All-In Podcast · · 1h 17m

Anthropic Goes Public, Nvidia in Crisis, Markets Tank, Open Source Crackdown?

ChorusBring the market's conversation to your AI. Learn more

Summary

The hosts discussed Anthropic's upcoming IPO and whether its valuation will be driven by its own business or by broader market conditions. Chamath Palihapitiya argued that AI megacorps are integrating from applications down to data centers and power, citing Palo Alto Networks' purchase of open-source model makers, and that Nvidia's falling stock despite record earnings reflects a shift toward distributed silicon. David Friedberg and Jason Calacanis were bullish on Anthropic's enterprise growth and AI adoption, while the group split on the macro backdrop of rising yields, sticky inflation and a cautious Fed. The episode also covered proposed restrictions on open-source AI models, with Friedberg and Calacanis arguing such limits would entrench closed incumbents and misread how open weights work, and Chamath warning that disclosure rules would erode firms' cost advantages.

Key takeaways

  • Chamath Palihapitiya (bullish on Palo Alto Networks): Palo Alto's acquisition of an open-source AI team was smart; market rewarded a security firm venturing into AI.
  • Chamath Palihapitiya (bearish on NVIDIA): Nvidia's over-earning era may be ending as compute gets distributed across many formats.
  • David Friedberg (bullish on Anthropic): Anthropic has the best enterprise product, long runway and a multibagger setup for five-plus-year holders.
  • Jason Calacanis (bullish on NVIDIA): Nvidia's stock fell despite record profits, which makes it look cheap against its earnings.
  • Jason Calacanis (neutral on NVIDIA): Nvidia chips likely reached China via rogue salespeople and resellers, not a deliberate corporate backdoor.

What was said, by company and theme

33 views from 3 speakers. Every quote is checked against the episode's transcript (how).

NVIDIA

Across the shows: 179 speakers in other episodes, 186 bullish and 184 bearish calls, past 30 days

Where everyone stands on NVIDIA

  • Chamath Palihapitiya · Social CapitalBearish

    Nvidia's over-earning era may be ending as compute gets distributed across many formats.

    Chamath says Nvidia's stock is falling because investors realize its over-earning era may be over. His reasoning is that the AI buildout is moving toward a multitude of compute formats and distributed silicon, not concentrated GPU demand. He says this is independent of Jensen Huang. He also notes that Nvidia reported a record quarter yet the stock is down meaningfully from earnings day, which he treats as evidence of the shift.

    “This, I think, is why NVIDIA is trading down, is this realization that the over earning era of NVIDIA may be over.”at 2:01
  • Chamath Palihapitiya · Social CapitalBearish

    Nvidia's Nemotron open models are good but behind, and under-resourced on compute or talent.

    Chamath says Nvidia was starting to invest in its Nemotron open models, but its checkpoints lag. He argues Nvidia could have given Nemotron more compute. Whether the shortfall comes from compute or talent, the models are good but not great, and the Chinese GLM models are also not great. The gap is a limit on Nvidia's open-model play rather than a financial forecast.

    “But you could have given more compute to Nemotron... If you look at the checkpoints on Nemotron, they're unfortunately behind.”at 53:07
  • Jason Calacanis · LAUNCHBullish

    Nvidia's stock fell despite record profits, which makes it look cheap against its earnings.

    Nvidia beat on the top and bottom line. Revenue reached $116 billion and net profit reached $65 billion, both roughly doubling. Jason says most companies in the stock market are not worth what Nvidia earned in profit this quarter. Yet the stock lost about $2 trillion of market cap over thirty days, down 35%, and dropped further after the earnings report. The gap between earnings and share price implies the market is pricing the stock too low.

    “Let that sink in... Most companies trading in the stock market are not worth what Nvidia just made in profits this quarter.”at 59:48
  • Chamath Palihapitiya · Social CapitalUncertain

    Nvidia is exceptional, but leadership cannot be settled until congressional hearings on Dec 1-2 conclude.

    Chamath calls Nvidia's founder a legendary CEO who built the most important company in the most important industry, but says he made mistakes whose extent is unknown. Reinstating him into a public company whose value has fallen this much would wreck governance and put the company at risk. Nvidia must first navigate congressional committee hearings on December 1 and 2. Until those are done, no decisions can be made, and public protests are counterproductive.

