Jensen Huang Resigns: NVIDIA's Succession Crisis and the Fate of the AI Supply Chain

0:06Ben Bajarin: Hello, everyone. Welcome to another episode of Circuit. I am Ben Bajarin.

0:13Jay Goldberg: Greetings, programs. I am Jay Goldberg. Ben, I'm a little nervous about this episode.

0:18Ben Bajarin: man this is like head shakers scratchers size all of the emotions and feelings as all of our listeners i'm sure know exactly what we are going to talk about on this early recorded and then early published because of popular demand episode of the circuit um yeah i'm nervous because we're

0:45Jay Goldberg: going to be delving into territory that is not immediately adjacent to semiconductor technology and i'm worried if we do too good a job we're going to end up like getting invited to go on

0:54Ben Bajarin: joe rogan and talk about conspiracy theories um okay well before we do that you you are participating

1:04Jay Goldberg: uh yes go ahead yesterday uh so we're recording this thursday november 12th Wednesday I was at the Silicon Catalyst event and I just wanted to give them a shout out because I think they're doing good work. Silicon Catalyst is a venture incubator for silicon semiconductor deep tech startups and they've been around for a while and I hadn't caught up with them in a few years and I was very impressed by their event. It was at the Computer History Museum in Santa Clara and it was standing room only i didn't get a chair the whole time it's so crowded uh room was full of people i think they had 20 something 30 companies in their current class they i think they their portfolio is worth eight billion dollars now they have branches all over the world like this is a like serious stuff they've come so far and i just wanted to give them credit if if you anybody listening is at a startup that is touches on things that we talk about in this space you should go look them up uh they're doing good work very cool yeah i've been to

2:15Ben Bajarin: not in recent years but a couple of their events way back um i think i spoke at one like 2014 2015 uh yeah they've been around a while back in the day um cool good stuff i i think we have multiple listeners from them to hit me up on on link dick all right um so the news of the week the shocker heard around the world i'm not gonna say what's heard around the world politically correct reasons but the uh the news that uh i don't even know how to say is that Jensen admitted to being unfaithful to his wife. This was speculated, I think sometime last week, some news broke, which might have been leaked by the government that they were concerned he was having an improper relationship with a woman who was tied to perhaps the smuggling of Chinese chips or any number of things.

3:17Anyway, we get fast forward.

3:19Ben Bajarin: That was, I mean, a whole slew of things happened Sunday. Fast forward to today, Thursday, November 12th, and he has been, he has resigned. We assume this was voluntary. We don't know, but we don't know. But he's no longer CEO. It broke, I believe, yesterday that he had an affair with a direct report, which didn't look good. More news has come out recently that there was more NVIDIA employees, at least one other one. Anyway, all things bad at the end of the day in what is just an unbelievable self-inflicted wound. The silicon industry leader at the moment, spokesperson, Jensen Huang, is no longer CEO of NVIDIA.

4:12Ben Bajarin: Your thoughts?

4:17Jay Goldberg: yeah it's just it's it is it is very hard to grapple with it which is surprising because both you and i have been probably talking about it non-stop for the last 48 72 hours um it's pretty stunning even still like i i feel like the world is smaller now without him in it it's like just suddenly semiconductors are just semiconductors we've we've lost our our little bit of glamour that he that he provided exactly right um i'm i'm not going to excuse his behavior i'm just but like the reality is like this is less than it was yesterday i think i and i i actually think that there is a fairly this is a fairly common sentiment like uh you know i think there are a lot of investors who were not particularly happy with the way this all played out um and not just like lots of people not just investors but they're the whole thing's a mess

5:16Ben Bajarin: that's so it's very sad yeah yeah so so let's just okay a couple things right to level set so um the only comparable i have to this is steve jobs and you know these are different situations steve Jobs, uh, unfortunately came down with cancer. Uh, I, to be honest with you, I don't entirely think, um, you know, Steve Jobs had a set, uh, uh, succession plan, but I think he was able to come up with one on the fly and appoint Tim Cook. And that kind of steadied, right. A lot of the, the, the storm both in between as he was dealing with the sickness. And then obviously, when he passed. Because people bring this up, they're like, why didn't he have a succession plan?

6:05I was like, well, what would he have needed one for?

6:08Ben Bajarin: I get you want to have one as a backup plan, but he had no assumptions he was not going to be the CEO for a very long time. Who was he in this case?

6:17Jay Goldberg: Jensen, sorry, Jensen, yeah.

6:19Ben Bajarin: And so it was kind of like, again, I'm not faulting him for not having one. Obviously, you should have one in a worst-case scenario, but he fully expected to run this company for a very long time. And so there's that part of it. But very similar, in my opinion, to Steve Jobs, he was the North Star of the company and the industry at the moment, right? We have looked to him to be the tone setter. He does these extra long keynotes all over the place, talking up AI. I mean, people, he had that gravitas around him, very much like Steve Jobs. So it's that type of a personality, That type of a who did this person mean to the culture and the vision and the leadership of the company in that sense is what has been lost here.

