Nvidia Bull vs. Bear, White House AI Executive Order & Coinbase Prediction Market | Nov 20, 2025

0:13Welcome, everyone, to the Informations TI TV. My name is Akash Pastrucha. It is Thursday, November 20th. We have got a great show lined up for you today. First up, NVIDIA. The company reported earnings last night. Revenue growth accelerating once again to more than 62 % on an annual basis. Jensen Huang came out strong, asserting right off the top that he doesn't see an AI bubble. And so we are kicking off the show today with our first ever bull bear debate on NVIDIA. We will get to that shortly. We're also talking about the White House today. It is deep in AI policy discussions, and we have some exclusive reporting on a new AI executive order that the administration is working on.

0:54And we're then taking you to the land of prediction markets, where Coinbase is now expanding into, according to our reporting. And finally, we are talking about the gender gap in venture capital with the CEO of AllRays to talk about the organization's push to bring more equity to the venture capital world. It is a big show, and so let's get right on into it. NVIDIA shares were marching upward this morning as investors expressed optimism around the company's quarterly earnings. But today, we are hosting our first-ever bull-bear debate on the show. We are joined by Ben Beharin, the CEO and Principal Analyst of Creative Strategies, who is our bull for our NVIDIA debate, and also by Jay Goldberg, Senior Analyst at Seaport Research.

1:37She has a sell rating on the stock. One of the only, if not the only, analyst on Wall Street with a sell rating. Welcome to the both of you. It's great to have you here. Thanks for having us. Okay. So, are we ready? Are you guys ready for it? Bring it. The gloves are on. Here we go. Okay. So, here's how we're going to do this. Okay. I'm going to give you both 20 seconds for opening statements, okay, to tell us why you're a bull, why you're a bear, okay? And then we're going to talk along this issue on five verticals. We're going to talk about the technology, the macroeconomic conditions, the valuations, the financials, the circularity that is ongoing.

2:16And then we're going to end with the Jensen factor, okay? Because that is something that we don't talk about enough, I think. So Jay, since you are our bear, I'm going to start with you for your opening statement. Again, you have 20 seconds. Lay it out for us, and then we'll get more into the details. I think the bear case comes down to demand. What does end demand look like? And I think there are signs right now and a growing number of signs that it's getting harder and harder to generate the demand, to pay for the demand, to finance the demand for all the people who wanna buy Nvidia chips. Okay.

2:51Right? That's 20 seconds. I gotta stop you there. But we're gonna come back to the demand. That's a good point. Ben, your 20 seconds starts now. Bullish because we're in early stages of AI. mildly adopted across consumer and enterprise. And this is the single most computationally complex software we've ever seen in the history of software and development, which justifies massive amounts of compute. Okay, we're early. We're early, but is there enough demand? That's, or will the demand last? These are good thoughts all around. Let's start with the technical aspects of NVIDIA's technology. We know that everyone says it's better by a long shot.

3:33Everyone's trying to catch up. There's a question whether or not they can do it. Jay, I want to start with you. What is the bear case on NVIDIA's underlying technology? Sure, NVIDIA has great products, but others are catching up. I think it's as simple as that. If you look at AMD just had an event last week where they outlined their roadmap. They're getting closer by the day. They have now competitive silicon. They're going to have competitive racks. They're going to have RockM software was lagging by years, you know, a year ago. and now they've closed the gap considerably. By 2027, they're going to be fully competitive.

4:10On top of that, and probably more crucial, is NVIDIA is competing with all its customers. All the hyperscalers are designing their own silicon. If you look at Google TPU, I would argue that you could make a pretty solid case that TPU is a better chip for doing large classes of AI workloads. So, I promise it NVIDIA, but the competition is real and getting better. Ben, what do you think? Yeah, I think there's a lot going in NVIDIA's favor in terms of two points that Jensen's like to make that I actually think are relatively valid and line up with the way that we think cloud software and cloud AI workloads will develop.

4:49One is architectural compatibility. So when you're a software developer, you want to write software for what is the single architecture or architecture compatibility across either multi-clouds as well as desktops. And then you want install base, right? And NVIDIA has these single large installed base of GPUs, both at the edge and in the cloud. And that number is increasing by an exponential. So as much as, you know, I agree, TPUs are great products and I do want competition and I want AMD to do well. There's something very specific from a software development, a deployment and a continued innovation standpoint from all these developers for technical architectural compatibility and the installed base that continues to favor, you know, NVIDIA.

