Chamath Palihapitiya
Co-host, All-In Podcast · Founder of Social Capital · CEO of 8090 Industries
“$725 billion in CapEx guidance in 2026 from but four companies. Amazon, Microsoft, Google and Meta. Amazon leading the pack with 200 billion, 190 billion each for Microsoft and Google, 145 billion for Meta... we are going to see a trillion dollars in build out over the next year”
“We had a massive build out of the infrastructure of the Internet in the late 1990s and into 2000. And what that caused was a lot of aggressive companies to do massive amounts of spending... look at the 2000 peak of Cisco. This is the most extraordinary chart ever. It took them 25 years to get back to that peak”
“Love Mark said it right. All these companies should go public and get evaluation and focus on the higher order bit. And all of these tiki tacky mechanisms that people can use to stay private longer need to get a bullet put in its head. And these SPVs are the worst. The layered SPVs. On SPVs. On SPVs.”
“the Chinese and the Americans have a very strong incentive to kind of divide up the world. Like, I think the future at this point, you have these critical resource inputs. China has a stranglehold. You have energy and intelligence AI, where America has a stranglehold on one and an effective stranglehold on the other, which is energy. And I think there's a trade there to be done. And as long as you can negotiate what a reasonable give and take is, I think they're just going to find common ground. Like, does China really need to be in Chile and Venezuela and Panama? Probably not. Does America have to have an incredibly strong point of view about the Strait of Malacca? Probably not. And so there's probably a trade to be done so that we can get their rare earths, they can get some more oil, everybody can be happy and we can all find abundance.”
“18 months from Taiwan not being an important moment of conversation the way it is today. Why 18 months? Because we are at a point where we're probably one to two nanometers away from being able to do what we need Taiwan to strategically do for us. And so as we scale up our chip fabs, as we get more capacity and interestingly, There are these orthogonal technologies being developed... Today it's economic. And if you take that off the table, I think we'll have a very different attitude to Taiwan. That's number one. Number two, sell the chips. And the reason we should sell the chips is we want Nvidia to win. We do not want to give enough oxygen for Huawei to then all of a sudden emerge and have a version of a chip that works.”
“the midterms, I think are probably going to swing more based on these recent gerrymandering rulings from the Supreme Court and what happened in the Virginia Supreme Court and what's going to happen in the state legislatures of these red states and some of these blue states, more than what's going to come out of this summit with President Xi. So let me just put that over here. And I think that money is getting organized. There was an article in the New York Times this week which really surprised me. But the largest donor in this election cycle is Andreessen Horowitz.”
“All these companies should go public and get evaluation and focus on the higher order bit. And all of these tiki tacky mechanisms that people can use to stay private longer need to get a bullet put in its head. And these SPVs are the worst. The layered SPVs. On SPVs. On SPVs. That's pretty odd by the way. And I will guarantee you this, once SpaceX goes public, once anthropic goes public, once OpenAI goes public, you're going to see a litany of these lawsuits back and forth between the purveyors of these SPVs. They should not be allowed.”
“If you put those two things [recursive self-learning + scaling] together, I think that you start to potentially live out this idea that there's an order of magnitude improvement on a yearly basis. So like this new form of Moore's Law. So then the model quality just goes absolutely parabolically just like this, straight up.”
“[Anthropic has] a decent lead on everybody else, whether it's three months or six months. Obviously they're probably six 12 months ahead of open source. Maybe they're three, six, nine months ahead of their contemporaries, but they have a lead.”
“across all evals, there is no single best model anymore. At the top of the leaderboard opus 4, 7, GPT, 5, 5, sonnet 4, 6 appear almost indistinguishable, separated by less than 3/10 of a percentage point overall...these things are getting commoditized way too quickly. And then you'd say, well, what's the ROI on all this incremental spend?”
“There was an economic and capital moat to training that is going away. It's going away in two ways. One is because we're getting these domain specific architectures at the silicon layer. And then second, we're rebuilding all of the core components...Elon was like, we've rewritten the entire training complex in C and it's an order of magnitude increase...Those kinds of innovations are going to make the cost of model training so much cheaper that it's like, why would we stick to the $10 billion training runs when we can have the $10 million training runs?”
“a lot of the folks that we see now in the Fortune 1000 and increasingly the Global 1000, they want abstraction above it. They want to sit in a control plane. They want to have the flexibility because they don't know how it's going to shake out. They see all the money being invested at the model layers, but they see the model quality asymptote. So they're like, wait a minute, what are we supposed to do?”
