YouSaid · the spoken record

Lip-Bu Tan

lines on the record: 110
first: 2026-07-15
most recent: 2026-07-15
sittings or episodes: 1
sources: podcast

Every line below is reproduced as it was said and linked to the record it came from. Nothing here is summarised or generated. Directory · Search · Corrections

“Not make these snap decisions, which often are the reason why he's amazing, but like some of these snap decisions are hurting him. I'm not sure if I can give Elon that much great advice because I think maybe it's just like get off Twitter and focus on like the products again, right? More, but he is working on that stuff a lot. Yeah”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“A lot of people at XAI are mad about the porn models, like porn stuff, it's fine. Like, you're gonna make a ton of money off of this. This is how you accelerate the revenue of that company. But he's losing a lot of talent and axing a lot of good projects. But Elon is a magnet to amazing talent and building stuff. So I won't bet against him, but it seems like since he left the administration and focused on stuff again. But I think, I don't know, I think he's focused on a lot of things. And I think like Robotaxi starting to look good actually again. I haven't written one yet, but I have some friends who ridden one. It's like looks pretty decent. He could like.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“It's unusable. It's like, what is going on? Like, you know, what is going on? Like, you need to shake the crap out of the company. I think they win a lot because they have the best business-to-business relationship with so many enterprises. Yeah, but like they end up not having actual product to sell them, which is like really scary. So they need to really work on product. Satya has done great on sales and stuff, but like, yeah.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Everything going for them. And there's just nothing, right? It's like GitHub co pilot is failing. Microsoft co-pilot is still crap, right?”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Model. They only had the best IDE, the best source code repository, the best enterprise sales force, the best model company as a relationship, and they were the first to market, right? It's like everything going for them.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Microsoft has the same problem. I think they were super aggressive in 23 and 24. And then they pulled back heavily, right? Now OpenAI is slipping through their grasps. There's that whole thing there. They cut back on data center investments heavily. They were going to be the largest infrastructure company in the world by like a factor of 2x, which would have been, you know, you could argue maybe that was like too much and maybe it wouldn't have been economical. But like they're losing grasp on open AI. Their internal model efforts are failing spectacularly. Like they're on LLM Arena right now and they're pretty decent there. But it's like that's just like a systemic model. It's a code name, but whatever. failing Azure is like losing a lot of share to Oracle and Coreweave and Google and so on and so forth, right? Their internal chip effort is by far the worst of any hyperscaler.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Think other people are going to be able to integrate user data and agents will be able to integrate all this user data and they'll start to lose control of what the user experience is as more and more gets disintermediated by AI being the interface rather than touch rather than touch pad and keyboard. And I don't think they've truly realized what happens when the interface to computing is AI. Like they market it, but that's going to shift computing really heavily. They have great hardware. And their hardware teams are working on awesome stuff and form factors. But like, I just don't know if they get what is actually going to happen to the world in the next five years truly well enough. And they're not building fast enough for it.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“I think more and more you'll see people Know gray Apple has this walled garden, but like they can only do so much to protect it, right? IDFA. Yes, they have access to the text and they have access to this.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“The funny thing is, like, some of their best AI people are now at super intelligence. They're building an AI accelerator. They have AI models, but they're just way slower. They did mention on the last earnings call they're going to allocate more capital to this, but it's like, guys, Apple, like you guys are going to lose the boat if you do not spend like $50, $100 billion on infrastructure”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Know if you read his website post about AI, I think he sees the vision, right? There's the wearables, there's integrating AI into that, there's being your AI assistant to do all this purchasing and stuff. I think he sees the vision, but I think he also needs to focus on actually releasing that faster, but also the products that they do outside of their core IP every time they launch something is kind of mid, right? Metal Reality Labs is doing well, but I think they should like go more explicit like have a chat GPT competitor, have a claud code competitor, like just start releasing way more products because they're really just focused on their individual gardens rather than branching outside of it.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah. And also, like, learn how to ship product. I think Zuck remains to be seen what goes on with superintelligence, but like they're trying to move super fast with the data centers. Screw it, we'll build tents instead of physical data centers because we only need these for five years anyways. The superintelligence moves, you could say whatever you want, but like, you know, trying to buy Thinky for 30 billion or SSI for $30 billion didn't work out. So then they spent not even that much on hiring, not $30 billion on hiring all these people. So I think that he recognizes the urgency with the models, with the infrastructure. So I really think he needs to like.