Nvidia's Jensen Huang Scandal & Anthropic's IPO Deflates to $1.4 Trillion

0:00Welcome back to the Market Maker podcast and for two straight years the institutional playbook was pretty simple. The S &P 500 was essentially NVIDIA in an oversized trench coat where Jensen Huang went global liquidity would follow until the day we're recording this that was yesterday. Last week Wall Street stared straight into the macro abyss. The Nasdaq 100 took a historic 6.4 % single day beating wiping out I love this stat, Piers. $2.3 trillion in market cap as post-election jitters and data center skepticism triggered maximum panic. But yesterday brought a violent counterpunch. The S &P 500 actually rallied 2.4 % its best session since April and chip indices jumped 7%.

0:49Yet in the middle of all the green, there was a glaring red anomaly and that was NVIDIA. They were down heavy at one point, I think like 8-9%. They fell 4.3 % at the close. That kind of divergence has happened exactly just three times since 2001. And I love this. I saw a journalist saying this. The umbilical cord between the market and its undisputed king has been snapped. And the catalyst isn't a missed earnings print. It's an espionage scandal. An affair indeed, admitted by Jensen Huang, an FBI inquiry and an imperial CEO presiding over a board with no chairman and no clear successor. Meanwhile, on the software side, we've had Darion Anthropic walking directly into the storm.

1:42Their IPO roadshow kicks off this week. Whispered valuations have deflated from$2 trillion down to the latest we've heard in the FT. Just an hour ago from recording was 1.4. I heard some rumors talking about 1.1 earlier in the week. Their defense, obviously dropping a new Mythos fable model doesn't hurt on the eve of the tour to engineer the optics and defend against open AI with Goldman Sachs, the lead left, also pumping some impressive revenue forecasts that they think for the AI giant. So is the AI trade maturing or is the circular financing machine showing some cracks? We're going to break down the decoupling, the boardroom crisis and the trillion dollar roadshow.

2:28So Piers, caught your debate, the bull bear case. Oh, yeah. I mean, listeners shouldn't even need to know who side I'm picking. I'm the perma hyperscaler bull. So there's no holding this back. Like, Stephen sent me a message. I think there was, it's going to be the funniest.

2:52Piers Curran: It's obviously, I'm going with the piss-taking side of things here, but there was a comment on, I think it was on YouTube, and it was like, because, yeah, we dropped the poll, right? Who are you with, me or Stephen? And they replied, it's like the Messi-Ronaldo debate. You know, you can't choose. Was there a question? theme i think were they talking about our arguments or uh or ourselves i'm not sure maybe but uh yeah i mean i thought i'm the messy of those two that's i thought i thought it was a slam dunk i thought there's no messy run out of debate here messy is the goat right well clearly i'm gonna get some comments on that but talking about goats or maybe fallen goats goats chopped out at the knees yeah jensen's That's a bit of a shocker, hasn't he?

3:47Piers Curran: Well, he has this week. By the time this hits people's ears two weeks down the line, I'm sure it would just be a forgotten blip. What affair. Well, there is one way this could... I mean, you know, yeah, I think you're right. What you're implying is this is just going to be like yesterday's news pretty quickly. And he's, you know, lines being thrown out. Too big to fire is the latest, isn't it? Which is absolutely true, unless it does go... If the FBI, who are investigating this, of course, and there's obviously a few different angles, but one of them is that this woman was a spy working for the Chinese.

4:34If that turns out to be true, and it turns out he was, I would assume, inadvertently passing information, then that is a real problem. And the problem being compounded by the incredibly unusual structure of the board, and NVIDIA, they're the only company in the S &P 100, right?

5:06Piers Curran: That's the 100 biggest companies by market cap in the US. They're the only one that doesn't have a chairman. And then on top of that, yeah, you've obviously got this situation where the directors on the board, well, they've been on that board for like since the 90s. The average tenure, I think if you go across the whole board, the average tenure is 15 years. That's double the usual average across the rest of the S &P 500. So it's a bit unusual. and he's got quite a flat sort of, I would say, management structure. And so he's got 60 people that directly report to him. So he, yeah, in terms of that business and operating that business, he is too big to fire unless he gets arrested.