“Getting public sooner, having the scrutiny of public markets, having the scrutiny of having to deliver sharpens the focus. Steel sharpens steel. Iron sharpens iron. And I think innovation tends to get better.”
“If you want to be unexploitable, the best thing you could do if you're trying to build a super God is have three or four entities in a room, close the door behind you, and then dominate those other three or four entities. And then you set the rules. If the refs don't understand the game, you'll run over the game.”
“There is no single best model anymore. At the top of the leaderboard Opus 4.7, GPT 5.5, Sonnet 4.6 appear almost indistinguishable, separated by less than 0.3 of a percentage point... which theoretically says that these things are getting commoditized way too quickly.”
“It hasn't done anything measurable yet. At the end consumption of these tokens, nobody is standing there and saying, look at my filing. Here is the lift that I have gotten. Nobody has said that yet... instead people are realizing I have this cover now to go and clean up poor management over the last five and ten years.”
“They want abstraction above it. They want to sit in a control plane... they want the flexibility because they don't know how it's going to shake out. They see all the money being invested at the model layers, but they see the model quality asymptote.”
“A gigawatt now costs $100 billion, guys. When I started this project it was 4 or 5 billion and it's increased by 20x. If you want to get all 3 gigawatts developed, I have to come up with $300 billion.”
“The marginal cost of production for a new internet user was effectively zero... AI is completely different. There is a real cost for every marginal user — everyone you stand up is taxing a GPU, needs electrons, needs memory. There needs to be all this critical infrastructure that even enables you to be on the field.”
“Companies need to start underwriting this next phase of AI: how do I have control? Do I want single-point-of-failure risk with respect to AI? The answer is you need broad diversity. A downstream scientist using the cloud APIs could trip it, and you'll get cut off from a very important source of business differentiation.”
“China has been smoothing energy consumption globally... we've kept a damper on $200 oil; we're sub-100. But if China runs out of reserves and has to go back into the spot market to buy an extra 3 million barrels a day, there's a very big risk oil gets well past 100, maybe between 150 and 200.”
“when you use our harness with Claude, it was simultaneously 1.4x cheaper and 1.5x faster than just using anthropic Opus4.8 alone. But if you wrap the open source model with our software Factory, it was 16.4x cheaper. Now it was 3 times slower.”
“You can't rent intelligence from the same place that rents it to your competitor. ... It becomes the lowest common denominator problem where you and your competitors now look exactly the same.”
“Right now, OpenAI Equity, I think is more reasonably priced than Anthropic equity. ... the reason is that OpenAI can fall back on a really healthy consumer business.”
“He said, write down our token costs are doubling every 45 days. Honestly, what we're finding out is that you need to use a lot more tokens to get to this next iteration of improvement because we've effectively already asymptoted.”
“The answer as far as all publicly available data was that the actual ROI was somewhere between 0 and 2%. At some point you're going to have to show an ROI that's above the risk free rate of return. Otherwise you're going to have some angry investors on your hands.”
“the real business model is not in the foundational model anymore. It's at the application layer above, and it's in the infrastructure below, whether that's the cloud or whether that's chips.”
“if they said no more open source, American companies cannot use open source... those things cost 50 to 100 times more than your other best alternative... So then Coca Cola has to get re rated. But then you look at the people who are selling those tokens... If the government comes in and actually tells you that there's no open source, their valuation will crater. Why? Because all of that revenue is artificially being propped up.”
Chamath Palihapitiya
One-line summary: All-In Podcast 'bestie' tracked here for capex-cycle framing and dot-com-2.0-style cautionary commentary.
What they're known for
Brief factual context — fill in.
Why they matter to stock-market
Why this person's claims are tracked here — fill in.
Said
Speaker-attributed claims extracted from diarized sources. Each bullet mirrors one entry in quotes: frontmatter — keep them in sync.
