounts during the next 18 months. I'm going to...Okay. You want to go to energy part " "Supply". Yes. Yes, and listen, this is what where is the rubber meets with proverb.

How am I, hmm, you know, people who are building these centers data processing, you know , the best way understand this- look at that one series "The Man from "And it's like Billy Bob Thornton... There you know, time Permian Basin. Permian Basin. You are you in the zone risk, you know. You are you really worried through drug dealers . You know, this is the tip. spear. It's like 110 degrees of heat. Ahem, you need organize thousands people in these remote places. These are atoms, not pieces.

And this really, really hard. Now, America, we are really very good when we are preparing for something something like that. America and capitalism is very solve these well problems. You all are you talking about that final limitation there are these turbine blades. AND you know, they are made only at two factories in America and Europe. AND every factory is, you know, gigantic building, and there is some 40-ton press. But, you know what? Now they work, you know , 24-hour shifts. They get, you know, they receive a 40-ton press.

The necessity, in essence , is the mother of innovation, and nothing like capitalism, won't do this... happens. Actually, boom, supersonic aircraft creates a new one jet plane for, you know, supersonic. And SpaceX has one on deposit. These jet planes, he like this: you know what? We first let's build manufacturing business turbine, because we have one high demand. AND then, Elon, I understand , there was a message in news that he bought company with turbine production, and he did it personally, because it tells me about that he just wanted to move quickly, or wrong?

And in addition to that, you know, I know, leftovers for private jets now extremely high. And big, most of The reason is because people remove the jets engines from old cars airplanes and repurpose them as turbines for power processing centers data at 100%. It madness. It madness. It's like big business that grows very quickly. And then, you know, you have there is Caterpillar, you have Commons, you have GE Vernova, you have Siemens Energy. They all rapidly expanding power. So I I think the question is supply, you know, Billy Bob Thornton is here.

, in America, you know, we We will work. with these companies and installation of these turbines on site. It policy, regulation, and that the fact that Texas, you know , announced, you know, that, in my opinion, there is the largest audit, that's what it's about you are talking. Governor Abbott announced this week, what is he going to do to spend energy audit. Not that he slowed down the process, not that it was moratorium, but what everyone must pass energy audit. If you bring own products, everything It will be fine.

But if in you have no solution for this is what you want connect to Texas networks, the wonderful state of Texas, you, perhaps, will need to pass some audit because we don't want the whole the network here is out of what's going on every January, when everything freezes. Yes, and now I'm telling you not much about AI good. He speaks. that the processing center data takes all water because someone made a mistake in book 3 years ago, and he was wrong, I I think 100,000 times amount of water on one volume of data.

use of the center data processing. Therefore, it's simple wrong. They increase the cost electricity, which just wrong. They reduce cost electricity. Wall The Street Journal just wrote an article about a city called Ellendale, where, as if it were was a city that was dying, was center built data processing, this revived the city, tax revenues increased in 10 times. This is the best, what ever happened. But no one doesn't tell these history, and to people necessary. And I think, You know, we are many. times discussed in show that Communist The Chinese party has a very active campaign against AI, against processing centers data, because that's what what is good for them and bad for America.

AND someone should tell the truth about AI . One of the best phrases in the Bible are true will make you free. Yes. Cannot be released you, if no one will tell the truth. Yes. And so. Yes. I mean that panic around processing centers data is evidence the power of the media to create hoaxes and literally hysteria because of nothing. Everyone aspect of the story about data centers false. This is...a lie. As you said if they do not use community water. Water use completely manageable, less than the field for golf.

Water recycles. This reduces expenses for electricity, if assume that you have data center, which independently produces electricity, therefore what are they will sell surpluses back in network and will pay expenses for network modernization. The real reason price increase electricity in " "blue" states consists of decarbonization and everyone else, you know, " "green" rules, which complicate production electricity. Um, What other accusations? I mean that they... Well, noisy. pollution...Yes, noise pollution- it is legal. So no one in good faith not obliged to put him on someone's backyard.

No one is healthy today's sense doesn't do this, but people did it in this country. Yes, people have been doing this for a while. Simply use good sense. Yes, common sense. You just use common sense, and no problem. Um, the idea that need completely to prohibit construction of centers data processing, simply stupid. And the communities that they are accepted, seen, that they have the best schools, better services, therefore what...they get a lot of income. Well, then, to yours. thoughts, Zach, when entrepreneurs something take, we suddenly we see how sunny energy, batteries, nuclear energy and more clean energy used for powering these things, and this demand over time will reduce the prices of these services.

Solar panels will become cheaper , batteries will become cheaper, therefore what are they will scale much faster than that good for humanity. It Simple, people. It very basic. Green materials are not can to scale in short-term perspective. All this runs on gas for new centers data processing. It the only way to achieve the necessary scale to the next a year or two. But in any case in any case, listen, Do you agree with this? Gavin? What I think is solar energy it will be sooner viable, but, you know, this is obviously does not mean the appearance natural gas.

Yes, I mostly I agree. I will say, you know, Elon builds new factory with solar production energy worth 10 billions of dollars in Texas. For...Reason. My home state. AND You know, I think that the world at last will work on solar energy, but for a while we we will work on natural gas. IN There are many of us in America. natural gas. It clean fuel. It the greenest carbon fuel. AND this media hysteria regarding processing centers data makes such disservice Americans with low level income, Americans with small towns, often American minorities.

It's simple. so good in everyone relations, and the media they make such a bear service, but no one takes the other side, and someone should take the other side and tell the truth. Yeah, well, it's, uh, I...I don't I know who it is now. relies on the media in my facts, but I I think there is a reason, why were we number one four in the world last week, so what, listen, we is actually part of of this development. We we know about it, and we we can explain to you this is a little more concise.

Noise pollution– This is a problem. Air pollution –that's the problem...that one gas. Keep these things. away from people, like Sachs just said, and in there will be none of you like that serious problems. THERE ARE a small problem, if you place them in a certain place in on a large scale, you can you raise the temperature of this valleys or anywhere where you will place it on a couple of degrees. This is what what is it worth control. That's all. completely controllable problems. And if you really want to get carried away by water, please, for the sake of God, stop it eat almonds.

For example, you know, the golf course is not must use billion gallons of water annually. I will say that with demand side and pricing if you will open this a short note that I have here, I think that Anthropic growth rate will slow down significantly next year. And this, obviously on the verge opportunities. They will never reach trillion in revenue next year. I I think that perhaps they could, you know, to triple, to reach 3 or 400 million. AND the reason is because we watching startups, as well as for some big companies that are now accept GLM 52 and Zuckerberg, who...We in a moment we will move on to release really wonderful models from open source.

Just take a look at GLM 2 prices compared to Claude Opus. And, you know, use these products 90% cheaper . I use them. Founders in which I I invest, use them, and more and more corporate America perceives these decision. Just them a little more difficult set up, and this Inevitable, friends. AND if you think that the point is not that corporate America will not accept open code, you did not return attention to the last 20 years when they did exactly that. A 20 years ago was big argument.

My God, corporate America will never trust open source. AND corporate America said that we are not we want us to deceived or to some big one technological the corporation held us hostage for the sake of their own technologies, in what they are experts. So open source is safer option for us. And this is what is happening today in corporate America and within startups. This will be very real. Gavin, you look as if you were jumping from there. your place. Yes, no, no, it's wider. question. Like, this definitely good for corporate America.

Exactly, open weekend code is good for the world. Exactly, it's good for America. But you know, Dwarkesh posted something. Dwarkesh, hmm, who? Is there anyone, a podcaster, thoughtful guy, close to the world artificial intelligence. He, um, he published something that really found I have a review. Finally, he said, that AI Anthropic in its own right constitution programmed do what Anthropic considers it best for humanity. And let's let's accept the judgment Ilona is like gold, and he is right. Its detector "don't be "evil" didn't work. Yes.