5:55Yeah, so yeah, I mean, one of the ways I'd read it is that you talk about key man risk and I was looking at how much of NVIDIA does he own? because you would assume kind of like Zuckerberg, Elon, others, you go, right, something super meaningful and they have super voting shares and all these sorts of things, meaning that if they get diluted, they still have overall control. He's actually the opposite. Hoang owns 3.6 % stake worth a measly$150 billion. He's been short changed there, I would say. But I guess it's, so that's one thing so he actually is this even though he's a he's kind of outsized in his role in terms of his personality but because he is the pin-up boy of this whole thing and a third point of terminology i really kind of i quite enjoyed reading this morning on the commute was nvidia has acted as a de facto central bank of ai yeah now i thought that was great in the fact that it's directly financing customers, extending loans, including Megaspeed, I think a few years ago, which is this Chinese firm.

7:06They've taken equity stakes in other firms. They're backstopping data center leases. So that whole thing about too big to fire, I mean, that came out of, you've always got to have context here. Who is saying these things? And it's Sequoia's Sean Maguire, who I'm sure you've got your little table of investors in the early days of a lot of these firms. I'm sure Sequoia's got their little bit of capital on the table here with some exposure. So is that the thing though? Is it because the setup of the company could go on and survive without him? However, this isn't just an NVIDIA story now. NVIDIA is the center of a universe that is driving our economic universe.

7:55Right. And so therefore, he's almost irreplaceable from a technological leadership standpoint. Yeah. His capacity to navigate Washington diplomacy, I guess that's going to be the interesting one to watch and how close is he now to this stuff. Yeah. But Trump is still around, by the way. And Trump wasn't exactly being super skeptical of some of Jensen's activities. Yeah. No comment.

8:23Piers Curran: But they are too big to fail. Forget Jensen for a second. They are too big to fail, the business. It is so, it's got its fingers in every single little pie of this AI boom. It's the absolute foundation of the whole thing. And if you remove the foundation, obviously the whole thing collapses. So, and that's not just the AI boom, right? That's the de facto economic growth story right now. So it's too big to fail. I think, yeah, you mentioned that Jensen's role in that kind of being sort of diplomatic. It's almost like the go-between between the Chinese and the US, to be honest. He's got to be yanked out of that sort of type of role immediately.

9:11Piers Curran: So this will have a negative impact, obviously, on the man himself, but I would say on the company. I'm just looking now in terms of negative impact on the company, it's down 5.5 % today. When we started this podcast, which was about, what, three, four minutes ago, it was down 4%. So it's now down 5.5%. They've heard that we've gone live. So they've... Quick. I'll point it out now. I do love a bit of technicals, but there is a monster technical support level in NVIDIA's share price, which is the March 30th low. It's the low of 2026. And that's down at 165. We're trading 169, just broken down below 170.

9:59Piers Curran: 165 is massive, huge level. So yeah, we've obviously had a really, really, really steep reversal, a second big step to the downside after the one that we had obviously back in the last week. Well, let's talk about the context of this week and last week. Because the NASDAQ 100 is 6.4 % one day full. I mean, I saw the volatility index, the VIX, to levels not seen since April 2025. I think people have got a pretty good sense check of that, even though we were there straight on it, saying people are overlooking the data center legislation in certain key concentrated areas geographically in the US.

10:44But now that's become what's being pinned on as the kind of catalyst, if you like. What about this reversal and decoupling that we've had? You just mentioned NVIDIA is down. They were down yesterday, but the NASDAQ had one of its best days in a long time. That's not normally normal when an S &P has index-weighted components components of which NVIDIA is one of the largest companies in the world. So the science or maths being if they fall, everyone falls. And that's been highly correlated, but it wasn't the case and hasn't been in the last few sessions. Why is that?