-
On ai-capex-to-power-and-materials-cascade:
"$725 billion in CapEx guidance in 2026 from but four companies. Amazon, Microsoft, Google and Meta. Amazon leading the pack with 200 billion, 190 billion each for Microsoft and Google, 145 billion for Meta... we are going to see a trillion dollars in build out over the next year" — 2026-05-01-all-in-podcast-openai-misses-targets-codex-vs-claude-elon-vs (2026-05-01)
-
On ai-capex-to-power-and-materials-cascade:
"We had a massive build out of the infrastructure of the Internet in the late 1990s and into 2000. And what that caused was a lot of aggressive companies to do massive amounts of spending... look at the 2000 peak of Cisco. This is the most extraordinary chart ever. It took them 25 years to get back to that peak" — 2026-05-01-all-in-podcast-openai-misses-targets-codex-vs-claude-elon-vs (2026-05-01)
-
On anthropic:
"Love Mark said it right. All these companies should go public and get evaluation and focus on the higher order bit. And all of these tiki tacky mechanisms that people can use to stay private longer need to get a bullet put in its head. And these SPVs are the worst. The layered SPVs. On SPVs. On SPVs." — 2026-05-15-all-in-podcast-trump-xi-benioff-saaspocalypse-openai-apple (2026-05-15)
-
"the Chinese and the Americans have a very strong incentive to kind of divide up the world. Like, I think the future at this point, you have these critical resource inputs. China has a stranglehold. You have energy and intelligence AI, where America has a stranglehold on one and an effective stranglehold on the other, which is energy. And I think there's a trade there to be done. And as long as you can negotiate what a reasonable give and take is, I think they're just going to find common ground. Like, does China really need to be in Chile and Venezuela and Panama? Probably not. Does America have to have an incredibly strong point of view about the Strait of Malacca? Probably not. And so there's probably a trade to be done so that we can get their rare earths, they can get some more oil, everybody can be happy and we can all find abundance." — 2026-05-15-all-in-podcast-trump-xi-benioff-saaspocalypse-openai-apple (2026-05-15)
-
"18 months from Taiwan not being an important moment of conversation the way it is today. Why 18 months? Because we are at a point where we're probably one to two nanometers away from being able to do what we need Taiwan to strategically do for us. And so as we scale up our chip fabs, as we get more capacity and interestingly, There are these orthogonal technologies being developed... Today it's economic. And if you take that off the table, I think we'll have a very different attitude to Taiwan. That's number one. Number two, sell the chips. And the reason we should sell the chips is we want Nvidia to win. We do not want to give enough oxygen for Huawei to then all of a sudden emerge and have a version of a chip that works." — 2026-05-15-all-in-podcast-trump-xi-benioff-saaspocalypse-openai-apple (2026-05-15)
-
"the midterms, I think are probably going to swing more based on these recent gerrymandering rulings from the Supreme Court and what happened in the Virginia Supreme Court and what's going to happen in the state legislatures of these red states and some of these blue states, more than what's going to come out of this summit with President Xi. So let me just put that over here. And I think that money is getting organized. There was an article in the New York Times this week which really surprised me. But the largest donor in this election cycle is Andreessen Horowitz." — 2026-05-15-all-in-podcast-trump-xi-benioff-saaspocalypse-openai-apple (2026-05-15)
-
On spacex-ipo-index-inclusion-mechanic:
"All these companies should go public and get evaluation and focus on the higher order bit. And all of these tiki tacky mechanisms that people can use to stay private longer need to get a bullet put in its head. And these SPVs are the worst. The layered SPVs. On SPVs. On SPVs. That's pretty odd by the way. And I will guarantee you this, once SpaceX goes public, once anthropic goes public, once OpenAI goes public, you're going to see a litany of these lawsuits back and forth between the purveyors of these SPVs. They should not be allowed." — 2026-05-15-all-in-podcast-trump-xi-benioff-saaspocalypse-openai-apple (2026-05-15)
-
On ai-capex-to-power-and-materials-cascade:
"If you put those two things [recursive self-learning + scaling] together, I think that you start to potentially live out this idea that there's an order of magnitude improvement on a yearly basis. So like this new form of Moore's Law. So then the model quality just goes absolutely parabolically just like this, straight up." — 2026-05-22-podcast-all-in-podcast-spacex-s-2t-case-nvidia-s-shock-selloff-america (2026-05-22)