And they really are good people, and I I actually believe this, who do what considered good for humanity. I actually I believe in this. The problem is because, you know, there is always unpredictable consequences, and the world chaotic and unpredictable, and I I just think that wonderfully open source code is that ...means that we have will be rich variety of artificial intelligence. We will have variety artificial intelligence, and it's a good world for people in America. We are not we want one, two or three dominant models.

And therefore, to things, you know, not for so that, you know, not I would always like to to talk about Grok, but it's great. I think , what is Elon...History kindly will judge Elon for that Grock dedicated self-searching objective truth. Objective truth. No, the whole point, Gavin, is that if you did bet, this means that you used yours intelligence and your own cleverness, so that to choose what, on in your opinion, is the winner, that actually, I think there is extremely accurate. AND I think Elon, um, clearly believes in open weekend code, and he believes that this is a technology that should be shared for the sake of her benefits.

And in fact, it was his first step in this matter, which consisted in supporting open artificial intellect, and manifesto open artificial intelligence was rewritten this week. The second topic, written by Mark Zuckerberg, enough paradoxically or ironically, in...Essay for 6500 words. It is called " The future for all: the path to positivity "future AI". He laid out extremely optimistic AI's worldview, where is it, as it is now in Open Facebook AI 2.0, will be released open models source code. He will create an agent for everyone. He will do its free.

He will provide everyone with world of one's own mentor and unlimited abundance . He picked up the crown. the abundance that Sam Altman lost when moved to private company. AND he says that superintelligence is invention, not automation, and what AI safety must to be based on balance of power, not on the only one centralized intellect. I I suppose, Sachs, you read, viewed did you receive illumination Zuckerberg's manifesto , and do you have any thoughts about him positioning? Yes, I mean, I I read it, and I found myself nodding nod in agreement with more of it part.

Well, I really like it. I like the idea, which he expressed quite early, where is he noted that many who is a loser, and I I think he's like pounced on effective altruists, and then "I'm fed up," he said. Dario, he said, listen, why are you like this hurry up to create the future, Which one don't you believe in? It such a dystopia. You know, if you What do you think this is? a bad future where everyone will be left without work, and artificial intelligence will fail and will create all these security problems, Why are you so excited?

its creation? So, I think he indicates fundamental contradiction. On this the question is the answer, which one actually is that there is two answers. On- first, they can to try to create a future that doesn't believe, with, let's say so , selfish considerations. This can to be profitable. Although I don't think it is the real reason. I I think most these guys— missionaries, and they believe that this the future is dangerous, and only under their enlightened control humanity can be protected. This follows directly from vision of the anointed, books by Thomas Sowell, written 30 years ago because where intellectuals throughout stories thought that if they maximum expanded, then they can to direct society in more benevolent direction.

And it always is had the opposite effect. This always led not only to empty ones or unfulfilled promises, but also became an excuse for totalitarian schemes and state authorities. And in fact Zuckerberg continues to say something that, on in my opinion, it concerns this. So immediately after he expressed an opinion about that that you know why you hurry up and build a future you don't believe..." What do you want? you don't want to He says that the idea that AI so dangerous, that the only safe one the way is extraordinary concentration of power, seems problematic for by its very nature.

Historically, so that the hope for that which is absolute the government is benevolent will provide humanity, if it is enough educated, not led to safer positive results". So, I I think that the essence, which he means, is that the main frame of the whole discussions about AI consists not only in openness or closedness. She also applies centralized against decentralized. And regardless of whether do you think centralized the result is correct or decentralized , I'm on the side decentralization. How can you not be on on the side of decentralization ?

I mean, this most obvious argument. Types E. Types EA are not...AND by the way, well, I experienced this when arrived at Washington. And, you know, all former people Biden just moved to government Anthropic affairs. Me too dealt with people from analytical center, which are part of this, like Click, and their point vision of AI was to because he is in his inherently dangerous, and what should we do centralize control over it in government. It must be, you know, just two or three companies, and they should essentially work together.

It must be converted to the cartel. Remember, Mark? Andreessen had meeting with these by Biden officials, where he stated: " Don't worry. compete". They deny this characteristic. Maybe they don't absolutely correct used these words, but I think that Mark was right in to his interpretation heard, because I heard things like that, and they really think that way to control what they are essentially considered malicious, or at least very dangerous, was like this: first, you limit everything The United States, not to export chips, do not export models.

Essentially, you keep full control over this in The USA, which, frankly, is impossible a dream, because the rest of the world will receive AI. The idea of ​​"you" "you can, but, but, but" was something like part number one, and then part number two was that, to create something on like a commissioner from nuclear energy or something like that, so that cartelize an industry , and then you have deeply rooted ties with the government. You actually have corporate merger and state authorities. So, that was it. extremely centralized vision of the future.

I think they are to some extent still in They believe it. Obviously, Zuckerberg listens to the All-in podcast, and, obviously we have free space next week. Mark Zuckerberg, or Can someone contact me? with him, come to next week. Thursdays will continue approximately from 10:00 to 11:00 in the morning at your time, regardless of whether Are you in Hawaii or, um , on Tahoe, we we'll send you Starlink, and you can do it from your email mail, guaranteed. But it would be great, if Mark told about what Zuckerberg allowed Kevin to attract, and, Well, let me.

just to summarize. I I think Zuckerberg goes to this, with what I OK, we want maximum empowerment and opportunities personality. This is the way solving this problems. We want decentralized results. We we want open models, as you say, Jake, we want data sovereignty. Look, they can also be these own models Frontier of anthropic and open AI. But we we want to create decentralized the result that expands possibilities individual people, and I I think Zach is on this. side, and I'm with this I completely agree. Gavin, I suppose You read this.

Yes, are you? I just realized that I didn't read what you have. there was no time for 6500 words? No, by the way, if you will write, by the way, if you write more nine, he will have secondary non-fiction a book that has approximately 60,000-70,000 words. So a businessman with Harper Hollis will call Zuckerberg and will start the other eight parts of this Manifesto. I read it. I agreed with him, and I would just completely agreed with everything that David said, and I would reduced it to the fact that I think that anthropologists and people from movement of effective altruism is considered that this technology too dangerous distribute.

And then that Mark Zuckerberg, and I think Elon and Jensen believes this that this technology too dangerous centralize. AND history has spoken word. And when there is a choice, always better distribute and decentralize. Actually, I can't. come up with none counterexample. AND just forget how American, but how man, I want AI, who takes care of me. Takes care of me. How about personality. To be a person. Sovereign the personality was key point. Yes. Yes, in my opinion about what is good for America, I don't want so that Dario Amodei or Anthropic was decided, what is in my best interests.

Hmm. I just fundamentally I don't want this. I am not I trust AI Jesus. I no...that is, AI Jesus seems great a guy, you know, although It's funny that someone said that his cursed with calm smiling face, and a little amazingly listen to him speaks about it terrible vision humanity, happily smiling. This is funny. He has literally this smile when he explains the scenario to you with the Terminator and that, that only he can save us, and that even he doesn't know that there is a 20% chance that he can not save us all .

Gavin, I like the wording that that centralization too dangerous, and I I think that another one is very similar moment concerns you thoughts about competition. So, I I think that the executives associations believe that what competition will create a race to lower level and lack security, while I I think that competition reveals best. And that's why they they think that she reveals the worst. Now, listen, there is examples, probably, both, but essentially they want centralize control and reduce competition. We we want decentralize and to have more competition.

And I think that I want it to be in constitution. I have the right to bear arms, and I think this is important. . I grew up in Texas. I I think this is mine. kind of line of defense from tyranny. I love Texas. Gavin, when are you Will you return home? I will be back soon, I promise. Good. But I want to be right. have your own AI. I I want the right to my own AI, where, you know, his values correspond to mine. Not someone so good, How can he be?