11:22Piers Curran: Yeah, it's, you know, what is the driving factor of the move? And actually, if we switch to the S &P versus NVIDIA rather than the NASDAQ, the S &P rallied 2%, which was its best day, one of its best days. What did you say? Anyway, for a long time, right? And NVIDIA dropped 4.3 % on the same day. So actually, outside of May 2009, this divergence has almost never happened since NVIDIA joined the S &P 500 back in 2001. So you've had a 25-year run where basically the correlation between NVIDIA and the index is very strong and positive and it's become ever stronger and more positive as NVIDIA has climbed up through the rankings to become the absolute top dog, biggest component in the whole index, right?

12:11Piers Curran: So this is a fantastically unusual event because the reason for NVIDIA's downside is there's a very specific story about its CEO. And whilst, yes, that might have implications for, right, how's the business going to be run? What's Jensen's role at a more diplomatic level? Is there going to be any sort of consumer backlash against NVIDIA? Well, we'll talk about that maybe in a minute. No. How can you backlash against NVIDIA? What consumers buying GPUs? No one, right? But because of this very unique thing about an individual person, this isn't a trade related to AI. Well, actually, the reversal of the whole index was a rebound from the big step lower that we had on Friday.

13:02Piers Curran: So you've got the whole market snapping and rebounding back from the heavy sell-off from Friday. So the macro underlying theme is rebound risk on. And you've then got a singular negative story that just so happens to be very negatively impacting the biggest company in the index, which just for that singular reason, it's just gone in exactly the opposite direction. So some stats to support what you're describing there. So Nvidia, well, this has probably changed, but at last check, just under 7 % of the S &P 500. And historically, whenever Nvidia sheds more than 4%, or falls more than 4%, the S &P 500 declines by an average of 1.6.

13:45That's the normal relationship that it would have. But the divergence on Monday of this week, the S &P was up 2.35 whereas NVIDIA was down 4.3 so is this where that terminology people often talk about which is idiosyncratic in this case governance themes being that there is an explainable thing so does this make you feel given the context of the debate you had with Stephen and the frothliness of the market at the moment with Anthropic about to list and all the rest of it and some of the market pullback we've had. Is this actually a little bit reassuring what's happened with the decoupling? Yes.

14:28Actually, that's an interesting spin on it, which I hadn't quite thought about. But you're right. When you've got the biggest player of all having a monster sell-off

14:39Piers Curran: and that can't stop the broader market having a really solid, strong rebound, then that is a really good sign of the strength do I say the durability of the rebound? Obviously, we'll see. I mean, the market has been the whole AI trade. And I would say media driven to a degree, the media are so anti the AI boom, it is ridiculous. And any, any, any, any negative story, it gets ramped and pumped and it's all over the headlines, right? So obviously, this is just another event where they're like, oh my god, this is amazing for us, the fodder, you know, bang, and they're all over Jensen. But the fact that didn't cause another leg lower, I think it's a really interesting signal about where the market is at with all of this.

15:36Yeah, I did see an interesting stat. I don't have it here, but I read it earlier today. And I think it was like a macro Goldman survey or snaps pulse survey that they did and it was looking at actually any of the holders of the index weren't selling in that move it was all the retail punters and so forth uh getting out and it is exactly to that intro and i kind of had the sensational intro a little bit on purpose for me you know working in and around the news for so many well two decades now it's funny it's like the invested interest of the size of capital and the influence behind the scenes involved it's a big thing to fail let's call it whereas what sells papers and generates clicks i tell you what does an espionage scandal an affair by the number one person who represents the poster boy of ai an fbi inquiry an imperial ceo as they're calling him because there is no chair yeah um and he has got no clue it's just It's almost like a script that's been written for journalists to go to town on.

16:47It's a Netflix script. I refuse to believe any other reason.

16:54Piers Curran: But as I said, people will forget it. Although, as I have said, the share price is down again sharply today. But as long as he doesn't get arrested, and look, that could happen. I have got no idea, right? We'll see what the FBI dig up. as long as he doesn't get arrested and he's in bigger, obviously, legal problems. As long as that doesn't happen, no doubt this will be a storm in Jensen's teacup and he will rise above it all. I'm not condoning his actions, by the way. That's just life and the way this is going to work. What will happen for the long-term benefit of NVIDIA, this, I'm sure, will trigger you know what's going to go on the board meeting you know agenda item number one the next board will be well hang on this has indicated to us that our key man risk is way too high we need to start now a process of reducing that key man risk so I think this can help Nvidia over the medium term but yeah as long as he doesn't get arrested this is all going to be fine maybe Jensen's hedged himself by becoming very close to Trump at this point.