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Queries, the monetizable ones, especially to purchasing agents, is going to really screw Google long term if they don't get their act together. I think they've gotten their act together on DeepMind. There's still some inefficiencies, but Sergei works within DeepMind a lot in their driving hard. They're still a little bit behind, but Think physical infrastructure TPU and how much money they could make and how much they could take the wind out of everyone else's sales if they start selling TPUs externally and reorg around building data centers much faster so that they do have the most compute in the world because they did but now there's certain companies that are going to surpass them potentially over the next few years if they don't really get their act together. So I think that's what I would say for them.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“I think they should open up the Kimono on TPUs, right? Like start selling them, open up the software, open source a lot more of the XLA software because there's OpenXLA and there's XLA, but the vast majority is closed source. Really, really open up the Komodo on that and be a lot more aggressive, right? They're still pretty not aggressive on data centers. They're pretty not aggressive on a lot of elements of the company. The TPU team's next generations are pretty not aggressive, partially because a lot of the TPU team has left to go to OpenAI, the best people that I knew. It was actually really annoying. I knew like four people or five people, and they all went to OpenAI. And it's like, fuck, like now I don't get as much. I met some other people, right? But it's like, you know, I think they could be a lot more aggressive in many ways across the company. They don't have to be, right? But they could because AI, you know, like this chat GPT take rate, the shift of”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“But the cash on their balance sheet keeps growing, and they're going to have north of $100 billion of cash on their balance sheet by the end of this year, I think. So it's like, what are you going to do with that? I think there's something moving into the infrastructure layer much more that they could do if he really wants to be the king of the world, right? Which I think he does.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“But they should accelerate the data center ecosystem with investments, right? Because really we think we can have very high degree of accuracy on what they're going to do next year in terms of revenue because it's just the number of data center watts that are being built, right? Like this is harder thing to shift up and down, right? Now there's a little bit of share difference between how much is TPU versus GPU, but like it's like you have to accelerate the infrastructure and you need to spend all of this capital that you're building, right? Like, okay, do you want to go the route of like doing buybacks and dividends? Like great. Like you're a loser if you do that, right? Like you can make more money by reinvesting in building a bigger company that's not just chips into the ecosystem or servers into the ecosystem, but actually like controlling the infrastructure end to end somehow. So I think there's something he could do there with this massive war chest. And there's a reason like NVIDIA's done some buybacks and they've done some dividends and increasing.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“If Jensen was here. I think he has a massive, massive balance sheet, right? Jensen does. He's cash free cash flow is like, you know, ridiculous. The tax cut, the new Trump tax bill institutes something really incredible, which is that you can depreciate all of the GPU customer cost in year one, which we put out like a note about how the tax implications to Meta are like $10 billion a year. And across each of the major hyperscalers, it's like massive. It's like, well, NVIDIA is going to spend tons and tons of cash or they're going to spend like tens of billions of dollars of taxes. Why don't you get into the infrastructure game somehow? Now this is obviously going to be like crazy because now they're buying their own GPUs and putting them in data centers and doing stuff and they're competing with their own customers, but they're already doing that anyways because their customers are trying to make chips.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“But I think, I pray someone does something, right? Like you get a big capital infusion. I don't know. The big hyperscalers are like muscled into like, oh, okay, wait, if TSMC eventually grows their margin to 75% because they're the monopoly, plus they intake all this stuff like co-package optics and power delivery and all this, like all of a sudden the cost is going to spike. So we should actually just throw $5 billion at Intel each, right? Screw it. And that could actually give Intel enough of a lifeline to potentially get to something and maybe be competitive. That's the hope.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“I think you should even run the company separately, right? But you can't physically separate them entity wise. Or they lay off like half the company, right? Which some could argue you need to lay off like 30% of the company, anyways. But there's a lot of bad things that happen if that happens, right? And they need to spend a lot more on building the next generation fad, even if they fix the fab. And they don't have money for that, right? So there's a lot more important problems than like physically separating the company, even though I think long term, yes, the fab has to be separate from the chips design software, right? Or chip design part of the company. Just like that's going to make each company much more accountable, be able to service their customers better, et cetera. It's just, that's going to take too long and they're going to go bankrupt by then.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“I think the goals are completely correct. I mean, I think the big challenge, just reflecting back in my time there, right? I think the big challenge is that right now, if you look at Intel, right, they have essentially software, the chip design, and then there's the core manufacturing part, right? And they have three very different cultures. And it's very hard to get everything under one umbrella, right? And so I think that is the big challenge. I think you should even run the compan”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Anything can happen. So they have to figure out how to get capital. So he has to figure out how to get capital. He has to figure out how to clean up all the crap, make the... Know yields go up, right? Make the product ship way faster. Like all of these things are basic products.