7:07Ben Bajarin: Now, obviously, Jensen isn't dying like Steve Jobs was, but we don't have any indication that he'll even be within the periphery of this company. And so my point is the comp we have is the magnitude of this moment, in my opinion, is very, very similar to who was a North Star of their company and the industry at the time. Yeah.

7:30Jay Goldberg: I mean, if you think about it, like let's not forget Jensen founded NVIDIA. He had some co-founders, but he's the founder and has been there all along, the only CEO they've ever had. He's arguably the most successful entrepreneur ever, right? And it is very difficult to think of NVIDIA without thinking of him. That being said, I think corporate governance 101 is every company has to have a succession plan. Man's flying all over the place. Private jets are dangerous. They have always needed a succession plan. And it really looks like they didn't have one. And that is a failure of the board. Now, maybe they didn't have one because he resisted them putting it in place.

8:25But, like, it's just basic, you know, this is how boards are supposed to be run.

8:31Jay Goldberg: You're supposed to have – boards don't have a lot of things to do, but, like, this is, like, one of them. One of them is, like, have a succession plan just in case.

8:39Ben Bajarin: Yeah. Yeah. I mean, I think too, like, I mean, again, right, I think there's just the gravitas of the situation in that he did sort of, I mean, I think orient this company kind of around him, right? I mean, I think you look at the way he managed it and it was really built around his kind of management style. Now, flat meant flat, but it meant that he oversaw his number of direct reports that then went and ran the company. So, you know, again, it's one of those things where NVIDIA would not have got to where they were today without world-class engineers, without people that Jensen trusted to run and execute his plan and vision.

9:30Ben Bajarin: That part, I think, continues, right? The technical merits of what they're doing. But I, again, I come back to all these same conversations we had when Steve Jobs was absent from Apple, which is can the culture proceed? Can the employees work the same, be motivated the same under the same central vision for the leader, right, who they work for? I think these are reasonable questions that now become issues we didn't have to deal with before, right, in terms of the future fate of NVIDIA. and that's why i say like yes right governance wise etc but again like i think i don't want to downplay how important he was as a icon if you will amongst other things to to nvidia and to this industry moment we're in yeah i think there's there's two pieces to it one is he was

10:24Jay Goldberg: he's central to the company and then he's central to the industry as well right and And most of the questions I've been getting today from people have been, what happens to AI now? And on that part, I don't think things change exactly. Like people are still going to – Anthropics are still going to go public next week. People are going to buy a lot of GPUs this year, spend a lot on CapEx. But I definitely have to think like we lose a little – AI loses a little bit of momentum here, a little bit of energy. Yeah, I agree.

10:58Ben Bajarin: Yeah, we don't have a replacement for that person, that personality. That kind of, again, North Star for the, we don't have a replacement for that. It's certainly not going to be Sam Altman or Dario. It's not Lisa's style to do this. Google really doesn't have anybody. Again, they're a hyperscaler. I think the question I keep coming back to is two things. is, okay, we could agree that NVIDIA's strategy isn't necessarily going to change. They're going to keep the roadmap that they're doing. But I guess I would have a question as to, like, for example, would the next CEO, or even in this moment with Colette Kress being interim, was everybody on board with the way that Jensen was using his balance sheet?

11:50Ben Bajarin: And does that now have impacts where they may or may not be backrolling or being the capital to customers, neoclouds, et cetera, the way that they needed to, at which point you could be, I guess, a little bit concerned about the buildout is sort of what I'm saying.

12:11Jay Goldberg: So I think I'm going to start from a place that I think AI demand is real and growing, and that's driven by the Frontier Labs more than nothing else. The frontier labs keep getting better. Demand will keep going. And if that continues, then somebody is going to build the infrastructure. And whether that's Amazon or 200 Neo Clouds, that's going to change now. I think you're right. I think the ability of NVIDIA to play the role of central bank to AI, that goes away. This$500 billion platform they're talking about, it has six of the biggest, most powerful private equity funds slash banks in the country or the world signed onto that press release.

13:04Jay Goldberg: I don't think anyone else – there are very few other CEOs who get all six of them in the room together to agree on that. and without Jensen I don't think the deal goes forward to add plus there's internal resistance at NVIDIA plus let's not forget that deal has not been finalized yet all they had was a press release I don't think they had a term sheet so that's not going to happen so that means there's a whole bunch of companies out there who are thinking they're going to get some funding from Jensen to start their NeoClouds all over the world I don't think that's coming to fruition right but amazon and google and microsoft are still going to build data centers and anthropic and open ai are still going to build data centers yeah right so we see it's a share shift i think for nvidia i i think there is a real problem there the neoclods exist because jensen willed them into existence he wants to have 300 small customers rather than three large ones and I think they will continue to provide backstops and some form of credit to NeoClouds but on a much smaller scale.