Right to wear weapons—good analogy because we we allow you to have weapons, even if you you can use it abuse. Weapon used, to kill, right? ? Whereas artificial intelligence has much more creative and much more useful using. He useful. In essence, this consumer technology, which everyone wants use. So, you know, if we let you have weapons, I agree, What should you do? allowed to have artificial intelligence. I'll correct you, uh, Sax, now that you with me in a wonderful in the state of Texas. Weapon also helps you to feed the family and protect you from evil forces in the world.

So, this is virtuous use of weapons. Yes, I understand. It self-defense, but I I say that you don't care. would use weapons, to build... Use gun to kill or injure a person in for self-defense purposes. I I understand. I don't mind. weapons. I think this is a hoax. Dario. If we, if we will think about it with these, you know, effective altruists, you you can search for this, people, but they think that, in essence, they are going to earn all the money in world, create one company, car for printing money, and then they will save us everyone.

This is huge. delusion of grandeur. You know, technology anyway will spread. This idea about what is possible keep the genie in bottle, absurd. But, by the way, to relate to this as I mean, I think, Sachs, how are they they perceive it, they are considering this literally like superintelligence and a nuclear bomb. And if you will convince yourself, that you are building a nuclear power plant bomb and you build The Terminator you you can't do it yourself to control, and this superintelligence, which will create a billion Terminators and a billion bombs...They literally cheated yourself, taking POD or LSD at Burning Man, what is it?

reality. This is not reality. You can turn it off at any time What time is it? They suffer from mania Greatness, period. This is not nuclear bomb. Well, you're right. regarding this. When I first time I came to Washington, it was dominant analogy, that they and their related analytical centers, their people in their group promoted the idea that that AI is similar to nuclear weapons, and we Commission needed atomic energy, global, so that regulate this. Quite a vain dream, which will never happen, but you are right that they used this analogy.

And I think what is the reason why they use this analogy, is in that you don't have the right to have nuclear weapons, create its nuclear weapons. IN you have the right to bear arms, but not nuclear weapons. So they wanted to put AI in this is a bucket. But, how Jensen said, to no one, to no consumer no need for nuclear weapon. To everyone to the consumer, to everyone Business needs AI. This is fundamentally different case. This is consumer technology. It consumer technology, before it happens military technology.

And this one the idea that this is, after all essentially, something like nuclear bomb, just absurd. I mean, justice for the sake of, Cyberdyne, justice for Gavin's sake, Cyberdyne Industries also thought, that create good technology. So science fiction literally soiled people's brains. If will be something about the Terminator, I want my own. Terminators, which I am I like it, okay? Yes, open source code, Terminator. Oh, but yes, but simply simply put boil everything else, as if I were an American, I patriot. If we let's ban open source code if we let's allow this control to be centralized, we lose AI, we we will lose geopolitics , we will lose everything on benefit of China.

That's right. simply. So, if you American patriot, you must to be for distribution, democratization, decentralization, individualization of this technology. Gay, To Jake Halpern, who Do you think? What are you? What do you think about this essay? Well, like you, Sax, I am all I nodded my head the way, because it seems he ran out in front of the parade, took up the baton and said: "I agree with best wishes here. And the best idea —it is open source code, expansion human capabilities, and these are not nuclear technologies".

I just I want to mention two more. moments that, in my opinion, opinion, remained out of sight, because he tried to cover many topics. Firstly, I want to give it to him. due to him a look at itself dismissal from work, for he says, “Listen, You know, it's possible. dismissal from work." But he says: "People are afraid that automation will overtake growth of individual persons, which will lead to job losses, and then it will come a difficult period when people will learn new professions. I have meaning that we all we can agree that this causes concern.

But there is no rule that artificial intelligence should increase automation faster than it increases people's capabilities or demand for new skills . Recent statistics show, which is perhaps more it is likely that growth people's capabilities will be equal to automation or will overtake her. I I think it's true. successful formulation, and this is just another one a way of saying that said, uh, Jensen: " Hey, you won't lose. their work through AI. You will lose her because someone with AI." "And this The game is on the field now.

You know, every time I I use this technology, three open new business opportunities, and I I'm creating two more. representatives. And people performed at five o'clock ten times more work, if they perceive this technology. And this big question if so it will be, and whether Will people do this? . The second thing I think quite interesting, this is, that he plays on the theme indexing rights anyone's content and learn from it. The quote he gave here says: "All AI models come from human knowledge." « Okay, fair enough.

Hmm, it's pretty simple. commonplace. Some tried to present distillation as harmful, but I think what is important to protect the principle that you can to learn from anything, what can you do watch. Others in words, he preventively says: "I I want to scan world data. "Well, in I am there for him. message, namely: Zuckerberg, explain, please, how is this the concept is compatible with the fact that you filed a lawsuit against every startup, in many of which I invested, which when -ever tried index or use social graph in Instagram or Facebook.

They have hundreds of lawsuits lawsuits, where they filed a lawsuit against people for trying to distill, you know, who are the main ones Instagram influencers ? Who is in my social graph? AND they blocked everything it. So, here's a little bit hypocrisy, which reminded me, Sax, about our discussion in previous episodes, where Claude tore old books and many times distilled the debate about distillation. Let me just come back up to this point about that which is huge the irony is that Dario mostly doesn't achieved his goal Washington during the time this administration, because I think that if he had achieved his goal, if he had done my dream about this highly regulated artificial apparatus intelligence, I don't think, that this company would be ready for extremely successful IPO.

Why? Because they have full right to charge this a huge prize for tokens that…” Jackal, that's exactly what you are. meant earlier: you are thinking, how much has it become possible because of all these open models that appear and offer tokens for much less money. All their pricing power, and the prize they can charge, is that they are for six months ahead of these open models. And these six months would be eaten in an instant, if only they were subject to approval regulatory organs. Dario himself continues to write these blog posts where says, "We need the FAA for AI.

"You know, how long FAA is required to to certify and approve a new one model, in this case for a glider? Minimum 5 years. I mean, if you want to change a table in your plane, it takes 8 years. Do you know how much? time it takes the FDA to approve a new one preparation? Years, OK? Are you going? apply this regulatory framework to releases of AI models? End of the game. Forget about it. Yours 6 months and will pass. Your model will be commercialized. On Gavin's opinion, Chinese competitors will catch up.

And this bad for United States, this bad for our competitiveness , it is bad for AI races. I have on I mean, obviously we don't we want China to us won. But, besides geopolitical measurement, it would be terrible for Anthropic as companies, because, again, all of their pricing power depends from whether they will stay leaders for 6 months. If they they will make mistakes someday, hold on. If they will make mistakes someday and will lose these 6 months, how will this affect income? That's right. So, now Look, I actually..." I think, their pricing policy may be enough stable, because, in my opinion opinion, people are ready pay.

Not all market. Not the whole market, but decent part of the market is ready to pay a premium for true pioneer intelligence. So, if the richest 3%, as at Sachs, if 20% the richest are ready pay 10 times more, they win , isn't it? Simply... But, as I said, this Apple vs Android, right? Yes? Most of the market ready to choose Android. It cheaper, but there is a significant part of the market that ready to pay a huge prize for Apple, because it's the best experience.

The same if you are a company that works on competitive market, and you absolutely want the best of competitive or other reasons, you pay a premium for true pioneer intelligence, but Anthropic and OpenAI are on this hamster wheel. They must to remain leaders, to justify this prize. And, by the way, I I think they can, and they think that in reality their advantage is growing because they ...on the way to recursive self-improvement. I just want one more time. to indicate movement direct connection. Zach had his account in Twitter from the very beginning beginning, when tried to buy company under a nickname thinkd, finkd, that seems to be was his nickname in colleges.

And for the last 30 or 60 days Nikesh from Palo Alto Networks, a friend of the group, started tweeting and use long accounts. Jensen opened account X. He started use long accounts. Thank you thanks for subscribing, Jensen . Uh, and then, uh, Zuckerberg activated your account and goes straight connection, and I want just to point out that Zuckerberg before, me I think I allowed others people to shape their own message and send it to market. And I think that he grew up and decided: "I will write your own message ".