5:28and NVIDIA getting better and better and better at processing these insanely big models, right? If Gemini was a 5 trillion parameter model, we're going to 10 trillion parameters, we're going to hit 20 trillion parameters. That demands unprecedented levels of compute. And that's just Jensen and NVIDIA's sweet spot. What about, Ben, what about the chips that are focused on inference? Why aren't those concerning to you right now as a threat? I think that that will become more of a opportunity going forward. And then again, you got to ask yourself, like and i firmly believe this that nvidia will begin to fragment their product line from one general purpose thing that's both good at training and inference to offer also an inference specific product because again it fits their strategy for we've got cuda we've got software developers you're already going to run and train on us so why not inference on us and so you'll build out inference data centers also with a skew that i believe nvidia will make we're just not at that part of the cycle yet in fact very few people if any actually have any successful inference only product.

6:27This includes TPUs as well. Aimed is Helios will also be for training and some for inference. So I think we're just in that stage. But I fully expect everybody in the space, NVIDIA included, and the names that I mentioned to offer also inference products, because you're right, it makes sense to just have inference data centers with inference specific products. Jay, what do you think? I think inference is too soon to tell, right? Too soon to tell. Okay. I think the market for inference will be huge, but I don't think we have a good grasp of what those workloads will look like. And until we do that, it's going to be hard to tell what the silicon footprint will look like.

7:02And I think it's anybody's game. And I think that's a vulnerability for NVIDIA going forward. Okay. So, Ben, let's move to the valuation and the financials. Look, the financials are strong. We saw them last night on the call. The valuation is extraordinary. I mean, you know, like these multiples are nuts. Like, how are you thinking about it. Yeah. I mean, one of those things where we just don't have a ton of context for this, right? I mean, you know, you just look at how much money they're making, how much money they're slated to make, how much money Jensen is forecasting, right? He's saying essentially right now you can basically book, we're going to do better than 350 billion of revenue in 2026, which I think is, you know, crazy.

7:45But if you look at most people's forecasts, again, along two parallel vectors, which is one, we're seeing the largest buildout of AI data centers that we've ever seen in terms of how many gigawatts is being added globally over the next five years. And those will be filled, right? You're talking about over 150 gigawatts ex-China for the rest of the world. I'm not putting China in there because NVIDIA is not selling to China. So let's just go ex-China. That is a lot of gigawatts for data center only builds between now and 2030 that will be filled with GPUs, right? The market is there. The market is there, basically.

8:19The market is there. And I think people are underestimating the gigantic platform shift in architecture that happens when you move from Hopper to Grace Blackwell, to then Rubin, to then Feynman. These are leap step functions in GPU compute that the software world will have access to that they did not have before. So it's like a huge leap in performance in compute. And again, all of that's going to go into all of these data centers now that are being built for AI. So the world's going to look very different, honestly, computationally, and in terms of the opportunity for software developers, for employment, for enterprise to deploy these things at scale in a year or two than they are right now.

8:58Okay. Jay, the valuation's high. What about the financials? Are you seeing any red flags there? Rebut what Ben's talking about here. Sure. Right. So valuation is, you know, price versus earnings. And I think there is an argument to be made that the earnings are starting to show some signs of friction. Right. One of the things on the call that, or actually not on the call, they didn't mention it, but in their press release, they pointed out that they had booked$13,$14 billion of cloud compute services. Right. Huge amount of cloud compute services they prepaid for. And I don't know this for a fact, but I'm pretty sure that's what we call backstops, right?

9:38Whenever one of these NeoClouds buys chips systems from NVIDIA, it comes with a guarantee that NVIDIA will backstop some amount of demand from that NeoCloud, right? NVIDIA doesn't need anywhere near that, anywhere near$26 billion of cloud compute services. This is essentially a discount, right? And so the question is, well, why do you have to discount that? I mean, there's a bigger question too, is which the NeoClouds all say were fully utilized, 100 % utilized. Well, if that's true, then why do you need the backstop? And if these backstops are helping with sales, isn't that a discount? Is pricing maybe a little bit weaker than the sort of headline numbers would indicate?