14:18Jay Goldberg: And that probably ends up meaning that the hyperscalers get more powerful. And then not for nothing, hyperscalers are very, very good at negotiating prices and I'm sure if I'm the purchasing manager at one of them right now, I smell blood. And the next time I have to go put in a purchase order for NVIDIA GPUs, I'm going to turn the screws a little bit. I'm going to push back on a lot of things. Because what happens, like that's a really, that's a tough look, right? Because right now they're sort of, it's interim CEO at NVIDIA, they're in turmoil. They don't want to lose business. The new CEO is going to come on board.

14:58Jay Goldberg: The last thing he's going to want to, he or she is going to want to deal with is, uh-oh, you just lost Amazon. So hyperscalers get a lot more power now.

15:09Jay Goldberg: And then on top of that, I'm sure the hyperscalers are also sitting there thinking, we've always wanted to have two sources of GPUs. We really need to accelerate that now. This just proves that point more. Let's go see what Lisa's doing.

15:27Ben Bajarin: Yeah, or can we get more TPUs? Now Google has picked a great time to be wanting to sell TPUs, right, publicly. But I think my bigger kind of point is, and this is a point for the investment capital to build out. Because I obviously agree with you. You know, we think demand continues to go up and to the right. But I think a lot of the reason why the balance sheet stuff that NVIDIA was doing was important was because they were helping people get the capital that they needed to secure the parts of the supply chain that they needed, whether that was, again, engineering talent, construction talent, people to go and do, you know, building and construction.

16:10Ben Bajarin: Because again, you've got to buy wood in advance. Like there's just a lot of down cycle financing that needs to happen. And so barring that, barring your, you know, NVIDIA or somebody co-investing with you in that, you are now just left to take all this risk yourself, which, again, maybe they're convinced the same way, right, that we are, and it's fine, but I just worry a little bit that kind of they're, they slow their roll, again, not necessarily because they don't think that this demand profile is there the same way, but they're not getting the investment capital, right, they need to help hedge some of those risks just in case, but yet they need to pop down a bunch of cash to go and secure everything they need to build those data centers.

16:55Jay Goldberg: But I think, so I think NVIDIA's role in that is a little different. What NVIDIA has basically done has, has provided some form of credit support, a backstop investment, whatever. And what that's done has, that's allowed the need that hasn't sort of made the, that hasn't, They haven't funded all of it. But what they have done is given enough support to the neoclouds that it lowers their cost of capital to the point that the project is viable. And so absent NVIDIA providing that support, interest rates that they have to pay get too expensive to support or they can't raise enough equity for their portion of the deal.

17:39The mechanics of raising that money become much harder to the point that they're no longer able to raise that money.

17:49Jay Goldberg: And so it's not entirely clear to me who could step in there. And AMD can't. They don't have the balance sheet or the interest. the sort of typical Wall Street sources. Like what we're really solving for is this gap between what the capital markets is willing to pay and what Jensen thinks the technology is worth. Sure. And he bridged that gap so the two could meet. And if NVIDIA is not there, I don't think the capital markets, anyone in the capital markets is in a rush to step up to that role.

18:22Ben Bajarin: That's agreed. That's kind of my point is if we needed something like this in order to secure the trenches of the supply chain to build out two years from now, which is kind of where you got to start doing these plans, like what's the threat to that if, again, right, that Colette or someone else who steps in decides to not use balance sheet the same way? Because I think someone needs to do it. And sure, maybe Google, maybe the hyperscalers can because they need it too. But I just – I keep coming back to someone needs to help play that catalyst because, again, these are very big and expensive checks you're building for something that's going to exist in two years.

19:08Jay Goldberg: Yeah, I think – I don't think NVIDIA stops providing support, but just nowhere near as much. They scale it back.

19:18Ben Bajarin: Yeah, which again then opens the opportunity for other players, right? I think the NeoClouds, there's been sort of the rumor that they've wanted other sources than just NVIDIA, right? So there's that part too, right? But again, I sort of keep coming back to like the conversation I was having, you know, in the back end of this with people. It's like, what does this impact to NVIDIA demand, right? Because right now, every wafer is accounted for for the next at least two years. NVIDIA is a huge chunk of that. So the question is, is there anything that threatens that demand model? At which point, somebody else gets some of those wafers.

19:58Ben Bajarin: And if so, who are those people?