This is a message, I I think it was written him, probably with with some help, but I missed it through the detector artificial intelligence. This is 100% human, individual, words Schumann. I think this is it's fantastic that he now moving forward forward and says that this potential...problems in the future, and it was It would be great if he did it with the girls -teenagers, influence social networks on children and addiction from Instagram and Facebook earlier. I would give a great credit to him for the decision that I will leave directly, I will write mine essay, I will tell you mine truth, and, man, X now it becomes exceptional in these dialogues and these messages.

It fantastically. Yes, that's why I'm completely I agree with everything you said. they said, and I want return to a few things that said Sach about calculations - about how open source actually does Anthropic is more valuable, and computational people efficiency- this is a bullish argument for people. But I think that what you said, very important. If you public figure, if you are a general the company director, and you don't have your own votes on X, you are you in a group risk. It's that simple.

Risk of what? Unpack this. You, after all, are you in the same department, which is true will make you free. The only person who can someone say personal truth, is she herself has her own in authentic words . And, you know, obviously, Mark...Zuckerberg has super-voters shares. He never will be excluded from the Goal. But there are many general directors who think, "Oh, you know, my corporate department of relations with by the public protects me. They- my voice." "Well, You know what? You know, maybe you will go to pension.

Perhaps, next general you are not the director you will like it. And you will be called before Congress. And I think, you know, in some moment, you know, Democrats, you know, talked about what in they will be in order daily accountability. And they are going to persecute everyone. AND I mean, I don't I'm sure that this is good message for the Democrats. But nevertheless, if you don't have own voice and own brand on X, you are in risk group. And the mother own voice Yes, it is.

protection. This is strength. It armor. And I think that this symbol. Yes, it is a superpower, and I...I don't know about you, Sax, but I am much I like it better. Jeremy Strong era Zuckerberg than era Jesse Eisenberg. Not doing these MMAs? I see there's a video of him. for something like barges or something like this. Floating platform. What is it? Barge on the lake Tahoe? Yes. Dude, dude, I would be careful not to don't talk about him bad things, Jay Kele. No, I am like this, I am like this.

Listen, me too...I started from a clean slate. I like this, I am in Zak 2.0. I hope so. he will like it Jay Kele 3.0. Come on, Zaku. I want him to was on the electronic Foley with American flag every year 4 July. And I want this broadcasted in live broadcast. And if he will fall, it will be for him to the maximum auras. If he does again will get up and write this down, and will do it live. Yo, he has to do it. this is on Kauai with bull sharks.

Give I want to see how he is chasing him fins. It generates aura a thousand times. Well done, Zaku. And we need more such authenticity. General Director , be direct. 100% dude, be direct. Protect yourself. Yes, and be yourself, dude. And he is like that, as if he is himself, and He is a good boy. It's okay, people. There are two points which, on in my opinion, really important. Open the source code does Frontier tokens more valuable. So, let's say so, let's Let's look at an example.

Let's say Anthropic, OpenAI and Grok have an IQ of approximately 250 or 200. Well, if you You have an intelligence of IQ 200. or 250, which can to manage this intellect with open weekend code with 150 IQ. It actually does intelligence with IQ 250 more valuable. This is because he can transmit task on outsourcing. And so only if we we are going use an analogy with Manhattan project, you know, everyone talking about Richard Feynman and, you know, Who is that guy? managed all this? AND, you know, there are these 10 really smart physicists.

Who the guy who did this led, uh, which one said: "I became death." "Hmm, who is that?" the guy who was driving Manhattan project? But that's enough, Someone has to. Oppenheimer. Yes. IN you were J. Robert Oppenheimer and you and there were 10 of you, you know, actually laureates Nobel Prize, maybe 30. But them you know, it was necessary , 2000, just, you know, 10 000 best physicists world. So, having more available intelligence, this actually does the frontier is even more valuable, and I think that very plausible the result is that, when frontier tokens will be from 65 up to 85%economic values.

Um, tokens from open source are approximately 80% of the volume, and that's good for everyone, and everyone win. Tell me, Gavin. Baker, what is the income? will have Anthropic until the end 2027. What is theirs? income in 2027 if The year 2026 will end exactly at level 100 billions? What is theirs? income in 2027? I I will take it upon myself. management. Over what? A trillion? grades. I am not going. undertake management". more per trillion. Okay, I'm going. bet on 500 billions or more less than Gavin Baker.

How about this? I I'm going to put more. Or 500 billion? No, I'm going. put more. Whatever it is Wall Street consensus for the year 2027, which I do not I know, I'll take more. by 25-30%, assuming, that this campaign is regulation is not will be successful. I'm taking a risk here by saying, that the underwriters will receive a large income. Okay, congratulations, Ben. Give me mine. respond. More than 500 or less? I think between 4 and 500. I I think from 4 to 500 billions.

You are less than 500 in zone. Good. as realistically and achievable. Can I still something to say? Wait, Sax, you less than 500 billion, Saxon. I was actually going to say from 4 to 500 billions, because output ARR on next year seems very comfortable. So a little less, good. Perfectly. For two Less. Yes, and the only reason, You know, it's not higher, because I just...I...I I think that somehow way at this stage you are facing physical restrictions. Yes. But maybe not, you know, maybe computational capacity will be enough.

I mean that there is also competition, ago I think that the next year they will reach 400 billion, and they are going 4X 4X 100 billions. Simply let it fall. It's going to be 120, so I think what does it mean that 4X it will be more like 500. Yes. I mean that it's a kind of total/ less actually... You know, I think that total/less actually amounts to 500 billion weekends ARR for the next year. AND I think they are think that they will make total/less.

We talked about them. at the level of 10 billion 10 months ago. 10 billions to 400 billions is 40X, Gentlemen. Will they be thank me for that that I didn't help Daria to shoot oneself in the leg? I think they are must...He literally said: " He put two "guns to two." ankles, and he just emptied both clips, and you They pulled out the bullets. You They They should give you 1%. I don't really think so. that the bullets were elongated. I think he held them aimed at the temple.

Oh, you ask if he Mao? He played in " Deer hunters? and Sachs pulled him down, and he just shot himself in the foot . That's where it comes from their hostility towards open models, I I think that if they are going slow down, they should also slow down open models. Otherwise they become commercialized, and their business disappears . However, the United States do not have jurisdiction over, what is happening in China, and China is not was going slow. It I think I always have been.

a huge disadvantage in this whole concept. These guys from EA thought that they are going to introduce global management calculations. Somehow they are going to get some global agreement with China. Complete a dream that never comes true will happen. If we Slow down, China will just race forward. I think it was always good create Parisian agreements. 2.0, and it's not It worked. So, Yes. Slow down, boy. Can I just say one thing that, in my opinion, opinion is also important regarding open source, and why I think AI will be useful for people.

He will be useful for Americans. This is what if we don't have enough computational resources, we are talking about what computational resources slow down growth anthropogenic development, because literally we just, you know, we're running out energy. We can't apply physics. We are faced with physics, and how do you you know, restrictions planetary scale, and I think that this is reality. just think as far as we are energy efficient. That bring our brain closer to our brain, to you need a center data processing, which consumes energy million American of houses, working, probably from 6 to 9 months, and then you you get our brain, and then every time when do you want to answer for what you need question, you energy is needed approximately a thousand American houses, okay?

That to function on the levels of our mind, you know, we are smart, we smart, but, like, and all these numbers will be... better. But ours capitalism is so energy efficient, which will be found application for the most effective in computational in terms of intelligences on planet Earth, and this human intelligence. Uh, that's so good. said, because if to look at what Cerebras, Groq are involved and Nvidia, they do conclusions 10, five times a year, and you're here chip master, Gavin . But they will be in five to ten times more effective every 12-18 months.