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“In the company that worked on the FAB side and were like, who's the worst people and friends? It's like, oh, this guy sucks. I explicitly told Lip Bhutan he had never talked to the guy because it was like four layers down. The company has like absurd amounts of hierarchy. It's like four layers down. He goes and talks to the guy and he's out, right? It's like he figures out like who's bad, right? And who's good? And he has to go in and he has to be like, hey, the vast majority of the team at Intel is the one who led the world in production and processed technology for 20 years. But there's a lot of like built up crap. So he has to go figure this out, right? He can't waste his time on, oh, all this structuring to split. I think it would be better if the company split. I just don't think he can spend the time to do that. And if the design side of the company is, you know, you're not really going to get into AI. You're not really going to, you have to make some money there. But the fabs, I think, could truly become a competitor. But they're going to go bankrupt by the time.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“IBM still makes more money every launch off of mainframes. So it's not like x86 is dead. It's like you don't get the growth rates, but like you could totally run this as a very profitable enterprise. And I think the same with PCs, right? There's some turmoil. There's some arm entry. There's some AMD competition. But I think it's a very, it can be a very profitable business if it had one third the people or half the people working on it. And so like Liputan to fix Intel needs to go into both the design company and lay off a shitload of people, but like keep all the good people and make sure that they're designing fast and they're launching from design conception to launch is two to three years, not five to six. And that's on the design side and make that profitable. And they're on the fabiest to do the same thing. There's all these people. One of the heads of fab automation at Intel, I explicitly told Lip Bhutan because we have a couple X intel people who are actually good.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“So, but if you look at Intel today, right, they still don't have a competitive entry on the AI side. And they will, right? So, what does it mean for their offering? I mean, they're still doing great on CPUs. They don't have a good AI chip product. Is it long-term sustainable positioning, right? I mean, as a standalone chip company?”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Then he never actually fixes the company, right? Intel's problem is that it takes them five to six years to go from design to shipping the product. In some cases more. And when they tape out a chip, right? Like, you know, you send the design to the fab, the fab brings back the chip, they go through 14 revisions in some cases, where it was like the rest of the industry goes through like one to three, right, revisions if they're good of like send the design in, uh, get the chip back, test it, send the design in, right? For a public launch. And they'll launch a chip in three years.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“I think the process of splitting it would take so much executive time and so much executive effort that you would have been bankrupt by then, right? And that's the big challenge. Like I think Intel should be separate, right? But to properly split the company and for all the management time that's needed, it was like absurd. And instead, like what you need is like, you need Lip Bhutan, who's the CEO of Intel. There's a lot of drama going around about him because he's one of the greatest semiconductor investors ever, right? He's invested in so many different companies first. You know, he was on the board of like SMIC, which is China's TSMC effectively, which is like a big drama or like some of the biggest tool companies in Chinese, the first investor in them. Because, you know, it was a multipolar world there and he was making good investments. But now people are getting mad about that. But it's like, no, he recognized that the companies he understands the supply chain. He needs to not spend his time on splitting the company because”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“And so, like, there is this difficult, difficult Thing to be done that, like, hey, there's one island that controls all leading edge semiconductors and not just all leading edge, like the majority of trailing edge production as well. Something needs to be done. Intel is behind, but not like absurdly so, right? Like if something were to happen to Taiwan, Intel would have the most advanced technology in the world, right? It's just, it's not economic.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“I think the world, well, the US needs Intel. I think the world needs Intel because, like, Samsung is doing worse than Intel on leading edge process development, in my opinion, based on even on various customers in the industry having done test chips at like Intel versus Samsung. I think industry generally agrees that Intel is further along, you know, the sort of the two nanometer class process technology than Samsung is, but both are way behind TSMC. And TSMC is a monopoly in some extent. The number one question always people ask is like, why is TSMC not making more money? Why are they only raising prices next year 3% to 10%, depending on what it is? It's like TSM sees a monopoly. Like they could raise a lot more, but they're good Taiwanese people rather than dirty American capitalists. If TSMC was owned or was managed by Americans, I think most ownership is actually American in terms of the stock markets on the New York Stock Exchange and all this. They would have raised prices a lot more.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Because at the end of the day, the expensive thing, right? Like this is why what Elon did would seem silly, right? They spent a lot more money on generators outside the data center and these mobile chillers to cool the water down for their liquid cooling instead of like the more cost effective option because it got the data center up three months faster. And so that three months of additional training time is worth way, way, way more on a TCO basis, right? The performance you got out of the chips and the time to market and all this is way, way faster. And therefore, it was the right decision, even though this part of the data center bloomed at cost. Everything else is still there and you're still paying for the chips. And if they were sitting idle, it's not worth it.