20:03Jay Goldberg: Yeah, I think it is – NVIDIA is hobbled here but not knocked out. I think their numbers, like you said, is everything sold out for the foreseeable future. But it gets harder, right? Who's going to go negotiate with TSMC? TSMC, they're pretty good at signing deals too. They're going to be a little wary here about whoever comes next and how much capacity they want to assign to NVIDIA. um you know there's you know i think jensen through his relationship and he's obviously very very popular in taiwan was able to get a pretty good deal for for nvidia at tsmc but he was bumping up against the limits of that this year they weren't going to give him anymore and you know i i think it definitely i puts tsmc very much in the kingmaker position they're the ones who were deciding this year and next who gets the extra allocation and i think you know now i think about it i think if they see nvidia really threatened they'll throw them a bone and give them extra capacity next year um yeah but it's it's uh

21:29Ben Bajarin: Well, I say that to say if I had to pick foot to the fire, who would get some of that if demand goes down? It's Broadcom, which would I think then benefit Google and TPUs and AMD. I mean, they're your two second largest customers after NVIDIA. I mean, well, Apple's there, but different wafer needs for monolithic versus HPC. um so that's that's again right it just comes down to a choose your assumptions on your wafer model and who you think gets uh gets that share if any anything happens to nvidia's demand profile which again i'm not saying it's going to happen i don't i'm not thinking that's the case right now but it's certainly the question to watch long term um which again just comes back to kind of the strategy side of this execution as you rightly pointed out the negotiation side of this like Jensen played a big role in securing those deals and doing those negotiations and negotiating pricing.

22:31Ben Bajarin: A lot of times you were up against him. It wasn't a random salesperson.

22:37Jay Goldberg: That's right. A lot of these deals are sort of CEO to CEO. Some of these sales are massive. So I can easily see that the real negotiation with Amazon is Jensen and Andy, Jensen and right it's at that level and you know it's that's it's harder now i'm i mean i imagine that colette is a perfectly capable fierce negotiator um but she's you know kind of fighting with a arm tied

23:08Ben Bajarin: behind her back yeah so so let's talk a little bit about you know where this sort of goes from here so colette is interim which i think is a good choice um the street knows her right it's not you have an unknown candidate. I had originally thought it would be Ian Buck, but it makes sense in the interim because Ian's never been on an earnings call. Never been on an earnings call. Not something he's had to do. Not something I don't even know if he's comfortable with. Even though I do think he should be a candidate for this job given how long he's been there and he does know these customers and whatnot.

23:42But this is basically a process.

23:48Ben Bajarin: Right. They're going to explore, this board is now going to explore all new, uh,

23:56Ben Bajarin: third party candidates to which I'm going to let you throw whoever your dream is. Like we've been talking about this. Like there's not, there's not a lot. It's a very short list of people who can run a company like this. When, again, there's a lot riding on NVIDIA, um, or an internal candidate who, again, I would think would be, you know, Ian Buck would, would make the most sense. so yeah so how do we see this process playing up and then as you've had to do some of these things before with boards and whatnot like how long how long could this take like is this could this take a really long time or is this the kind of thing they might want to do quickly i think it can happen

24:33Jay Goldberg: fairly quickly but not days it's going to take a couple months okay right because the board has to figure out what they're looking for they have to i'm sure they've already reached out to recruiters but that takes some time to get them set up. They have to figure out the candidate list. So give the board a week or two to just figure out who they want to talk to. They'll bring in outside experts to consult with them, but some of those outside experts will themselves be potential candidates. So, you know, oh, we have the holidays coming up. I mean, I think they get somebody Q1, like March is my guess.

25:14Jay Goldberg: I'm going to make up a number. March. Let's say they get somebody by March of next year. That's, you know, maybe they can squeeze it a little bit faster than that. But I think March is sometime around. Maybe for GTC they get the new CEO. That's probably as fast as they can go.

25:31Ben Bajarin: I mean, something like having a goal for GTC would make sense? Because right now I'm thinking through like their event strategy, and I don't know what they do, right? They've got GTC DC, which was coming up in a few months. I don't know who does that. you usually ces was prime jensen stage for ces do you even send someone to ces at this point if so who i mean it's certainly not going to be the draw it was so that's an issue and then yeah and then gtc is the big one right all the news all the announcements the roadmap everything and so like that's going to be tough right without somebody at the helm let alone jensen but just somebody who can come out and represent the company yeah but it's it's tricky because the

26:13Jay Goldberg: board can only move so fast sure i get that i get right no no but i mean it's like it's like there's we get into some pretty serious legal territory here because the board has to be shown to exercise thorough due diligence and consider all the candidates right this board is now under immense immense scrutiny um and they're going to have to demonstrate that they conducted a deliberate and thorough process, which is going to slow things down. And I don't think – the lawyers are in charge now and the lawyers don't care about CES or GTC for that matter. This is true. They're going to dictate the pace, right?