Then look at the models, and, by the way, congratulations, you Zuckerberg on ass. Ask on MySpace, ask Snapchat how it is- Zuckerberg's mother rearview mirror kind. It's literally a T-Rex , which comes on advanced models. He- a beast that copies, imitates and competes. He will be a hero. open source with America. He will have the best model with open source next year, I I guarantee it here, on All-in, I can take, I can take less. I want And here's another one...There's more one American hero. Good.

Jensen Huang, and how I I think you can see Nvidia at Open Source 2023. Regarding this. I heard feedback. He says we don't we are doing Open Source business. I have on I mean, we have Nemotron, and we cannot provide you have a full stack, but one of hyperscalers, to things, Gavin, one of hyperscalers, told me that they just done nine-digit double reverse Lindy, having moved from Frontier models for 100 billions of dollars in infrastructure, they said: "This will be Open Source. Not Frontier.” So people looking at this and they say that perhaps time to move on to some Open Source models.

Good, Listen. All-in Summit will take place from 13 to 15 September in Los Angeles Angeles. Past we talked about this week to you about giants: Jensen, Satya, Jared, Gwen. I have on I mean, this is everyone's club. in first place. But Here, Steve Hilton. Running for the position of governor. with California. He will come. Spencer Pratt . Will tell about his unsuccessful performance in Los -Angeles, but successful in that he captured the imagination people. Abigail Schreier , author fantastic book " Bad therapy, will be with us from very special guest- surprise.

And, hey, Bestie Gavin, our boyfriend, bestie gavin, may even appear . And my boyfriend, Blake Scholl, General Director of Boom Supersonic, which I mentioned earlier in the program. AND this year we let's transform appearance temple on festival experience from you, for you, center of excellence technologies with demonstrations artificial intelligence and robotics, and also a gift set. Hmm. Set gifts, as if you celebrity. Merch.com created for you unbelievable comprehensive set gifts, wellness set. WITH red light therapy for face and droppers, if Hey, you're late, Saxi-poo. This is sometimes happens.

Not skip these important conversations. And, to things, we are going announce everyone DJs. and parties, that's why Friberg is does. You are this will amaze. Submit application on summit.com today. Boldly submit an application. Gay, by the way, about ours Jensen's boyfriend, the sixth best here, in the group. Nvidia collaborates with Goldman, BlackRock and others to to attract 500 billion in calculations based on artificial intelligence. Jensen made his point. vision in article X, as we already mentioned. Jensen is now on X under called "Calculations" based on artificial Nvidia's intelligence is becoming investment grade assets".

Companies will create independent platforms financing, raising capital third parties. Idea is to consider graphics processors as financially viable generating asset income, as possible securities, secured by a mortgage , or hopefully not with the same result. Here's how this works on the base equal. Instead, for the company to pay billions for purchase chips upfront, they borrow money. They buy systems on based on Nvidia and use them, to start to receive income, that is, they rent calculation. Uh, yeah then they can use cash flows for loan repayment.

Nvidia's role is to mediation, connecting customers, who need calculations, with creditors. It is obvious that GM, Ford have their own, uh , financial departments. Gavin, what do you think? about this statement regarding transfer of funds and disturbing the concerns we have spoke in the same podcasts a year or two ago? Well, I think that here what is important is that in essence happened, for example, Blackstone, KKR, these firms, Goldman Sachs, they would do this didn't do it. They wouldn't did that episode of CNBC with Jensen, if not believed that calculations on graphics Nvidia processors have financially suitable asset.

And what is here? really smart, yeah this is what you get , you know, some of the smartest asset managers in world, you know, which ones they charge 20 dollars for many products to essentially check this market. And Nvidia is matchmaker, and what they essentially they say this: " Hey, our calculations, because they are like that flexible, will have long enough service life so that you could finance them at lower rates, than other, um, types calculations. "And I I think, and I think that this reasonably.

This is good for everyone. And it's interesting if look at the basics architecture of three, I I would call it big Chinese models with open source, and, maybe even to single out a few or four; you know, if you we have, um, Qwen, if you we have Kimi, if we have Deep Sync, um, and then in we have GLM, they actually are developing very in different ways. AND this architectural variation works on in favor of Nvidia and on the benefit of graphics processors, because this means that you need these more flexible computing, to be able to to finance it, to have to think that you you can have a long life.

But what Nvidia does what I think really smart, that's what they say: " Hey, we're going. help create this market, develop this market. But they are going to provide two very important functions." They'll say, "Hey, Blackstone, Goldman Sachs, KKR, Apollo, BlackRock, um, I'm someone I miss it, and I I'm so sorry. Their six. Contact us to us. If you have deal, you can offer it to us, and we will provide a guarantee residual value. And I think this one the remainder that is still will decrease more cost financing, maybe to be included in underwriting these companies.

And that's essentially it. means that they they say that in 3 years, 4 years we we guarantee that these graphics processors it will be possible to rent for a certain period rate. And then the risk they carry, it is a risk between this value, um, and market rate. AND everything is going straight now above. Nvidia has the best telemetry than almost anyone, on demand and offers computational techniques. So they can place this in very, very a smart place, um, even more to do it financially profitable.

You're taking a risk, you know. they only carry 25% risk. The world is one of the smartest in world. underwriters, you know, from the point of view of reputation. What could go wrong? Gavin? I remembered about mortgage securities papers, etc. AND, obviously, the term services is key question. Various hyperscalers said : "Hey, 4 5 6 years old." « Amazon, and took place big debates. But that, what we saw, concerns you opinions on models with open source, which they can be customized for work on the oldest equipment.

So, maybe we will get 6-7-8 years of them. Is this the key to this? than this. Cloudweave said, that they rent Ampere economically at favorable prices today for the year 2029. So, Ampere, which came out in 2020, will have 9- one-year service life. And I think that's exactly what Nvidia sees, you know, this like old people American cars, their sent for border, just from the idea that these graphics processors can go wrong? Here I will tell you where it is. it could go wrong theoretically, okay.

In theory, simple so that we...we...we...we... became... The greatest risk for I'm not from the demand side. The biggest risk is that in there are too many of you calculations and you will receive excessive build-up. And so just like we had dark fiber after dot-com crash, if You had a dark... Graphics processors– it would be a disaster for everyone, especially if you were building your own computer infrastructure, waiting for the spot price from 30 to 50 dollars per watt, like, you know, Elon said, or wrong? So, if suddenly too much many people will compete for to ensure these calculations, and now there will be a surplus offers, and the market will collapse, it will be risk factor.

Strangely enough, everyone political obstacles, I think they are insured. from such result, because so hard to build data centers for all the reasons mentioned we talked. There is a whole moral panic {slash} hysteria {slash}, what exactly are these political obstacles, I think almost guarantee that no there will be a surplus offers compared with exponential growing demand. So, in a strange way, you protected from this. But, J.K., do you Can I have a second? switch and to tell you that so brilliant in because of what you did here Jensen?

Another part in return for this. They receive a share income above this limit . Who Who Gets Income share is higher minimal? Intel. Yes. So, I think the agreements will work so that they will be ready guarantee sale on a level, you know, far from 30 dollars per watt. We are ready guarantee sale at a very low levels of risk that we are satisfied with our very clear vision production capacities and future trends models. And instead we get a certain higher income share, probably higher equal.

And Morgan Stanley wrote a wonderful a note, it seems, two days ago. Nvidia can to become, and this is actually royalties, these shares income, they can to become very fast a very large cloud with low capital. Let me tell you why I believe that what Jensen did it, it was so brilliant. It because the numbers are so big, that TAM is limited ability finance it expansion. And what he does, this makes it easier financial constraint, so that he can grow up, eh -eh, so big, how much in reality there is a TAM, right?

It removes this restriction. So, just for example, Elon wants add somewhere around 6 up to 8 gigawatts next year. We we know that this will cost from 3 to 400 billion dollars capital investments. Company just attracted 100 billions of dollars in within its placement of shares and debt obligations. So, obviously, they I'll have to do it somehow. to finance. And, as we spoke in previous episode , the simplest in a way funding would be get funding from seller from Nvidia, especially considering, what is the payback period may be only 1 year .