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“80% of the cost of a GPU data center, if you're building Blackwell, is capital. It's the GPU purchases, it's the networking, it's the physical data center conversion, power conversion equipment. All of this stuff is like 80% of the cost. And then 20% is going to be your land and your power and your cooling and your cooling towers and your backup power and your generators and all this stuff. It's like nothing, which is why it doesn't matter if you spend 10% or 50% more on that.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“But outside, like, it's like in Europe, that number is not moving up that fast. And all these other countries, I think. We need to build a lot more power, but it's not like some crazy, crazy amount. It's just doing it properly is the hard thing. And again, the cost of power, you go look at these deals.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“If you want to service it, you're screwed, right? So, like, the same with power, it's like we talk a lot about the power is not actually that expensive. It's just hard to build, right?”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Like cooling is like the physical cooling of a data center are like, you know, there's this whole narrative about, oh, yeah, I use this so much power. And it's like not really, you know, farming alfalfa uses like 100x the water of AI data centers even by the end of the decade it'll be the same and it's like alfalfa is like worth very little so it's like i know there's like it's like cooling is like not that you know people have like experimented with like you know undersea data centers to reduce the cooling cost”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“So, what's the endgame for data centers? I mean, we need more power, we need more cooling. Well, all data centers will be next to a nuclear reactor, lots of solar, you know, next to a deep level, like deep seawater that we use for cooling or something like that.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“They're getting many, many Blackwell from them, right? And the same with Oracle and the same with Microsoft and all these other companies are renting tons of chips to China anyways because it's more cost effective to do that than build it yourself. So it's not like China has this mentality where we only have to, well, the government does, but the infrastructure companies don't, like Alibaba 10 cent, Bike Dance, et cetera.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“In China, they don't have any of these problems, but they just haven't spent the capital yet. But capital is an issue as well because of the scale of what's being spent, right? Like Nvidia's revenue this year is going to be like over $200 million. And next year expects over $300 billion. Plus Google is going to spend like $50 billion on TPU data centers, right? And it's like, and Amazon's going to spend tons and tons on trainium data centers. It's like the scale of dollars is quickly growing to nation-state level stuff. And what's more important is being able to decide to spend the dollars and what's cost effective. And so to some extent, China's still constrained by that, but they can smuggle chips in. They can build data centers outside of China. They can rent data centers outside of China and have the most cost effective Blackwell chips or whatever, right? Byte dance is, you know, either the biggest or the second biggest customer of Google Cloud for a reason, right? And they're getting.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“In the US. Power grid interconnections, transmission, substations, all of this stuff. Like electrical contractors, electricians in Texas, if you're willing to be a travel electrician, it's like oil pay, right? Like it used to be that like if you're physically adept, you could go make 100 grand in West Texas. But like, who the fuck wants to do that? Now it's like, well, you could go like 200 miles away from Dallas and what's still a reasonable town and build a data center and work on the wiring within the data center and all this other stuff, the transmission stuff. And your pay is up like 2x now versus what it was just a few years ago. This labor problem is a challenge too. And it's, yeah, I think.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Even if you didn't do it in your own self-built data centers, there's still a lot of challenges in the open market. There's a deficit, right? And that's constraining American chip bailouts heavily. Yes, others could maybe do it a little bit faster, like Corieave or others, right? Oracle's got an open mind as well. But it's still constraining U.S. build-outs heavily, even though the capital has been spent, right? The chips are 60 to 80% of the cost of the cluster, depending on what chips you're getting. So it's like they've already bought the chips. They just can't put them anywhere because the data centers aren't ready. Supplies to Google, applies to Microsoft, applies to Meta, applies to a lot of folks.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Because they need the data they want the power. They need the power, right? And it's like all the hyperscalers have said screw off to my sustainability pledges because they need power as fast as possible. They're doing things that are not, that they take a little bit longer to move the ship, but like”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Going through his mind, yeah. So I think what's really important is that Coreweave doesn't care, right? They're like, oh, crypto data center, I will convert it to AI data center, right? They bought a company for like $10 billion that's doing crypto mining, which is worth like $2 billion a couple years ago. And it's not because they're Bitcoin mining business is growing. It's because they have powered data centers, right? Like anywhere and everywhere, people are trying to build power data centers and companies like Coreweave and Oracle are moving to the actually today Google just didn't bought 8% of a crypto mining company called Terrellwolf, right?”