26:51Jay Goldberg: This is true. Right. And then let's also not forget the board has other things on their plate as well, such as – what are we at? Four, five ongoing federal investigations, congressional hearings coming up. right um those are serious serious matters uh i don't think we have a good sense of how what the board is doing about them but they have to do something yeah um

27:20Ben Bajarin: well yeah and lord knows what other information comes out are there more shoes to drop who knew what like the geopoliticalness of this is gonna get messy you've seen senators and whatnot come out and tweet like we better make sure we never sell chips to china like yada yada yada which which fine um let's come back to that because i got theories on that okay right but like but i think

27:48Jay Goldberg: i think what's interesting here like one of the things the board has to be extra double super careful of is especially with internal candidates let's say they go through this process they get somebody in march from an internal candidate and i'm not going to name names i'm just going to say generic internal candidate then in april the congressional hearings begin or the investigation comes out maybe and then it turns out that that person knew about something that yeah of course they exactly they have to know who knew what absolutely right and it's going to take them a while to figure it all out

28:30Ben Bajarin: the more we talk about this whole thing I just I can't believe it I just it's really it's the self-inflicted wounds of all of this that just you know there was nothing we could do about Steve Jobs cancer you know like that was a tragedy in and of itself this was yeah that was a whole slow tragedy go ahead so so

29:04Jay Goldberg: yeah i i so this is this is where my conspiracy theory starts which is jensen has obviously made a long series of very poor decisions about his personal life um but i am i have been very suspicious of this since the so it all started with a New York Times story just about a week ago that that claimed he had an affair with an alleged Chinese spy and if you look at that story it was reported in Washington DC that wasn't some reporter in Beijing who had a source in China's MSS This was a story coming from Washington, D.C.

29:50Jay Goldberg: And that smells a lot to me like a political motivation for embarrassing Jensen. And then the way in which it sort of rolled out from there was, in my mind, almost certainly an orchestrated campaign to remove Jensen from office, from his position. Right? It just – it really, really smells like a Washington political operative maneuvering, using the press, leaking to the press selectively. I mean I was – like I've been saying to people all week long, like there's going to be another shooter drop because that's how these things always go. You create a sense of urgency, a sense of panic. Oh, we have to do something.

30:35Ben Bajarin: You have to, right.

30:36Jay Goldberg: And then just as they're formulating their defense, the other shoe drops and then boom, everyone's so all wound up. We got to do something. And then, right. And like I had that conversation with a reporter. I was talking to a reporter about it yesterday. And like we got to the end of the interview. And just as the interview was ending, the other – the next New York Times story about the other affair with the internal direct report came out. and like i looked like i was like i could see the future like because there was the other shoe right there and i i think it's a shame and i there are you know there i i you know look i i don't know anything more about what goes on in dc than anyone else who reads the newspaper every day but there's lots of clearly jensen has made a lot of enemies in in dc um there are a lot of people who didn't like the way that he was lobbying Congress to loosen restrictions.

31:36Jay Goldberg: Like not the idea of loosening restrictions, but the way in which he went about lobbying people, rub people the wrong way. There's this, you know, there's an element of him being very close to the president, which I'm sure made other people just jealous. They didn't really care about the policy. They just didn't like, they were jealous of that closeness. And then there's another element, which is he could just be collateral damage in a fight between two politicians, right? He's a little closer to one, so that guy's enemy undercuts Jensen to undercut his political enemy. There's all of these things playing out.

32:09Right? And so, you know, like, it's just, it is super frustrating that that's how it played out.

32:23Jay Goldberg: And I don't know what he could have done differently because it certainly looks a lot like – I don't know if they caused it, but certainly the people in D.C. who feel like they've benefited from his removal are the people who are most hawkish on –

32:37Ben Bajarin: I've seen a lot of these – Restricting chip sales to China and the like. All these conspiracy things. Again, I think you see what some of the senators have put out today. You could see them being like very, very angry about we should not be giving any chips, right? to china and now you heard them saying we need to cut off china's access to even get access to our chips via clouds any cloud regardless of where they are like there is a lot of that there's the as somebody was telling me there's the there's the china hawks and then there's the china hawks like really big ox and there's that's differencing opinions so I guess I just keep coming back to the like I mean again I guess I don't I shouldn't give any intelligent credit to politicians probably but like you have to realize how important this company is also to your national security not just your competitors trying to get access like your national security like why would you do something that jeopardizes this like it's I don't know

33:51Jay Goldberg: Yeah, yeah. And there are all kinds of reasons why. I understand very well, personally, how much China is a rival to the United States, a significant adversary and competitor. But maybe this isn't the right way to go about confronting them. Right. It's it's not clear to me that restricting chip sales to China is ever going to work. Right. There aren't enough inspectors in the BIS inspectors in the world. Right. BIS is woefully underfunded. Right. So even if we have super strict rules, Chinese companies are still going to get chips. Right. If they can't get them in Malaysia and Singapore and Korea and Japan, they'll get them somewhere else.