18:09I was just having a quick look, and legally, according to Article 2, Section 2 of Clause 1 of the US Constitution, might I add, he can be pardoned. Well, yeah, but can he at any point? I think presidents have that thing.

18:26Piers Curran: When they leave office, they do a whole load of pardons, don't they? But that's not for two years. So can he just do an off-the-cuff pardon at any moment? well it's i think it doesn't shield him from other investigations state civil lawsuits removal by his own board all that sort of stuff but if the fbi investigation to megaspete allegations uh those export control violations they had because you know they were saying they were sneaking in the nvidia chips which would be a violation of that um or espionage leads to potential federal criminal exposure, Trump can issue a full preemptive pardon. Okay.

19:10It's kind of, he's hedged himself here. He's a clever guy.

19:14Piers Curran: Trump's not good. Trump will turn on a dime. If there's any sniff that Jensen has been involved in facilitating the legal smuggling of NVIDIA GPUs to China, or if there's any hint he's been involved in any kind of espionage scenario. Trump will die. He will turn on Jensen in a heartbeat. But if NVIDIA fails, and the plug gets pulled from the AI bathtub, history will remember Donald Trump as the worst president that led to catastrophic loss. Donald's probably in his bedroom at the moment just buying up NVIDIA stock here before he then announces the pardon. Well, look, let's move on and let's talk about some of the other major headlines here because the other interesting one, of course, is whilst all of this has been going on, anthropic, not the best timing that one would have hoped for.

20:19I mean, it feels like they were probably six months too late. If they just got this away, maybe just ahead of SpaceX or just right the two weeks after, They could have absolutely hit top of market. But it looks like compared to the 2 trillion whispers, actually the stat line is that was five weeks ago. Right. It was early October. Wow, is that right? 2 trillion was still being... Banded about. Right. Now it's gone down to 1.4 to 1.5 trillion. And now some are saying prospective Rochelle bidders is sat at a 1.1 to 1.3. So yeah, when there was an FT article out earlier today, what was that giving us in terms of some signals and how this might go?

21:08Piers Curran: Well, so the FT dropped it. They're targeting 1.4 trillion is the latest number. Obviously, well, they're looking to raise, and this has been, so this is a bit more interesting info. They're looking to raise now apparently$64 billion, which, can I just pause? You've got to love Elon, not only for the timing of his IPO, absolutely just the timing's genius, right? And will he now hang on to his largest ever IPO on record status? and it looks like he will, right? In terms of both the valuation, because I think, correct me if I'm wrong, SpaceX was 1.75 trillion valuation and he raised$85 billion. So here we're now saying Anthropic, 1.4 trillion valuation, raising$64 billion.

22:07Piers Curran: So Elon will hang on to his records. So what they're looking to do is plan. and Anthropic are looking to sell 825 million shares. They've issued a price range, which actually, remember, SpaceX didn't. SpaceX didn't even have a range. They said, this is the price. Take it or leave it, which is very unusual for an IPO. And actually, yeah, it'd be interesting. If Anthropic did go earlier, a month or two back, I'm sure they would have been in the driving seat. They're the ones with the power, and maybe they would have gone with a singular price point, but they've lost the initiative and they're on the back foot and the AI trade is absolutely crumpled.

22:53Piers Curran: So now they're going the range between 70 and$78. So if they do sell the full 825 million at the top of that range, which is$78 per share, that's where you get your 64 billion raise from. So just talk to me then, strategically, why not delay it? Why not let the ripples just kind of settle? Is it that it could get worse? You have to go now. It's going to look bad if you pull out. What am I missing there? I think it could get worse. And, you know, they need the cash, you know, in the end. They don't have the luxury to go, oh, it's fine. We'll just wait. Let's give it another 12 months. That's not how this game works.