So now it's Jensen creates, so to speak say, credit line, using these big banks, using these large direct funds investment, and he makes it available, and this will now bring benefit from all this buyers below current. I don't think so. This is a cyclical situation. This is not cyclical. the situation that Gurley worries. I I think the Wall Street banks street and private investment companies provide financing for based on expected cash flows that will be provided these graphics processors. In a way, Nvidia becomes central artificial bank intelligence.

Federal backup system artificial intelligence. This is a kind of way conceptualize that what they do. AND same as Federal backup system, as ...A lot of monetary politicians work through private banks. Here you will have, um, and I, probably more banks will agree to it. Do you have private markets that charge commissions, guarantee these deals, and Nvidia helps this promote. So I don't I think it's true. cyclical financing. Yes, I watched it too. this special edition of CNBC. One moment that I really I remember when boys from private school capital said that it's a bit like, what do they do with financing airplanes when the airline buys airplanes, they can to finance it is not only on the basis of creditworthiness airlines, but also because the planes themselves have value.

And this makes it much easier financing, because you know that even if the airline will get into trouble situation, you protected, right? IN you have funding, asset-backed . And that's what they, by the way essentially, they do it here with graphic processors. Now, to make this happen, Nvidia should standardize, they must create, in essence, reference designs etc. so that all this can be done it was enough to standardize in order to these guys from Wall- Street could now do this pack and securitize it, to turn into valuables papers.

And this, in essence, what they do, as with a mortgage. Yes. I will say that there are two reasons why I don't I'm worried about this. build-up. There are two... Um, there's a regulator, something. like a throttle flaps, Gavin, if you want. THERE ARE regulator, which corresponds to physics real world, about which we said when it was about construction of centers data processing. Also there is a dead switch human, that is, you know, switch that need to keep pressed to start the train, but if you have a heart condition attack, your hand breaks away from switch, this is Anthropic and Open AI.

These are two the largest customers. One of them shortened their ambitions for the Open AI with about 1.4 trillion to approximately 600 billions. And then there is Anthropic, which we are talking about just said: " Okay, they are going to reach 400 billions of revenue next year or something like that. Or they will end the year with these pace." If they don't calculations are needed, or calculation will become so effective, or open models code will create headwind, which no matter what, they will remove the hand from purchase buttons.

And this one dead switch slow everything down to we did not receive valuables secured papers mortgage. Well, it can lead to car accidents. But in fact, that's why, I think the Anthropic IPO really important for market–getting these quarterly income and the opportunity to see their numbers quarterly. I think what will happen, probably most important a signal for the whole industry. Why? Because all these people say that AI is a bubble, these capital expenditure not justified, not will be profitable investments. And everyone will look at quarterly income Anthropic is like a pace car.

Not only them, it will they, OpenAI and SpaceX, but you have to to say that Anthropic is now a pace car. AND they must to look at them to to see if there is a signal demand. And if some will arise delays or oscillations, you know, this is first will appear there. Yes. If Anthropic sharply will slow down, then there will be a whole pile companies, because Anthropic earns a hundred billions per gigawatt computational power. That's why they can pay SpaceX, say, 50 billions per gigawatt, you know, the spot price for computing power.

That's why SpaceX can then pay Nvidia 30 billions for chips for developments. That's why Nvidia can pay TSMC, Micron, SK Hynix and all the others . This is all food chain. So, if there is some fluctuations, if someone sharply will slow down, if the pace car slow down, everyone the industry will feel it, but still better that these figures were published quarterly. That's right, checked and complies with GAAP. So, Gavin, how do we here we say, there is emergency stop switch, yes brake, how do you do it you can name it.

Or Is there enough space? Or sufficient distance between cars to there was no crowding cars? Others in words, if Dario will press sharply on brakes, if any headwind, or everyone crashes into them, is there enough space, so that the entire industry said: "Okay, yes, we are going to leave 20% slower. We We are going to go on a 30 %slower". It's very simple. If he brakes sharply, because there is no demand, there will be a crowd cars. Okay, it will be...Oh, it will be cluster cars.

Not there is enough space between cars. Yes, if it's because of demand. Now, if Dario overtakes SpaceX or OpenAI... OK. Or Soccer Aid, or Microsoft, or Salesforce, or Azure, yes. There is no such thing, there is no such thing there is no such thing, What there will be no crowding. But I believe that Entropicus will be true important. And he will break your brains to many people. AND the reason is that, on In my opinion, there are many macro and value investors who are very confidently do these predictions about AI, based on assumption that tokens subsidized.

Anthropic generates money. And they, I would say, already famous in the world. And I never actually talked to Anthropic, but , you know, it consists of the impression that there is enough signals that, if to look at what it was said publicly, they generate money. They profitable. Tokens from open source profitable. Anthropic profitable. Open AI is not generates money, he, certainly, will generate. Um, SpaceX, you know, the same thing. AND when I hear these macro- and valuable investors who are ignorant, they very smart, but they don't know the facts , and they assume, what are tokens subsidized, and all this will collapse in some kind of cyclical financial wealth, hmm, they just are wrong.

The vast majority profitable tokens for everyone in the chain. Everyone. And I think that Anthropic S1 to some extent this will clearly clarify, but when I listen to these macro and value investors who talking about AI, it a little about what I am I say, you know, I'm very pessimistically set on world economy, because oil is expensive $500 per barrel. And you know what? If oil cost 500 dollars per barrel, that's there would be a good reason to be pessimistic. This is simply not true. As if they just are wrong.

As if a simple fact is false . They are paid to they were...They are paid , in a certain sense, for to be scrollers fate and merchants of fortune. I I can tell you, you know, one position, which I am working on, and exactly the number of people, employed in United States, for example, 150 million, 160 millions; what are theirs total salaries, 10 trillion, 12 trillions, whatever it is was. There is no such thing. a world where I can't see corporations and people in economy, which spend 5 or 10% their salaries employees on equivalent in tokens.

In other words, if someone earns money in average 80,000 dollars a year, he will spend $8,000 , 10% of your salary or at least 4,000 dollars to to raise your efficiency. We have analogy for this. This is software SaaS provisioning and computational computers on their desktop computers, mobile phones etc. Just theirs general using technologies. If add these numbers together, it turns out 8 trillions of dollars artificial costs intelligence only in United States. It , obviously, it will happen. And if I'm wrong, and this is half, and this is 500 billions, well, that's approximately the same trend, where are these companies are developing, and this not even talking about the rest of the planet.

I I think we have everything. It will be fine, friends. Good. Let's...Yes, let's finish on this. I have two more. stories I'm into I want to reach. We are fast Let's look at this. Very interesting story from New Jersey and New York around Amazon. General Director New Jersey of the past filed a lawsuit a week ago on Amazon. They claim that the company is cheating, essentially, hiring drivers through subcontractors . This is well known. It called DSP. This partner concept delivery services are very smart, maybe too much a smart way to Amazon to protect itself, according to these lawsuits that begin accrue.

Ago what if the Amazon driver will get into an accident, there is responsibility, and this frees you from some things for which you, perhaps, would have to pay, if they were by Amazon employees. On Monday, the mayor of New York before teams, Zoran Mondani, rushed into fight. He expresses your support and took one of his smiling broadly socialist, democratic socialist videos , thanks to which he now it is becoming known, where he speaks and requires through municipal laws, which will oblige Amazon hire the drivers themselves, I quote: "Corporations, such as Amazon, are building billion-dollar business models, isolating themselves from responsibility through the system exploitative subcontracts".

It the problem has been around for a long time. Unfortunately, it happened. several deaths, so what on these subcontractors seems huge pressure, and people tragically die when encounter people because they working in real world. And since drivers technically hired DSP, Amazon states: "You know, we are not an employer, so we don't responsible." And to us no need to enter in the trade union negotiations or work obligations. So, Sax, I'm curious. What do you think about this? from the point of view general surge people's faith in socialism now and belief in that corporations are cheating on the margin.