    “He is exceptional, that company is exceptional, but they are in a very unique moment where you cannot make any decision. I repeat, you can't make a decision until all of the congressional hearings are done.”at 1:04:41
  • Jason Calacanis · LAUNCHNeutral

    Nvidia chips likely reached China via rogue salespeople and resellers, not a deliberate corporate backdoor.

    Jason points to reports that servers and cloud providers gave Chinese companies backdoor access to this hardware. He draws on his experience with how sales get flipped and passed along. From that he predicts that the investigation will show rogue salespeople and resellers operating in a gray area, not a deliberate company-level backdoor. He does not give a directional view on the stock.

    “This has always happened where sales of items get flipped and then provided to people. I bet you this is what's gonna come out. A bunch of rogue salespeople,”at 1:08:57

Palo Alto Networks

Across the shows: 10 speakers in other episodes, 9 bullish and 1 bearish calls, past 30 days

Where everyone stands on Palo Alto Networks

  • Chamath Palihapitiya · Social CapitalBullish

    Palo Alto's acquisition of an open-source AI team was smart; market rewarded a security firm venturing into AI.

    Chamath calls Palo Alto Networks' purchase of the team behind open-source models a smart move that rips off the Band-Aid. He notes the stock rose 8-9% at one point even though it is a security company buying an AI company. Nikesh's press release framed it as venturing outside security. Chamath reads the market reaction as a vote for companies building full AI stacks by acquisition during high disruption.

    “I think the deal that Palo Alto Networks did yesterday rips the Band Aid off, and I think was very smart.”at 0:37
  • Chamath Palihapitiya · Social CapitalBullish

    Palo Alto's big open-source bet is right and could make it a trillion-dollar business over a decade.

    Palo Alto Networks was one of the few stocks rising on a risk-off day (up 5.86%) after making a large open-source bet, which Chamath reads as the market rewarding a move that is both philosophically and mathematically sound. He sees the deal as a smart use of cash flow and stock to get into the AI game, and says the CEO has a good chance of building a trillion-dollar business over the next decade.

    “because they made a huge bet on open source. Freeburg is right. And the market will reward you.”at 1:12:16

Anthropic

Across the shows: 154 speakers in other episodes, 270 bullish and 136 bearish calls, past 30 days

Where everyone stands on Anthropic

  • David Friedberg · The Production BoardBullish

    Anthropic has the best enterprise product, long runway and a multibagger setup for five-plus-year holders.

    Friedberg says Anthropic is growing beyond internal expectations and is on track for $100B of revenue. He argues it has a lot of runway because it is under 1% penetrated among potential high-spending users. He says it has the best enterprise product, ahead of Google and OpenAI. He thinks its cost per unit of intelligence is falling, creating a natural monopoly moat. Open source is the biggest risk but would not eat the whole business. He sees roughly $50B of true float against a multiple of $100B demand at IPO. He says long-term holders could make a multibagger from this entry point.

    “if you're willing to buy the stock and own it for five plus years, you could hold a multibagger from from here at this at this entry point.”at 20:14
  • Chamath Palihapitiya · Social CapitalUncertain

    Anthropic is a great long-term business, but its next few months' valuation depends on the market, not the company.

    Chamath calls Anthropic a great long-term business that is doing the right things by going from dirt to application. He thinks its valuation over the next three or four months will depend mostly on market conditions rather than on Anthropic. He points to Nvidia, which had a record quarter while its stock fell, as an example. He sees elevated volatility, sticky inflation, a rising 10-year yield, and an uncertain Fed as reasons investors are skittish and de-risking.

    “the valuation of Anthropic over the next three or four months has mostly nothing to do with Anthropic.”at 22:15

Open-source AI models

Where everyone stands on Open-source AI models

  • Jason Calacanis · LAUNCH

    The disclosure rule shows a fundamental misunderstanding of open source: forkable, self-hosted models have no backdoor.

    Companies don't buy open-source models from China. They download them from GitHub or a hub, fork them and run them on their own servers, so they control them. Any backdoor would be spotted immediately by many researchers. Concern would be more reasonable about Chinese hardware such as Huawei chips, which are hardware-based. The proposed rule therefore reflects a basic misunderstanding of how open source works.

    “Friedberg, if you look at open sources, this shows a fundamental misunderstanding of how open source works. You're not buying it from the Chinese. It's on GitHub. It's on a website. You take it. You fork it. You put it on your own servers. You use it. You control it.”at 40:41
  • David Friedberg · The Production Board

    Curbing, restricting or labelling open-source AI would create regulated monopolies and threaten freedom and prosperity.