23:45Piers Curran: These data centers, these GPUs, they ain't cheap, right? And their whole model is to build out this infrastructure. And the capital outlay is just ridiculous. ridiculous so they don't have the luxury although i mean what was surprising when they dropped their s1 was that they had i think it was 100 billion on their balance sheet in cash so i mean so they do have some cash they could delay it but yeah i just think what happens if you delay it and then six months later we've got a big massive crash and then they're like ah damn it do we delay it again And ultimately, that runway of cash runs out.

24:26So I think they are right to just plow on and take the hit.

24:33Piers Curran: And look, to be honest, it's not like they're going to be selling much of the business from a percentage perspective. So whilst the valuation,$1.4 trillion, let's just say it's that, whilst that is a lot less than two, they don't, I mean, do they really, the valuation is not the most important thing here for Anthropic. right it's just not it's just not the most important thing so therefore i think they're right to just crack on even though they're going to be going in a bit of a discount to what we were thinking a few weeks back yeah there was a strategist at hsbc chap called max kettner and he was talking about this talking about a sub one trillion or a conservative uh pricing structure might be a healthier route to prevent a broken IPO or a SpaceX-style dump in the secondary market to avoid.

25:27As to your point, they're thinking about where are we going to be in 2030 and beyond? Momentary, you know, they're still going to raise tens of billions of dollars to fund the build out, the acquisition of the hardware, components, whatever they need. Yeah. Like, why get obsessive? Elon's won that race now. Yeah. quite clearly.

25:49Piers Curran: And look, if you take the Elon SpaceX example, where they did go at 1.75 trillion, which was a kind of crazy valuation, their share price in the secretary market since then has been mega volatile, like huge, huge swings. And ultimately right now is trading about 15%. 15 % it's underperformed the S &P by 15 % since it's June IPO day. So it's massively underperformed and it has been crazy volatile, which is not that Elon cares, but for a lot of businesses, that would be really difficult to deal with from a kind of customer or sorry, shareholder relationship perspective, right? Elon doesn't care.

26:38But one thing you mentioned there was,

26:41Piers Curran: does it go for below$1 trillion? Now, there would be a really big problem with that. So I was looking back at the valuations of Anthropic through its history, very short history, given it kind of was founded in 2021. But we do have certain valuation timestamps, and that's each time they've raised capital in the private markets, you know, directly from investors, were able to have a little mark-to-market at that point. What was the valuation of that raise? And you go back down the list. I'll come on to my main point in a second. There's some very interesting people on this list. I know, I just saw the name.

27:19You are joking.

27:21Piers Curran: So they did the Series A in May 21. And the first two raises, Series A and Series B, these haven't been entirely verified in terms of the valuation. So the estimate, series A, in May 21, they raised$124 million at a post-money valuation of$550 million.

27:45Piers Curran: Dustin Moscovich is the most famous person people might have heard of that was involved in that. Series B, April 2022, they raised 580 million. Can you believe that? Within the first 12 months of their life as a business, they'd already raised$700 million. Anyway, that Series B was at a 4 billion estimated valuation. So it got an 8x valuation in 11 months. Who was all over that series b is your man sam bankman freed baby there he is the disgraced ftx founder who's now well got his feet up in jail somewhere um he was all over the sea what is he in line for then when this thing goes public well four billion right if he'd if he'd have just made it through God knows what the valuation of his holding is now.

28:41Piers Curran: But anyway, Series C was May 23. I won't go through all of these. Actually, Series C was interesting because that was another, well, May 23, so 13 months later. But actually, the valuation was flat for that 12-month period, April 22 through to May 23. Sam Bankman-Fried overpaid in 2022. But then anyway, you go up the list. By the time we get to March 25, we're at a 61.5 billion valuation. September 25, that was just like, what, six months later, it tripled to 183 billion. Two months after that, November 2025, it's kind of two and a half X up again at 350 bill. Feb 26, it's at 380 billion. And then the key point here, May 2026, Series H round, where Sequoia were involved, it actually went at 965 billion valuation.