34:40Jay Goldberg: Right. They've like they're just right. Right. There's, you know, there's so much interest. There's so much demand there. And there's what, 2000 BIS inspectors total in the world? Or no, that's a whole agency. Like, it's just there's a mismatch there. All right. So if that's the case, then. Right. Limiting. Limiting access to chips is not it's not effective and it's not going to actually it's probably not the best way to do it. There are all these bills that are going through Congress now, varying degrees of harshness, varying degrees of direction. And it's clear to me that some of them are better than others.

35:21Jay Goldberg: Just cutting off China from chips is probably not the right way to do it. There's the other part of this.

35:28Ben Bajarin: To the degree that China, I mean, again, retaliates or could retaliate. Like there are a lot of components in the supply chain that we need that come from China to even make the boards that those chips go on. And if they felt like it, right, they could impose a lot of other pain. But all that would do is prove that they were getting stuff illegally. Like, hey, we're mad that you cut off our illegal attachment. And so now we're going to figure out how you can not get materials you need for PCBs and substrates and trailing edge components, et cetera. So I don't know. I guess I'm a bit worried about how this plays out from a government standpoint as well, just the variables that are in play.

36:26Jay Goldberg: Yeah. Yeah, it's – I think – I don't know. We could get very tinfoil hat over this. I don't think anything is clear yet, and I think – I don't think all of the most hawkish people's plans are going to come to fruition. But I think it is safe to say that we will see more restrictions on chips out to China. But let me ask you this. Okay, so if this is the line you're taking –

36:54Ben Bajarin: Which is not great for NVIDIA. Why aren't we cutting off CPU sales from AMD and Intel? Because those play important roles in AI inference. And why are we drawing the line just with GPUs?

37:22Jay Goldberg: I'm not even going to respond to that. I'm just making the point. have to say something insulting about politicians and

37:31Jay Goldberg: let me i would i will i will say i think it's safe for us to assume that there is not a high

37:36Ben Bajarin: degree of but again at the same time like a crowd let's just exclude china from this right people were like why does singapore and why does malaysia and all these other places have a bunch of chips i'm like okay that's a fair point but at the same time every nation not just china and the united States is going to be building data centers for compute. The hyperscalers want to serve businesses and people there in their regions. They need those data centers. Something's got to fill those. Like, sure, it's in close proximity to China. Okay. But at the same time, like the nations are going to be building these things and inevitably we're going to put NVIDIA GPUs in them.

38:12Ben Bajarin: So you can't just say, hey, you can sell to nobody, but our friendlies, Europe and the United It's like that's just like these companies need to be able to sell their chips to other places, you know?

38:32Jay Goldberg: It's even worse than that because you just – you said Japan. Like there are serious Chinese-affiliated data centers in Japan, right? That's like – I don't know that it's larger than what's going on in Malaysia, but it might be, right? There are Japanese sort of NeoCloud equivalents, which have sprung up very quickly. Right.

38:53Ben Bajarin: Exactly. They're serving Chinese customers, right? And so where do we go along? We just can't export chips anywhere? That's what this week has been like. So here we are. All right. You had a name you said you were going to talk about for NVIDIA CEO that you're floating to me. so well

39:14Jay Goldberg: because that's

39:15Ben Bajarin: she's not going to do that

39:18Jay Goldberg: like you know my crazy dream would be that Lisa Su takes over I had said she's not going to do that no

39:26Ben Bajarin: Pat Gelsinger said that'd be pretty cool but I know people I know he's not really an option you know and then anyway alright alright let's talk

39:42Ben Bajarin: so so okay so i've actually spent time lately they're like okay for the first time ever on

39:48Jay Goldberg: on polymarket tracking tracking the odds of right yeah oh yeah yeah there was there was this was up a few days ago so the leading candidate right now is uh is ian buck yeah because he was there

40:01Ben Bajarin: forever colette is number two the uh renee haas from arm is number three i have no idea where that came from that's right that's right okay like i guess coming back and they don't come back number four is jensen huang wait wait what's the poll sorry what's so what what's the percent though for jensen like what's that yeah i mean it is it is all right yeah okay six percent that's oh man right it's higher than tim cook tim cook is five

40:33Jay Goldberg: percent and the president wants right so so i mean here here's here's the crazy plan here's how that would take place is if you find that are there are enough shareholders that are upset again as long as there you could run a proxy campaign like to bring him back some you could

40:49Ben Bajarin: You know what I mean? Like something illegal. As long as that's the case, sure. I mean, yeah. Yeah. Anyway.