Is he cheating? Amazon? Is this too much? cunningly, and perhaps they need to give up from this? Me Your opinion is interesting, because this is gaining momentum, and this will be one of the main problems in by-elections and, apparently in 2028. Well, I mean that this is a common practice. She is not focused. in New York. I think that Amazon has been around for a while uses these networks of independent contractors as driving force. And they they say that if you get rid of them opportunities use these independent contractors, this will lead to increase in value delivery.

And it was research that showed that the average household can pay hundreds dollars more if you will change it. It slow down service. It put under the threat of a thousand jobs in small subcontractors enterprises. And Amazon says that it can lead to their transfer activities outside the cities. In addition, one of the things this one does DSP model, this is what it allows for quick scaling for peak loads. So, festive bursts, dense city ​​routes, lower costs. It much more flexible scheme than if everyone were just by Amazon employees.

So, here's how the market developed. I I think that when they started, Amazon, probably did everything himself, and then started decentralize it to these small networks business, these contractors. It actually creates more jobs. This creates more entrepreneurship. I I think that these suppliers digital services (DSP) hire thousands people in New York. These jobs will be at risk if you essentially make them illegal. So I I think this is the one. case when... What is your position on about this, Sachs? What is your position on about this?

I would just let it be. freedom of contract and voluntary exchange prevail here. AND look, this market structure developed not just like that. For employees, this liked, for DSP owners it is liked it, Amazon it liked it. It mutually beneficial agreements. I am not I see the reasons why Monder has to interfere in this city . Gavin, what is your opinion on this matter ? What would I do if was Amazon, and I'm sure that they don't will do, I would just introduced a surcharge. Hey, this is Monder...This is a collection Zohra Monder, who imposed on you.

Many people use Amazon, not only wealthy people in New York. Many people use Amazon. And this is the collection of the Zohar Mondera as a result this ill-conceived politics. I would include its like a point in every score. Because it is to This is what it will lead to. When to compare Amazon service with anything, what the government provides, incredibly, how effective Amazon. You are ordering her thing, and she can be there the same day. Or, at worst, case, its will deliver to you next day.

Unbelievable, how wonderful Amazon service. He optimized for low prices for consumers. And we, by In fact, let's get into this, and the government will intervene in Amazon driver network? What city service? provided by Mayor Garcetti remotely, what is it? as effective as that, what does Amazon do, what does it is going now tell them how run your own business? It madness. If you look at that What Starbucks did when it was run by Schulz, then he gave to people work that is worth doing was a mother.

And he was proud of what he took the lowest paid employees in organizations and gave them subsistence minimum. And this livelihood the minimum is somewhere around 20 dollars, 22 dollars in these cities. Hereby DSP drivers actually pay pretty good, 18 dollars, 19 dollars, 20 dollars. But this is not was created, uh, on free market, like you as if you were saying, Sachs, was exactly what everyone wanted. No, it was specially created in order to benefit Amazon and shift the burden these workers on American taxpayers. AND They deserve it.

criticize New Jersey and New York. AND I think you know, and I am a long-time Amazon customer. I is a Prime subscriber from the very beginning. For there would be minimal to make expenses these workers regular employees. And you you can still have some DSP for scaling, Of course. But this is for them the opportunity to do these employees the leading part Amazon family. Delivery in New York will cost them only 25 cents from 18-19 dollars for an hour, to have a little to raise them salary, perhaps, on 21-22 dollars, and provide them benefits.

It's easy. make. And what is our the industry has a crazy blind spot, so it is in because they are looking for efficiency, and Amazon is the most effective company in the world, as much as to the fact that she is treated cruelly to the most vulnerable members of our society and opposes corporate spirit capitalism. And this level of capitalism, who needs self-correction. And I I think the government will have very important here arguments. I think that it would be great for Amazon to get ahead of this, so what is that democratic socialist, too known as socialist communist, in my opinion opinion is very important argument that will like it to Americans who believe that this a group of people use, and that we we can easily to afford pay 25 cents more for delivery parcels.

Yes, Sax, you You are right. This can add 100 dollars. Okay, then why should we Uber won't do it either, Jay Kele? Of course, Uber already did. Because Uber actually criticized for that namely, using independent contractors instead to make them complete employees. Sax actually makes a good point regarding Uber, and that would be equivalent to, if Uber had three or four thousand people, which represent their drivers, and they would not have any influence. Uh, Uber actually has huge insurance influence, huge lawsuits, with which they have to to fight for what these drivers do.

So that's a completely different matter , and they went through it. I am not a fan. socialism, but I think for Amazon, which one will be automate business as usual , there is a possibility a little, er, er, cut your amazing business, but then to be on the right track side of history and not to take from one's own workers' aid on unemployment, food stamps and health insurance and don't give it to the population. And these socialists are going to prove this is absurd in the next two election cycles.

And this will matter. Socialism is playing. Socialism breaks through, he infects people's brains, and capitalists should say: " You know what? We don't you need to be like this cruel and not you need to be like this "smart." So, this is my message as long-term Amazon shareholder. I have very big position in Amazon, and I continue it increase every year. They must act. correctly and to get ahead of this, to they could say: " Hey, we're incredible. employer, on unlike that we really we are trying hide and redistribute these costs".

This is mine. the only opinion on this reason. Not a socialist , not an idiot, not interested in to, you know, mess around with capitalism. I I think they are here. a little tricked. It my only opinion. And I I think they are must do correctly and to relate to the most vulnerable employees in lowest groups income. What is it? Now I understand why. you brought it up consideration so that you can to start your trouble. Well, no, no. I took it out. this is under consideration because this is one of the main news.

And I endured it. for consideration, because we we are talking about socialism every week. AND we are talking about everyone advantages of capitalism . And there are disadvantages capitalism, when we we can become too cunning, and then try to take on expenses themselves. Let's . You want Amazon hired all of these drivers. By the way, they they don't do this, I think they should to hire, I know. You said that Do you want a bunch? people left with DSP, can it happen, or you decide who is who?

I think they could would do it proactively. This is great. question. Andy Jassy could proactively decide how it is Schultz did when he wants to get rid of DSP. He wants to make them all employees. The compromise is here. is that there is a DSP for scaling, when there is demand during holidays. It seems that this great idea. But I I think he should. get ahead of it. Since you admit, that DSPs are better because they offer more flexible conditions the jobs they want people, why are you at all are you interfering in this agreement?

This is not true. People who working in DSP, DSPs themselves, people who own these fragmented , very small companies that go bankrupt, just get into an accident, to protect Amazon, these employees are very would like to have shares Amazon. They would like receive RSUs, and they would like to receive medical maintenance and benefits. Some of these drivers Many people work for Uber. time. I actually can. And they switch between the driver's work on Amazon and driver work on Uber. They like it flexible schedule work. I misinformed.

You You are wrong. These are people, who work full time changes. They work. eight to twelve hours. I think that actually these are changes in 10-12 hours. They are the ones those workers... Uh, you can. to speak for them name, if...They can speak for themselves for themselves, and they are organized, and Amazon is trying to destroy efforts organizations when they are trying to unite in trade unions. And this tool for the destruction of this, forcing them to be part of DSP. Such a game in the field, Sax.

Well, you already admitted that Do you think DSP must stay. So, I think some of them must do this, and I think Andy Jassi and the Amazon team must be ahead this, because it's bad looks. Not in interests capitalism to be so cruel. I I think he should do it proactively. I don't think it's him. should be forced do. This is mine. personal belief . Just like Schultz said that there is possibility provide a barista subsistence salary and benefits, and it was one of the best Starbucks solutions.

And this really what it's about capitalism must start thinking. What our role? What is ours? role in promoting movement towards socialism among young people? And this an example of one of reasons why people are beginning to believe in socialism, consists in because they believe that such companies like Amazon, a little cheating on margins and perhaps are trying to take the expenses they must bear, and translate them into American taxpayers. AND You know what? Mondami and New Jersey will win these lawsuits. I predict this is now.