29:42Piers Curran: So there's nothing worse than what we call a down round. That's where you raise capital at a lower valuation than a previous raise in the past. So, you know, they are not going to want a down round here. So it's definitely going to need to be north of the 1 trillion and look they'll get it away at north of a trillion you know if they go for 1.4 people are going to buy it i know i know the market is puked here and a lot of corners of the market are panicking but look you got your medium term hats on and people are going to hoover this up two two things there what i do like is that it was the very partner of sequoia who's saying don't worry about nvidia don't worry about it not a bubble the same guy is absolutely into nvidia into anthropic being a success here getting in right on the last round uh stumping up some serious change so this ain't no bubbles according to uh sequoia which is quite funny the other thing is uh you mentioned there about the only real anomaly was the flat uh valuation between april 2022 and may 23 makes sense because that was when chat gpt went public in november of 2022 yes so probably they raised money i'm assuming six months after chat gpt had gone wildly successful on its launch that's exactly right and that would have been why I would have thought.

31:14Yeah, absolutely. Yeah, and then like other stuff like Goldman's who are leading this, lead left on this IPO deal.

31:25Piers Curran: They've been trying to reassure markets themselves, obviously for Goldman's. They're fully invested in getting this IPO successfully away, you know, selling all of these 825 million newly issued shares, selling all of them at the highest price within reason that they can, that 70 to 78 price range. By the way, what happens? So they're on the roadshow now, right? So they're out there talking to investors at the moment, and that's the price range 70 to 78. What they'll be asked to do, the investors, by Goldman, they'll be asked then to submit indicative bids. In principle, if you're interested, how much would you be interested?

32:08Piers Curran: How many of the 825 million shares would you be interested in buying and at what price. This is then a book building process, which enables the investment bank to gauge demand. And therefore, with that demand data, then set a singular IPO price, normally between the range, 70 to 78. That's not fixed, though. It could end up going above the range or below the range, depending on how that demand gauge looks like in that book building process, right? But Goldman's will want to be getting this away at the top of the range, right? So obviously, they're talking their own book. So they're saying, ah, Alphathropic, they're expecting to make$600 billion worth of revenue by annual revenue by 2030.

32:54Piers Curran: And that's up more than 130 times from its sales and revenue last year. So they're pumping, you know, really bullish projections here on revenue growth. Yeah. And in combination with that, yeah, the spin doctors are working hard on their stage crafting dropping the i saw the the mythos and fable models the day before the roadshow begins it's like oh by the way we've just upgraded these uh humanity destroying models even more because destroying planet earth wasn't enough we're now going bigger so by the way we do have this uh even though uh there's going to be lots things around it that we won't have details on but i guess that's to serve your point uh the sales desks the syndicate bankers it's this kind of fresh product momentum defense of the uh the valuation multiple against some of these other concerns of tech multiple fatigue and so forth so yeah they are going guns blazing this is what you pay goldman's for though right That's right.

34:02But unfortunately, at the end of the day,

34:04Piers Curran: there are things that are under your control and then there are things that are not. And what is under their control, well, we can pump out bullish revenue growth expectations. We can set the price range and let's have an amazing investor roadshow. Everything in their control, obviously they're experts at. What they can't control is the macro market sentiment. What they can't control is Jensen sleeping with a spy. They can't control those things. And unfortunately, from a timing perspective, it's shocking. But I don't think they have a choice. I think, well, as we know, they are cracking on with the IPO.

34:47Piers Curran: And I think that is the right call. And maybe as a segue then, and to conclude, the other thing they can't control, even though it's a little bit circular in financing and so on and so forth, but they can't control competitors. And the only competitors in town that can really change, move the needle at all are the hyperscalers. And one of those that I saw that's also come out in the last day or two was Alphabet talking about custom silicon variants, these TPUs. And I know you've talked about TPUs a number of times on the show over probably the last 24 months. yeah talk to me about that what what is google's uh product here that they're offering this is google is amazing because they've got so many little sort of diamonds that they're sat on and i think they just don't it doesn't really get much coverage um whether or not you can argue it's not priced into their share price i'm not i'm not too sure i i would go quite there but um um they've got these things well you remember that nvidia their chips are called gpus right well google have something called tpus which is slightly different but equivalent okay so that they are a straight up competitor to nvidia at the chip level um we are in a world where nvidia GPUs are sold out.