41:02Ben Bajarin: It's possible though. Yeah. I guess like I saw somebody post something that he owns. It's completely nuts. As long as one person owns more than 3 % or a group of set number of people, they could appoint or make recommendations to appoint board members who could then increase voting share, oust current board members get yourself like anyway i don't know dude i'm just telling like this is a popcorn situation this is nowhere near from over all sorts of other news and drama is going to happen all at the time where like one of the things i wanted to talk about on this show prior to this whole situation was like you basically have increasing degrees of market slash investor sentiment that none of this is going to last beyond like 2029 like like that was what was going into this situation already and now you have this and so you're just like well goodbye everybody like this is like this is uh just an absolute cluster on so many levels at a time where everybody was already reaching the conclusion that compute was a solved issue ai compute is a solved issue and after 2029 nobody's gonna need this degree of compute anymore at all and now this and now this which you know before we talk about that every week we always say at the end of our show we'll see what next week brings and here we go just alright let's talk about

42:38Jay Goldberg: yeah

42:39Ben Bajarin: sure did not I did not have Jensen sleeps with a ledge Chinese spy on my list of risk factors lay lay the uh the framework on that and then uh yeah

42:58Jay Goldberg: core weave and nebius both reported today or no this week and uh the market's verdict seems to be core weave good nebius bad core weave was up 20 nebius was down close to 20 um

43:16I think the issue with Nebius is very straightforward.

43:21Jay Goldberg: They had to delay the start of a data center. It's in construction. The completion date and the powering on date got pushed into next year, so they had to lower guidance, and people didn't like that. That's pretty straightforward. The reason Corweave was up so strongly is less clear to me, and I think what people saw there was the numbers were good all around, and Corweave in particular had been so beaten up in the last three weeks. So many concerns about its credit quality, all that stuff, and its ability to raise capital. And they came out with a set of very strong numbers, and I think that, I guess, to some degree assuaged concerns.

44:08Jay Goldberg: but I think in both cases what it shows me is that what really matters most to these to these what matters most in these companies is their ability to bring on large projects for large customers it's less about the day-to-day of their business it's just their sort of construction schedules and power cycles I mean core we've had a whole bunch of other things but like I it's hard to tell how much of their their commentary is about the core business and how much of it is just around sort of very anecdotal things, like they're raising prices on things. But that's not really the core business. So NeoClouds are very confusing.

44:45I guess without NVIDIA funding 200 other NeoClouds,

44:51Jay Goldberg: maybe things get a little bit easier for Core Web and Nebius because people still want their compute and they don't have hundreds of competitors anymore. But it's still, you know, these businesses are very much live or die by the large contracts they have. I think they said it was like 130 ,000 GPUs.

45:17Ben Bajarin: Sorry, they have 130 ,000 Hopper GPUs? Hopper. Yeah. Okay.

45:27Jay Goldberg: They have 130 ,000 hoppers that are on a contract.

45:30Ben Bajarin: Right. So the assumption was they might be able to charge more for those than they did at the time.

45:35Jay Goldberg: In I think Q1 of next year.

45:37Ben Bajarin: And is that a fair thesis?

45:42Yes. Yes.

45:46Jay Goldberg: So the pricing bears that out, right? The contract price for – the spot price for Hopper is up this year. I don't know when the contracts were started, but it's safe to say that the contract prices for Hopper are up. And so in theory – so actually, let me take it back. Contracted prices are almost always lower than spot prices. And so Hopper prices since the time of the contract signing have come up a little bit, but then you factor in the discount. you're actually they're actually going to have 130 000 gpus come on stream in q1 of next year that they can put on the spot market and they can raise prices yeah by what is effectively a pretty hefty amount that certainly if if if they came off contract today and they had to sign up people that would be true um i don't know what hopper prices are going to be like in six months that's the real risk there um and that's kind of what i was getting at which is they have lots of little examples of like oh we're raising prices we're raising prices do you think they i know

46:52Ben Bajarin: the contract and stuff is still like when they're different yes but you know there had been sort of this discussion we talked about it too where parts of their fleet they were leaving open to spot because it's becoming kind of like an auction because of demand does core weave give any indication that they're kind of maybe going to favor like one of these or the other leaving some first spot so they can get higher revenue versus long-term contracts which again are never not going to be at the same the same rates okay

47:29Jay Goldberg: loosely speaking yes they kind of frame it a little differently but that's the basic idea which is prices for all these things have gone up let's just let's let's stop having such so many contracts. It says owner's contracts, let's have more flexible, shorter-term contracts, higher prices. The other key thing that they claimed was that they're moving away from this form of structured contract with fixed pricing and more importantly, moving away from these long-term contracts they have with the hyperscalers, where the hyperscalers fund the capex and instead they're moving to uh something that's much more like prepayment right so you don't necessarily sign up for three years but you pay you you sign up for a year and you pay all of that up front and that that's seen as as a positive right i i would worry though that um yeah they're able to do that right now i don't know how how sustainable that will be going forward right they They don't need – what they're effectively losing is they're losing the credit support.