    Friedberg argues that efforts to restrict, label or litigate open-source AI would hand government agencies and incumbent closed-source companies control over what people can do. He compares it to Apache and Mozilla, which kept the internet open, and to Android, which put open software on about 70% of phones. In his view, the same choice now faces open AI, and suppressing open-weight models would limit prosperity and individual liberty and infringe free speech. Regulating harmful outputs is fine, but restricting the software itself is not.

    “any effort to try and curb open source, restrict open source, label open source, and give... the government the tools to create effectively regulated monopolies”at 41:51
  • David Friedberg · The Production Board

    Open-weight models are inert, auditable data, so labelling them by country of origin is technically misguided.

    Open-weight models are not executables or runtime software. They are a file of weights that you run inside your own environment, so no hidden backdoor can take over a computer. Anyone can inspect and test the weights, and builders are liable for whether the software works and is safe. Because the model's origin is hard to define on an anonymous hub like Hugging Face, the rules being discussed don't work. Government labelling of models is therefore a misunderstanding of the technology.

    “It is a set of data. It is a piece of data, a chunk of data, a CSV file with a bunch of stuff in it. That's all it is.”at 47:19
  • Chamath Palihapitiya · Social Capital

    Non-model-native companies should commoditise the model layer and make money in apps and chips instead.

    A company at Meta's scale that is far behind on frontier models should open-source or commoditise the model, since model-native firms have to make money from the model itself. Other companies can monetise through applications, chips and infrastructure. Chamath says spinning up a competitive model is much harder than it looks, which is why cash-rich public companies should stand up open-source model initiatives, with Meta the obvious play.

    “if I was an entrant at that scale with that balance sheet two years ago and I was far behind, it made a lot of sense to commoditize the model there if you were a non model native company.”at 55:17
  • Jason Calacanis · LAUNCH

    US open-source models will likely sit between Chinese leaders and French Mistral within a year.

    Jason ranks the open-weight field. Chinese models such as Kimi, DeepSeek and GLM are leading. A French model, Mistral, sits between them and the trailing US models such as Muse Spark, Nemotron and Reflection. He expects the US models to close some of the gap within a year, so that they sit somewhere between the Chinese leaders and the French model.

    “The French one is right between the three trailing US ones and the three leading Chinese ones. We'll sit here in a year, and The US ones will be somewhere in between.”at 57:47
  • David Friedberg · The Production Board

    Open-source proliferation is needed to stop AI from becoming a monopoly or oligopoly.

    Friedberg argues that a large, unconscious effort is under way to monopolize AI, which he calls the greatest technology advance in human history. He says AI will keep advancing whatever anyone does, so the way to prevent an oligopoly is to allow open-source proliferation and competition. He frames this as speculation, but it is backed by a clear argument that concentration would hurt everyone, regardless of political affiliation or economic position.

    “I'm speculating on the fact that there's a large scale unconscious effort to monopolize artificial intelligence, the greatest technology advancement in human history.”at 1:11:11

US stock market outlook

Where everyone stands on US stock market outlook

  • Chamath Palihapitiya · Social Capital

    Hard to justify being mega-long equities given rising yields, sticky inflation, elevated volatility and political uncertainty.

    Chamath lists factors he sees as bad for equities. The 10-year yield keeps rising, volatility is elevated after a 10-12% AMD drop on a short report, inflation is sticky, the Fed has little room to cut and may hike, and a new Congress creates uncertainty. He says the market is trying to will itself higher and that the balance of trade is to de-risk. He asks what setup would take the S&P from 6,500 to 8,000 and says it is hard to see.

    “the ten year unfortunately keeps ticking up. That's bad for equities.”at 23:28
  • Jason Calacanis · LAUNCH

    Stocks are bouncing along the bottom, and the AI-driven upside is underappreciated.

    Calacanis says he thinks stocks are bouncing along the bottom independent of a black swan event. He acknowledges a wall of worry from rates, debt, inflation, wars and political turnover, but argues the market is overheated in people's minds while the AI product velocity and adoption are underappreciated. He says he is long, which is why he takes a shot at the bull case.

    “I think we're bouncing along the bottom independent of, like, a crazy black swan type event, COVID, etcetera.”at 24:35
  • Chamath Palihapitiya · Social Capital

    Market momentum is turning down; he is risk-off ahead of AI-related congressional hearings and shocks.