36:20They're sold out for till like 2029, right?

36:25Piers Curran: So Alphabet sat on this very interesting product. Now, who's going to get first dibs on Alphabet's TPUs? Google, obviously. So this does give them an incredible sort of moat. And I've talked in the past about Google's, you know, complete, they've got a full vertical stack, right? In that they've got the chips, they're building data centers, they have the cloud, which is growing gangbusters, and then they've got the frontier models as well in Gemini, right? They're the only ones at the moment that are occupying all layers of the stack. Now, that'll change because all the other big hyperscalers are busy building products in all layers of the stack, right?

37:11Piers Curran: And who knows, in like 10 years time, you're probably just going to have four or five companies that just do everything. Like NVIDIA, they're going the other way, right? Yes, they've got the chips, but they're building, buying Hugging Face and, you know, building models themselves now, right? But anyway, obviously, you know, for every player in the game that stumbles, there's a competitor that benefits. And so whilst NVIDIA and Jensen, you know, do their thing. Well, Alphabet are nicely placed, of course, from a TPU perspective. They've just launched a couple of new products on that front, TPU-8TX.

37:52Piers Curran: That's for like confidential computing. They pair that up with AMD's CPUs. And then you've got something called TPU-8i Max, which is a dual chip module optimized for long context inference, things like maths and code generation and stuff like that right so um blueberg intelligence projects external tpu hardware software revenues are basically bloomberg are saying that that they're going to cross the 135 billion dollar mark by 2028 so that's google selling chips um you've got companies like jump trading um which i'm sure many of our listeners will be familiar with, client of ours, shout out to Jump, but they've disclosed deploying thousands of these TPUs.

38:40Piers Curran: And actually, Jump is saying that they're about 30 % lower cost versus NVIDIA. So Google have been placing themselves in the market at a cheaper price point. But look, whilst all that sounds good and wow, Google are going to come in and soak up all the market share, there's no market share it's all sold out you know what i mean so in the in the short to medium term none of this matters for nvidia in terms of them selling gpus unless jensen gets arrested otherwise fine google crack on but are they going to properly benefit from this nvidia wobble i don't think you can necessarily connect those two dots.

39:26Yeah, I think the looking at politics as a precedence, it's the big thing until the next big thing comes along. And Jensen is the big thing of this week. And just like politicians always do. And I think COVID, the post COVID environment, which is very lumpy for many politicians around the globe. I think the strategy that you ordered probably the best results was just to stick it out. And in time, the next concern arises and the previous one is forgotten. And one of the things I was trying to sort of put together is in my head, there's always this analogy I have of cops and robbers. And the problem that you have when it comes to say, in this case, what I would say is like, you know, these frontier labs are like the robbers, in the sense of their ability to move at the speed that they can and to be at the cutting edge of technology and change and adaptation to take advantage of this opportunity.

40:29Whereas regulation moves at the speed of congressional committees, which is at a snail's pace at best of which most of it never comes to fruition anyway because of the composition changes just like we've had in the midterms where things just get stopped started changed reinvented and by the time it's come the robbers are already you know around the cunt around down in the next town

40:57Piers Curran: yeah absolutely and and anyway thank god jensen's thinking well thank god goblins are still advising Anthropic to go ahead with their IPO because that'll be the next big story in a couple of weeks time. That is very, very true, actually. Yeah. And then how long have we got to go? I mean, it's just before Thanksgiving, isn't it? So he's got to hunker down. Yeah, less than two weeks. Just get through. By no way am I saying what Jensen did was right. Yeah. His actual thing that he did by no means am I condoning but by point being that yeah let's just see how this plays out and washes out is he indeed too big to be fired let us know what you think drop us a comment both that and also Ronaldo or Messi who is the greatest of all time let us know in the comments but cool thank you very much Piers see you next week pleasure see you later