10:17You know, it's not a smoking gun, But it is definitely, it's not as easy as it looks from the outside to sell these. Okay. So, vector number three that I want to get into is the macroeconomic situation and the, well, yeah, I mean, the macro situation at large. There are trade issues. We're in a current, we're in, you know, a certain type of interest rate environment which could change. Jay, what is the bear case around this? Do you have any confidence that things will get better on the trade front? i well no i the trade stuff is so complicated and it's so wrapped up in political issues rather than technical or commercial things that you know i i could talk about i i think the the dynamic with china is going to is going to be very challenging um and it's a big it's a big vulnerability for for nvidia um but also i i am you know there's this meme going around that the whole US economy is sort of depending on NVIDIA's earnings.

11:14That gives me some cause for concern, right? It worries me a little bit that it seems like a lot of companies out there are really dependent on NVIDIA and this whole AI build. And not just computing companies, but energy companies and electrical companies and piping companies, industrial companies. There's a lot riding on NVIDIA's prospects. And that's probably not a great long-term dynamic. Okay. Ben, this doesn't worry you. I mean, I agree with Jay on the trade stuff. I mean, government's government, man. Like, that's just going to be a hot mess across the board. I would come at this the other way, though, where I look at just this opportunity in terms of, again, scaling out and we're in a build-out phase for data centers, which is coming from the hyperscalers, right?

11:59So when Microsoft CEO and Google CEO and Amazon CEO tell you they have never seen an environment like this for demand for their cloud services, when again, they were already building out new infrastructure and new clouds just for basic cloud software because everybody, all enterprises, are moving their workloads to the cloud. In fact, most statistics tell you less than 45 % of organizations have or are fully deployed to the cloud today. So there's just a bunch of enterprise money that's going to go for cloud deployment. That's not including AI, where less than 10 % of IT budget today is for AI.

12:32So they are building capacity for the world to move to the cloud for software. And that was just in general software. We're not talking about AI. So they are seeing the demand. They're building the pipeline. And we have friends in all forms of the supply chain, from power to electrical, to equipment, to manholes, which is my favorite. We can't find enough manholes to build an anecdote that I share lately. And that's coming because hyperscalers are building and driving demand for all this. So it's really the hyperscalers to drive this. So then the question becomes, how tied are they to NVIDIA, right?

13:05And that's, I think, the central part of this. But whether or not that's NVIDIA or something else from an economic standpoint, the hyperscalers will keep building, and they're going to use the best solution, whatever that is, to fill their data centers to make money on tokens so that they can fund their clouds with enterprise software. Okay. I want to go back to that because like how much of Microsoft and how much of Oracle's at, you know, at capacity data centers are actually going towards open AI, right? It's a pretty fairly significant, it's a pretty fairly, it's a fairly big chunk of both Microsoft and Oracle's needs right now is fulfilling open AI.

13:42It's a really narrow base. It's like four or five companies are holding the whole thing up, right? And in the case of open AI, like full credit to them, they have incredible models, they have incredible technology, but let's not forget they are essentially a startup and they're burning a lot of cash and they've they have these big plans to you know to spend a trillion dollars in the next five years we don't know where they're going to get that money from they don't know where they're going to get that money from how much of nvidia's sort of long-term outlook has open ai baked into it it's a pretty big number um and i think that's you know i don't know how reliable that is ben yeah i mean open ai aside like i just have zero doubt on where the cloud guys are going, because again, they are right now spending less than 30%.

14:24You're basically putting it under the cloud. OpenAI side, you're saying the cloud, the hyperscalers are spending enough on CapEx. The hyperscalers are spending enough on CapEx. And again, that's what they're spending right now, less than 30 % of their cloud revenues. So they have opportunity to keep investing, keep monetizing, keep seeing the AI-centric portion of that road. And then again, remember, these are not one-trick ponies. And I think that one of the challenges for OpenAI and some of the frontier labs is that's the only revenue model, like Microsoft, Amazon, Google make money from a lot of other stuff as well.

14:52So they have the opportunity to continue to invest and fund. And a lot of ways, like I think what gets missed in this, oh, you know, how much money should they lose? Like, let's remember that for a decade plus, Amazon lost money investing in their future of capacity of books and warehouses and infrastructure before they turn to profit. And that's like, that's not dissimilar to what we're seeing now. there's just such a big belief that AI will monetize, that people will see revenue will be recognized that this is the build out phase. And that's the hope, right? And I, again, I'm okay with us taking a giant swing at this, if not from the essence of, I think America should lead and win in global competition on AI.