-
On (no topic linked):
"[Anthropic has] a decent lead on everybody else, whether it's three months or six months. Obviously they're probably six 12 months ahead of open source. Maybe they're three, six, nine months ahead of their contemporaries, but they have a lead." — 2026-05-22-podcast-all-in-podcast-spacex-s-2t-case-nvidia-s-shock-selloff-america (2026-05-22)
-
On cuda-moat-erosion-at-inference:
"across all evals, there is no single best model anymore. At the top of the leaderboard opus 4, 7, GPT, 5, 5, sonnet 4, 6 appear almost indistinguishable, separated by less than 3/10 of a percentage point overall...these things are getting commoditized way too quickly. And then you'd say, well, what's the ROI on all this incremental spend?" — 2026-05-29-podcast-all-in-podcast-anthropic-s-digital-god-pope-vs-ai-job-loss (2026-05-29)
-
On ai-capex-to-power-and-materials-cascade:
"There was an economic and capital moat to training that is going away. It's going away in two ways. One is because we're getting these domain specific architectures at the silicon layer. And then second, we're rebuilding all of the core components...Elon was like, we've rewritten the entire training complex in C and it's an order of magnitude increase...Those kinds of innovations are going to make the cost of model training so much cheaper that it's like, why would we stick to the $10 billion training runs when we can have the $10 million training runs?" — 2026-05-29-podcast-all-in-podcast-anthropic-s-digital-god-pope-vs-ai-job-loss (2026-05-29)
-
On cuda-moat-erosion-at-inference:
"a lot of the folks that we see now in the Fortune 1000 and increasingly the Global 1000, they want abstraction above it. They want to sit in a control plane. They want to have the flexibility because they don't know how it's going to shake out. They see all the money being invested at the model layers, but they see the model quality asymptote. So they're like, wait a minute, what are we supposed to do?" — 2026-05-29-podcast-all-in-podcast-anthropic-s-digital-god-pope-vs-ai-job-loss (2026-05-29)
-
On ipo-comeback-public-market-value-capture:
"Getting public sooner, having the scrutiny of public markets, having the scrutiny of having to deliver sharpens the focus. Steel sharpens steel. Iron sharpens iron. And I think innovation tends to get better." — 2026-06-06-podcast-all-in-podcast-the-ipo-comeback-why-tech-giants-are-finally (2026-06-06)
-
On ai-safety-as-regulatory-capture:
"If you want to be unexploitable, the best thing you could do if you're trying to build a super God is have three or four entities in a room, close the door behind you, and then dominate those other three or four entities. And then you set the rules. If the refs don't understand the game, you'll run over the game." — 2026-05-29-podcast-all-in-podcast-anthropic-s-digital-god-pope-vs-ai-job-loss (2026-05-29)
-
"There is no single best model anymore. At the top of the leaderboard Opus 4.7, GPT 5.5, Sonnet 4.6 appear almost indistinguishable, separated by less than 0.3 of a percentage point... which theoretically says that these things are getting commoditized way too quickly." — 2026-05-29-podcast-all-in-podcast-anthropic-s-digital-god-pope-vs-ai-job-loss (2026-05-29)
-
On ai-job-displacement-debate:
"It hasn't done anything measurable yet. At the end consumption of these tokens, nobody is standing there and saying, look at my filing. Here is the lift that I have gotten. Nobody has said that yet... instead people are realizing I have this cover now to go and clean up poor management over the last five and ten years." — 2026-05-29-podcast-all-in-podcast-anthropic-s-digital-god-pope-vs-ai-job-loss (2026-05-29)
-
"They want abstraction above it. They want to sit in a control plane... they want the flexibility because they don't know how it's going to shake out. They see all the money being invested at the model layers, but they see the model quality asymptote." — 2026-05-29-podcast-all-in-podcast-anthropic-s-digital-god-pope-vs-ai-job-loss (2026-05-29)
-
On ai-capex-to-power-and-materials-cascade:
"A gigawatt now costs $100 billion, guys. When I started this project it was 4 or 5 billion and it's increased by 20x. If you want to get all 3 gigawatts developed, I have to come up with $300 billion." — 2026-06-13-podcast-all-in-podcast-anthropic-s-fable-backlash-nationalizing-ai (2026-06-13)
-
On ai-capex-to-power-and-materials-cascade, mega-issuance-peak-to-ai-capex-derate:
"The marginal cost of production for a new internet user was effectively zero... AI is completely different. There is a real cost for every marginal user — everyone you stand up is taxing a GPU, needs electrons, needs memory. There needs to be all this critical infrastructure that even enables you to be on the field." — 2026-06-13-podcast-all-in-podcast-anthropic-s-fable-backlash-nationalizing-ai (2026-06-13)
-
On ai-safety-as-regulatory-capture:
"Companies need to start underwriting this next phase of AI: how do I have control? Do I want single-point-of-failure risk with respect to AI? The answer is you need broad diversity. A downstream scientist using the cloud APIs could trip it, and you'll get cut off from a very important source of business differentiation." — 2026-06-13-podcast-all-in-podcast-anthropic-s-fable-backlash-nationalizing-ai (2026-06-13)
-
"China has been smoothing energy consumption globally... we've kept a damper on $200 oil; we're sub-100. But if China runs out of reserves and has to go back into the spot market to buy an extra 3 million barrels a day, there's a very big risk oil gets well past 100, maybe between 150 and 200." — 2026-06-13-podcast-all-in-podcast-anthropic-s-fable-backlash-nationalizing-ai (2026-06-13)
-
On open-source-share-shift-bullish-for-compute, frontier-lab-vertical-integration-to-sovereign-ai-stack:
"when you use our harness with Claude, it was simultaneously 1.4x cheaper and 1.5x faster than just using anthropic Opus4.8 alone. But if you wrap the open source model with our software Factory, it was 16.4x cheaper. Now it was 3 times slower." — 2026-07-03-podcast-all-in-podcast-ai-sovereignty-wars-palantir-nvidia-deal-scotus (2026-07-03)
-
On frontier-lab-vertical-integration-to-sovereign-ai-stack:
"You can't rent intelligence from the same place that rents it to your competitor. ... It becomes the lowest common denominator problem where you and your competitors now look exactly the same." — 2026-07-03-podcast-all-in-podcast-ai-sovereignty-wars-palantir-nvidia-deal-scotus (2026-07-03)
-
On anthropic:
"Right now, OpenAI Equity, I think is more reasonably priced than Anthropic equity. ... the reason is that OpenAI can fall back on a really healthy consumer business." — 2026-07-03-podcast-all-in-podcast-ai-sovereignty-wars-palantir-nvidia-deal-scotus (2026-07-03)
-
On ai-roi-reckoning:
"He said, write down our token costs are doubling every 45 days. Honestly, what we're finding out is that you need to use a lot more tokens to get to this next iteration of improvement because we've effectively already asymptoted." — 2026-07-11-podcast-all-in-podcast-more-trillion-dollar-ipos-anthropic-3t-zuck-s (2026-07-11)
-
On ai-roi-reckoning:
"The answer as far as all publicly available data was that the actual ROI was somewhere between 0 and 2%. At some point you're going to have to show an ROI that's above the risk free rate of return. Otherwise you're going to have some angry investors on your hands." — 2026-07-11-podcast-all-in-podcast-more-trillion-dollar-ipos-anthropic-3t-zuck-s (2026-07-11)
-
On open-source-share-shift-bullish-for-compute, frontier-lab-vertical-integration-to-sovereign-ai-stack:
"the real business model is not in the foundational model anymore. It's at the application layer above, and it's in the infrastructure below, whether that's the cloud or whether that's chips." — 2026-07-24-podcast-all-in-podcast-the-fight-over-open-source-ai-anthropic-s-1-5b (2026-07-24)
-
On open-weight-sputnik-to-frontier-lab-derate, open-source-share-shift-bullish-for-compute:
"if they said no more open source, American companies cannot use open source... those things cost 50 to 100 times more than your other best alternative... So then Coca Cola has to get re rated. But then you look at the people who are selling those tokens... If the government comes in and actually tells you that there's no open source, their valuation will crater. Why? Because all of that revenue is artificially being propped up." — 2026-07-24-podcast-all-in-podcast-the-fight-over-open-source-ai-anthropic-s-1-5b (2026-07-24)
Sources
- 2026-05-01-all-in-podcast-openai-misses-targets-codex-vs-claude-elon-vs
- 2026-05-15-all-in-podcast-trump-xi-benioff-saaspocalypse-openai-apple
- 2026-05-22-podcast-all-in-podcast-spacex-s-2t-case-nvidia-s-shock-selloff-america
- 2026-05-29-podcast-all-in-podcast-anthropic-s-digital-god-pope-vs-ai-job-loss
- 2026-06-06-podcast-all-in-podcast-the-ipo-comeback-why-tech-giants-are-finally
- 2026-06-13-podcast-all-in-podcast-anthropic-s-fable-backlash-nationalizing-ai
- 2026-07-03-podcast-all-in-podcast-ai-sovereignty-wars-palantir-nvidia-deal-scotus
- 2026-07-11-podcast-all-in-podcast-more-trillion-dollar-ipos-anthropic-3t-zuck-s
- 2026-07-24-podcast-all-in-podcast-the-fight-over-open-source-ai-anthropic-s-1-5b
Related
Cross-links — fill in.