Well, listen, Jay. Kele, all this sounds very beautiful, and I want save capitalism just as strong as you, and if Amazon really will save capitalism, then I guess I'm ready for this, but I don't I'm sure that this is will happen. What are we we know for sure, that's it that this will lead to increase in value parcels for approximately $5.20 per package, and it will cost families for $664 more than a year, and also endangers about 5,000 workers places. So there are real costs if to do things your way you say.

When people wear your hat voters, yes, they are exposed socialist to the fanfare, but then when they are consumers, when they wear their consumer hat, they just want pay less for parcels. They will hate that the fact that the parcels cost 5 dollars more. And, as he said Gavin, they should to include in your orders surcharge Mem Donny. And it's not 25 cents, it's 5 dollars. 5.20 dollars per parcel. 664 dollars on household on year in New York. It madness. These are real costs. If to talk about cost adjustment life is huge cost adjustment life.

Jay Callas has there were problems with the internet, and we just moving further. Gavin, in last issue, which Are you thinking about Grok Bot? What Are you thinking about releases? that xAI released this week? I think it was Grok Bot, and then also Grok 4.6, right? And people they said that he close to the limit or on the edge, but much cheaper. So I claimed again last week that we agreed on a duopoly advanced laboratories with Anthropic and OpenAI. Was I wrong? ? Is this number three?

Is he more flexible than me thought? Yes, I think so. I I mean, I think, what could it be Pareto frontier graph? This is quite strange. Now, on the left, this is shows something on kind of quality or intelligence on the Y axis and value on the X axis. So, you want to be in upper right quadrant. And you you can see here that you just want to be beyond that. And you can just to see, as far as revolutionary is Grok pricing 4.6. Right. So, how are you?

...Sorry. Sorry. Just like you move to the right along the X axis, it becomes cheaper. It becomes cheaper. And the Y axis is opportunities, and they are becoming more and more more capable when you are moving up. But they invert the X axis, so, moving upwards on the right, it's better, isn't that right? That's right. And I have to meaning that it is now, and this is on Cursor Bench and XAI, you know, SpaceX almost probably has the right buy Cursor, they work with Cursor. So, maybe this evaluating your own home work.

Well, right here. on the right, this is from Databricks, which just attracted 190 billion dollars. Hmm, that's very a real company that very developed, and you you can see that even in this you you are far ahead Fable 5, which, you know, I I think most people would agree, What is this, hmm, gold? standard. That's it, in you are of higher quality than at a slightly lower price. So, I think Grok 4.6 is, Well, you know, there are numbers. numbers. You know, on X, I I think we are released.

Um, people, you know, we will have to wait and see the vibrations, but I think that this quite interesting. Um, DHH, which I think created Ruby on Rails, um, David Heinemeier Hansson. Yes. Yes, he published something really positive. Basecamp, Basecamp. David Heinemeier Hansson from DHH. 37 Signals, great, great guy. 37 Signals, Basecamp, yes. He posted something really positive. Em , you saw the rating Databricks, have you seen our Morq-Morqor rating, and you saw that I think that early vibrations really good at these benchmarks, which, um, are confirmed in the real world.

In other words, Elon caught up, and if you explained why he caught up, that would be actually two reasons. First, he bought Cursor, it has this team, they a close-knit team. Secondly, he shuffled xAI management and invited his friends SpaceX killers. Yes, Gavin, there are two reasons. He invited, you know, he invited, you know, SpaceX aces, along with predecessor aces and some new ones people, and this... And he did it for six short months, Gavin. Never throw away Elon's accounts. He is often late. He never is wrong and prone to exceed one's obligation when it happens.

And this is for his own recognition. He says: " I am always late, but in the end we "We'll fix everything." Right? Absolutely. And this is still relatively small model. These are models on 1.1.5 trillion parameters. Bigger Grok 4.7 model, which has to be significantly more powerful, it will turn out, you know, in a few weeks. GrokBot also seems the same important. This is similar to next moment open paw for AI , for such AI personalization, for him democratization, for in order to do its more like this, as Zuckerberg said in in his manifesto, he wants to create bots for everyone, and that's what colleague of Claude and Claude do extremely good, like Hermes, Perplexity Computer and OpenCloud.

For me, some of them such, and I try to be very technical, but I like it "Rock, scissors, paper" ". Yes. Not too difficult use, but it's much easier for people, scissors, paper much simpler in use. I have meaning OpenCloud and Hermes, you need to be Hermes? Hermes. Scarf? Yes, that's what I call them. They are Hermes. They just such. These are very fluffy. If you have ever- was this channel ever it's something like that. Saks. How are you, you, you, you, ours? wives from time to time they order some channel and share with us bread.

Well, this is It's quite surprising that in 6 months he caught up, but Saks, this attracts attention. Hmm. Yes. This creates a small voltage. He places Anthropic. He is catching up. Anthropic. What do you think? about this tension that he there is, Amazon Web Services , Elon web services, while competing with them more and more? What I think about it, yes this is what he creates these huge computational clusters, right? AND this gives him opportunity to teach large model, potentially large advanced model. So you can think about it's like a call option, to catch up and to become real advanced laboratory.

But if he doesn't will succeed, he has put option to just sell calculation. Yes, spot prices advanced laboratories. And he has a law about the right to take these calculating backwards from Anthropic. He turned this on. point, according to messages, in agreement with Anthropic. So he can take away computers back, if he wants. Right. Both the parties have a 90-day cancellation, that means that if spot price calculations will increase, it wins. By the way, if it will fail, it can harm him, because either side can view conditions within 90 days , isn't it?

So, by In essence, Elon can to charge spot price minus 90. Hmm. So, it's a risk. But it all comes back to what we are talking about talked about earlier, but namely: do you think, that are computational resources will be limited in next year and in a year? And if so, then it is very profitable dynamics for Elon, because again, he very close to that to charge real spot price for yours computational resources, but he can also essentially redirect them to own service if will be able to take on initiative.

Yes, and whether can I say that interesting? Sorry, I don't ...Sorry about that I interrupt, but no, no, sorry about that I interrupt. I spoke with by many investors about SpaceX, okay? Yes. And what just makes me What is striking is that everyone will have a different one A look at Starlink. Maybe they skeptically configured, perhaps, they are skeptical are set on Starlink directly on mobile communication. I think the plan using, um, these distributed Starlink, in fact, almost, what do we call small cells, femtocells, there are interesting.

Everyone has will be peculiar orbital vision calculations. Regardless because they are bullish or bears, they will have your vision. They have will have its own vision ground computing. Very few of these investors in general talking about Grok. And he is Pareto dominant for by many indicators . And we have proof. the existence of that Anthropic very fast increased from a billion to 50 billion. And if Anthropic costs, you know, two or two trillion for FTT, you know, three or four trillion, you know, maybe the number will be set on the previous one episode.

Um, I said that, on In my opinion, this is the price. market clearing. I am not said that's right, in my opinion, he cost. Um, you know, maybe people should start consider Grok when they are thinking about SpaceX. IN SpaceX has so much ways to win, and I I think this is higher. order is here, and that's it, and we all, obviously we are talking about our own matter. People from this podcasts are among the first investors in SpaceX. I was part of management, hmm, a small part second-hand management round of 10 billions.

Hey, there's one thing that became known during show that we are sure about worth talking about, but exactly what Silver Lake, possibly buying Workday. Workday grew by 17%, and if this is true, then I I think this is important. moment for stock market. You know, software software is in death spiral, which one actually started, you know, you probably know, approximately 3-4 months ago, but it was in in a difficult situation during 12, 15, 18 months. And the return private joint-stock company capital on software ensuring really changes the investment landscape.

And I don't know, what do they think potential buyers Workday, but I know that growth of open code and growth competitiveness open source–this find for American industries software software. Hmm, this is an absolute find.

Go deeper

• All All-In Podcast episode guides