48:39Jay Goldberg: Right now if they borrow money, they use their customer's credit to guarantee the loan. That's how they're able to fund it. They're saying they're moving away from that and instead they're going to sign up people for prepayment. Sure. You pay a year up front and that's – they'll raise the money that way to be able to get the loan they need, which again is possible. I don't know how long that, like, I don't know if they'll be able to do that forever.

49:06Ben Bajarin: I mean, the thing is, too, is like, look, you look at how both these companies position themselves as hyperscaler alternatives, which, you know, I don't want to necessarily say either of them fall in a hyperscaler camp, but they put themselves in this conversation as an alternative. To be an alternative means that enterprises or business customers come to you directly for these things. and if you're a business, a SaaS company or whatever, right? You're not coming to them and going to pay GPU auctions to serve inference. You'll do this for training, right? It makes sense. Training are short worth slows.

49:39Ben Bajarin: You just need to suck it up, put those costs in. You got to do it. If you're inferencing something, right? I mean, you're providing a software as a service or an AI, whatever, as a service, which is what they're trying to say. If you're an AI native company, come to us. those companies are not going to have variables in their margins for spot pricing and gpus they are going to want long-term contracts and predictable pricing and so my point here is that this that that point is is doing spot pricing is fine for training but if you want to be an inference cloud that's not going to work and i just those those tensions sit in my mind about what they say they want to be versus how we know customers use them which is either a hyperscale proxy or for training, not for long-term enterprise viability of inference.

50:26Ben Bajarin: And those are the kind of long-term things I would rather see them prove that they're getting the long tail of the enterprise or again, have some form of economics that makes them attractive as a hyperscaler alternative for AI lab or AI SaaS startup or whatever, right, that needs to come and pay for compute and do so with predictable pricing.

50:57Jay Goldberg: I think they would maintain that they are, in fact, doing that. The problem is that piece of the business where they're going after the long-tail enterprise is very, very small relative to these massive contracts they've gotten from the hyperscalers.

51:13Ben Bajarin: Sure, of course.

51:14Jay Goldberg: And we've talked about this before. And like you get a whole bunch of money, how do you decide who gets which capacity? And like how do you – right? It's so easy to get distracted by the large amounts and then you have to still build the business. And so there's a sort of incentive timing problem there that makes it tricky. But they're both – I think it's safe to say they're both working towards that.

51:38Ben Bajarin: Position themselves, prove that they can get the type of customers who are long-term durable in these contracts, that they're not just a training cloud, that they're not just a cloud for the hyperscalers. And these are the little nuggets that I'm looking for either on calls or whatnot where we can judge that they're getting businesses that fit the positioning that they have, which is hyperscaler alternative for enterprise software.

52:10Yeah.

52:11Jay Goldberg: And I will say this. Like I said, they're both – they're all working. They're both working towards that. And Corrie has now gotten their customer concentration down to – the two biggest customers are down to 60 % of revenue, which is – it was 70 % last year, 75 % of the IPO.

52:26Ben Bajarin: So they're heading in the right direction. It's just going to take a while. I didn't see what Corrie was at today. The market is still just doing all sorts of market-y things. But like I said, at the end of the day, there's still just doubt that any of this is going to be around. in two or three years. So that's the world we live in. Until you believe, the market believes, we are stuck in this cycle.

53:01Jay Goldberg: You know, the world's still going to need semiconductors. In two years, the AI boom goes away and we don't have Jensen and everything is less exciting. we'll just go back to being the normal semiconductor industry that you and I have covered for a very long time. And so, you know, like, look, if all the excitement goes away in 2029.

53:22Ben Bajarin: Semis have to eat the world, everybody. That's fine. We'll still be here. Semis eats the world everywhere.

53:26Jay Goldberg: And we'll see. You're right.

53:29Ben Bajarin: Okay. Well, yes, quite the week. Hopefully next week doesn't bring anywhere near this amount. Just to also say, Jay and I will be at the TomorrowX Summit in Austin.

53:46Jay Goldberg: You just jinxed us. You just jinxed us.

53:49Ben Bajarin: If anybody's going to be there, do reach out. We are hosting an after party on Monday evening. My company is with Funda, which will be fun. So we have an event. If anybody wants to come, just reach out. Let us know. And yeah, we will talk to you then. Have a good one.

54:16Jay Goldberg: Thank you, everybody. Tell your friends and tell your agents.