15:30And so I'm all for these companies taking these giant swings. That's a very American thing to do. And I think that this is like, again, but I'm optimistic that the returns we see from AI are already so positive early. And again, less than 10 % of IT budget, money will come. How much? Is this$5 trillion? I don't know, but money will come. But this gets us to point number four, which is a circularity here, which is, I think, Jay, what you're getting at here. Ben, the circularity, and I know we're talking about OpenAI here. We're not talking as much about the hyperscalers, although they are involved in this to some extent.

16:03The circularity of these funding deals, that doesn't concern you at all? I mean, again, where's the money going to come from? This is an absolute sort of critical point. But there's a handful of people in this industry that do this, right? It's hard to look at this and say, like, let's penalize industries, people who have four or five customers, because that's pretty much common throughout really a lot of the semiconductor industry and the semiconductor supply chain, right? You have your giants who invest in your partners or customers, just because there's a handful of people that make you a bunch of money, right?

16:36And I'm not really that worked up about, you know, Jensen's point about, look, we're going to continue to invest in the ecosystem. You have a company, again, that's going to generate over$500 billion of free cash flow. What are you going to do with that money? It makes sense to invest in your ecosystem. It makes sense to invest in accelerating CUDA and make that as wide. Like, that's a strategic thing that anybody in strategy, I don't think, would disagree with, right, when you want to deepen your ecosystem. So, So again, that would be a problem, in my opinion, if there was no demand. I think the demand is insatiable and everybody says it.

17:08And so it's hard to say, like, with the demand you see, keep investing because there's no signs that that demand is slowing down. Jay, the Amazon analogy that Ben made, do you think that that is a parallel here? Are there caveats that come to mind for you? So I think this is to some degree a question of timing. Like I'm willing to believe that long-term AI is going to be incredible and is going to deliver all kinds of important gains. I just think no technology as significant as AI is supposed to be gets adopted in a straight line, right? And between today and us getting to AGI, conscious robots, talking computers, you know, sending in computers, between here and then, there's going to be some pretty serious corrections.

17:57and we're going to see that across numbers. And I don't think we can say today what the compute footprint will look like when we get to our compute overlord future. So the last vector here that I wanted to talk about is the Jensen factor here. And the reason I wanted to talk about it, and I guess this one is, to be honest, it's less of a debate, it's more of a question that I have for the two of you, which is how much of NVIDIA's valuation do you think is attributed to Jensen alone? And I know, you know, it's a question that has been asked, for example, with Tesla, with Elon, right? And you could raise the question, if Jensen stepped down tomorrow, what would happen to the company?

18:36Would it maintain its competitive edge? Jay, what do you think of that? So full credit to Jensen. I think he's one of the best CEOs in semiconductors in tech, probably in the country. But he is, one of his superpowers is his immense charisma and his speaking ability. He loves to talk about his big visions for how the future is going to play out. And I admire that. It's compelling to listen to, but it can also get us into trouble, right? Jensen will be talking about the future and leading us off a cliff someday, right? And I think he likes to talk. He's pretty good about it, but sometimes it gets him in trouble.

19:17And we see that with China and the U.S. now. He is having to walk this very fine line between China and the U.S. and he said some things that have gotten both sides upset. And so the last few weeks, he's been pretty silent on the topic. And I think, in general, he's very persuasive, but he's not going to tell you that there's an imminent stop coming. We're going to just follow him until it all falls apart. I realize that we've kind of teed this up now, Ben, for you to end us off telling us about all the ways that Jensen is awesome, which I trust is not that hard for you to do. Yeah, no, I mean, obviously, I think Justin's charisma, his technical understanding, just his passion for this industry and the way that I think he's predicted relatively good shifts in the industry is true.

20:02But in terms of like, again, how much is this him and tied into his future? I would just offer this analogy, which is Apple, right? The same conversation came when the tragedy of Steve Jobs' death happened, that it was really who is going to, can anybody be Steve Jobs? Is Apple completely screwed? And there's just something to when you have really good products that delight customers and that add value and create value that is sort of a fundamental point about why I think these products can continue. And, you know, that basis of the product, like the product merit, what the product's going to do speak for itself, which is really what happened with Apple, right?