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“No, don't go anywhere. Jensen likes them as well, to be, yeah, yeah, that's very important as well, but they'll go because it unconcentrates the ecosystem, which is better for NVIDIA.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, yeah, for sure. Because companies like Coreweave, you know, why is Coreweave valuable is really because they build infrastructure really fast, right? And their software is nice, I think, but a lot of their customers are bare muddle, right? Just replace the GPUs whenever they're broken and network probabilities.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Isn't this to some degree also coupled to their unwillingness to sell them to a broader ecosystem? I mean, if they want to be confined in their own data centers and they're didn't run data center build out for their own hypersecurity cases quickly enough, right? Then yes, that constrains them. If they were on the open market, would we still constrain?”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“You still have the problem of like, well, dollars spend to AI output in tokens or in whatever is going to be lower because these chips are worse. So power is not the gating factor. It's always capital, right? At least today, right? Now China can spend a lot more capital if they wanted to. They're subsidizing the semiconductor industry to the tune of like $150, $200 billion a year through SOEs, through CapEx that's not generating revenue, et cetera. So it's not like they couldn't do this to the AI ecosystem, right? Given Meta's CapEx is like $60 billion, right? And Google's CapEx is like $80 billion, right? Like they could totally spend way more than that on a single effort. They just haven't decided to. And I just think for the US, our buildouts are constrained by power, right? Like Google has a ton of TPUs sitting waiting for data centers to be powered and ready, as does Metal with GPUs, right? We posted about how Meta”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“I don't think so. I think, again, like there's a lot of like. What we see is that, like, even with H20 being sold to China and future versions of the chip, the H20E and other chips, we still see Chinese companies like Alibaba renting GPUs outside of China because the GPUs they can get outside of China are just so much better on a dollar spend performance basis. Renting them or even like going through sort of like a Singaporean company that is effectively a Chinese company and building data centers and putting chips in them. So it's like, I don't think China's limiting the power per se. It's that it's, you know, you can only, if you can spend like Chinese companies are growing their CapEx way more than US companies on a percentage basis next year. The absolute dollar number is obviously the US companies are spending more still on AI. The percentage basis Chinese companies are growing.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Versus selling on the chips, right? The economic value derived from selling them the chips is not as large as being able to somehow sell them AI services.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah. And this is how NVIDIA argued to the administration. They're like, if we don't do this, actually, I think it's like a very powerful argument that, for example, within Triton, which is a common ML library, like Bite Dance has open sourced some stuff that plugs into this that is super awesome. And there's like all these other libraries. It's not just models that China open sources. It's like software for NVIDIA that Chinese companies open source. In a sense, like by NVIDIA selling GPUs is NVIDIA's argument again, like was like they were able to stop Huawei from building up a software ecosystem and the Western ecosystem is better. But then on the flip side, it's like, again, if you believe the models deliver more economic value to society than the hardware, which I actually think they do. It's just there's a value capture problem today, then you're giving China way more by giving them H20s. And soon a version of Blackwell that's cut down, like Trump said, right?”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“So, like a back door wouldn't even matter. I don't know. I think. It'll be interesting to see because China can definitely deploy way, way, way more power to AI the moment they decide to. But there's these competing interests, right?”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Which is a big challenge in America, right? There have been companies that were like, you know, Jensen keeps saying he couldn't give away H-20 in America for free. But I've literally heard companies say, like now say like, yeah, no, I mean, I wouldn't because I only have this much power. How am I going in data centers ready to go over the next year if I bought an H20? I'd literally have less compute capacity and then I'd lose, right? Even if it was free, like it doesn't make sense. Whereas China doesn't care. They can build these things. They have the muscle. I'm curious how this all shakes out. You know, China's posturing really hard. They even like put out something that I was like, we're investigating to see if there's backdoors in the H20. It's like there's no backdoor in the H20. Like, chill. You know, it's like, you know, GPUs usually like firewalled from the public internet anyways. You step through stuff before you get to the GPU cluster.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Well, what's interesting is that efficiency is just not so much less of an issue in China than here because they just have the power infrastructure to be able to support. So even if they're running less powerful chips, you would imagine that it doesn't really matter because China has just such an infinite supply, infinite supply of power that, you know, they'd sort of be okay with it. So it's interesting.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source
“Yeah, defense supply chain. And then they get that, right? Like, so it's like, and Letnik himself said we had to do this for rare earth minerals. It's like interesting. China, there's like provinces in China that have rules that say the H20 is not efficient enough to be deployed, which is like super bizarre because it's clearly the best AI chip China has. Huawei is still a little bit behind.”
2026-07-15 · a16z Podcast · From the Archive: Can Anyone Catch NVIDIA? | The Future of Chips and Infrastructure · IDENTIFIED FROM THE TRANSCRIPT · source