20:36It was a great culture. It was a great product. Tim Cook is a great CEO who continued to scale and operate that company, but it was because of the culture, the vision, the infrastructure, as well as the product vision. Although I will say, this is a question people have now, is Apple's innovation has kind of stagnated as of late. And I know we're talking about different points in the company lifecycle. I mean, this is now many, many years after Tim Cook has taken the baton. But look, I mean, you look at Apple right now, you could say that it's kind of stagnated. Yeah, and I think history knows that hardware innovation happens in cycles, right?

21:17There just becomes the limits of physical technology. What can you do, right? And I would say, let's wait and see what the next 10 years looks for Apple at a hardware innovation point. Right. All right. Well, I want to appreciate both of you for, I appreciate both of you coming on. And it's a great discussion. And look, we're going to try to do these for all of the Mag7 companies. And NVIDIA, it was a logical starting point. I'm excited to get both of your thoughts along the way on maybe some of those other Mag7 companies. And until then, thank you both for coming on. We appreciate it. Yeah, thank you.

21:51Thank you. See you soon. Okay. The White House has been thinking hard about AI lately. And a new story my colleagues Sylvia Varnam-Oregan and Aaron Holmes published yesterday, reveals new details about an executive order the administration is working on. And I want to bring them on both to tell us more about what they've found. Sylvia and Aaron, welcome back to the show. It's great to have you here. Hey, gosh, great to be here. All right, so let's talk about the White House. Sylvia, what did you find in your reporting? Yeah, so yesterday we obtained a copy of a draft executive order that, as you said, the White House is working on.

22:27And it's effectively trying to prevent states from regulating AI. Now, this order, which is still in draft phase, directs the attorney general to create a litigation task force to sue states who enact AI laws that the administration is unhappy about. It also instructs the Commerce Department to withhold federal funding to certain states as well over this issue of AI regulation. So it's a really interesting document, obviously raises a lot of questions as well. Okay. And Aaron, if you think about what a regular or an executive order like this, what impact it would have on Silicon Valley and on companies, what stands out to you?

23:14Yeah, I mean, this is essentially something that a lot of tech companies have been asking for from the White House for a long time. You know, we saw Mark Andreessen essentially come out this week in favor of sufficient action to block states from regulating AI. And we've also heard that Meta, Google and OpenAI have been asking for, you know, similar action. And the basic reason for that is that, you know, these state regulations make it more onerous and costly to comply with as they build AI models. And those companies don't want to have to deal with a patchwork of regulations across different states.

23:49So they're essentially, you know, using their lobbying with the White House to ask for some sort of action to preempt those state laws. Sorry, I was going to ask you a question and then I started choking. That's a first time on the show. But as you can see, it's just a very exciting topic. Sylvia, my question for you, if my voice will allow me to ask it, is has something like this been tried before by the White House or by any federal regulators? Yeah, so this is a really interesting area because it has been tried before. People inside the White House were, in addition to people in the technology industry, lobbyists were trying to get a provision included in Trump's big, beautiful bill back in July.

24:42They wanted a state AI moratorium added to that bill, but it was very unpopular and it was actually killed by the Senate in a vote of 99 to 1. The White House is also trying to see if they can revive that provision and include something similar in another piece of legislation. So it seems like they very much want to find a way to do this. But I don't know if it's got the popularity needed amongst Republicans as well as Democrats, of course. Right. And talk about these state regulations that the White House is trying to block. How many of these state regulations are there even? And specifically, I guess, if you could talk about a few states that have implemented them, you know, I'm just wondering how big of an issue this is at all.

25:35Yeah, well, there are more than a thousand bills that have been introduced at the state level, and several of them have passed. The executive order or the draft executive order that we saw takes particular aim at Colorado and California, where acts have recently been passed, and these address issues like the potential for algorithmic discrimination and transparency issues. So there is certainly a lot of activity at the state level. The people behind the executive order and Trump himself have advocated for a national standard as opposed to this patchwork that Aaron mentioned. Although it's not really clear to me at least when such a national standard would materialize, what that would look like.

26:19There is some instruction in the order for David Sachs and others to begin that process. But it's not clear when or if a national standard would come to pass. Aaron, as you think about watching this issue moving forward, what kind of questions come to mind in terms of how you think tech companies will be affected by this? But also, you know, is this going to encourage more lobbying from the tech companies to sort of push what they want, given that it seems like it's kind of working? Yeah, I mean, I think it's going to be really interesting to see what sort of legal arguments the Justice Department tries to make to challenge these state laws.