    Chamath cites the day's tape as evidence: the S&P 500 down 1.39% and the VIX up 10.83%, with a rally only in a few names like Palo Alto. He says the balance of motion in the market is down, not up. He expects more economic and AI-trade shocks, specifically the first-order effects of the congressional hearings on chip diversion, which he thinks will clarify the picture after they happen. He also points to daily cyber hacks and shutdowns as added risk, so he is staying risk-off.

    “What I'm saying is I'm risk off because what I'm telling you right now is the balance of motion in this market is down, it's not up.”at 1:13:08

AI: overall outlook

Where everyone stands on AI: overall outlook

  • Chamath Palihapitiya · Social Capital

    Expects 8-10 AI megacorps that vertically integrate from land and chips up through applications.

    Chamath argues that over the next decade each AI megacorp will go from the application layer all the way down to the dirt. He cites Palo Alto Networks buying open-source model makers, Anthropic and OpenAI moving down into data centers and up into apps, SpaceX owning Cursor, and Meta owning the full stack. He expects a few haves who control critical assets, a rich app ecosystem with open and closed models, distributed silicon, and land and power as the binding constraint. He says the market is now betting on which companies become conglomerates, and that this explains recent stock flows.

    “is gonna go from the application layer all the way down to the dirt.”at 0:30
  • David Friedberg · The Production Board

    AI's biggest value is on the frontier solving new problems, not cutting costs in existing processes.

    Friedberg argues the biggest AI opportunity is on the frontier, where businesses and people can do things never done before, such as hard engineering or science problems. Most people instead use AI to improve existing processes and save money, which he calls a small return on token spend. He argues that frontier use could expand the frontier by 100x or 1000x. The return on tokens is much larger when they produce things humans have never produced before.

    “I think the biggest opportunity for AI is on the frontier.”at 31:33

Enterprise AI adoption & AI software

Where everyone stands on Enterprise AI adoption & AI software

  • David Friedberg · The Production Board

    Only about 1% of people use AI tools daily, so pessimism reflects poor understanding, not weak demand.

    Friedberg says it is striking how few businesses spend money with Anthropic and how few people use AI tools every day, roughly 1% of the population. He argues this explains why people remain pessimistic: they have not yet understood the technology. He expects digital intelligence to reinvent every category of commerce, as the internet did.

    “it's amazing how few businesses are actually spending money with Anthropic today ... call it 1% of the population, which explains why people remain pretty pessimistic because they haven't really understood it yet.”at 8:59
  • Jason Calacanis · LAUNCH

    AI token usage will grow at least 500x as workers shift to agents and hours of daily use rise.

    Calacanis says his 100 startups a year already use AI five to seven hours a day, working alongside agents more than humans. He expects big companies to follow in two or three years, with usage moving from minutes or an hour a day to six to eight hours and more people using it. He says productivity could rise 10-20x and revenue per employee could double every three years, as at Microsoft. He thinks the question is where the tokens come from, and that the market is underestimating the technology, comparing this to the dot-com era.

    “What that means is at least 500 times the token usage is what's gonna happen.”at 26:29

AI & jobs

Where everyone stands on AI & jobs

  • David Friedberg · The Production Board

    AI job-loss, welfare and UBI narratives are wrong because new industries emerge before old jobs vanish.

    Friedberg argues that AI can incrementally improve any job, but that does not mean mass job loss. He says that, as with past frontier moments, new jobs and industries get created before the old ones become irrelevant. This creates a bridge. He cites San Francisco becoming the second most populous US city during the gold rush, when people moved to the frontier because the value there was greater than in older industries.

    “You can incrementally improve any job. Which is also why the job loss narrative is bull ... it's the new jobs, the new industries get created before the old ones are irrelevant.”at 33:27
  • David Friedberg · The Production Board

    AI is creating more demand for jobs and capital than it destroys, so no jobpocalypse; the real risk is a K-shaped economy.

    Capital and talent keep flowing to the new AI frontier, as they did to San Francisco's earlier tech booms, before old industries are wound down. Because that frontier is attracting more capital and jobs than the old economy is losing, unemployment isn't rising and there's no rush toward UBI. The media narrative is wrong. The real problem is a K-shaped economy: not everyone is riding the wave, and the resulting backlash could stall the frontier.

    “And as a result, there's much more demand for jobs, for capital in that new frontier then there is a loss of jobs in the old world.”at 34:56

Software & SaaS

Where everyone stands on Software & SaaS

  • Chamath Palihapitiya · Social Capital

    Markets will stop caring whether a firm is software or services and reward high gross margins and cash flow.

    Chamath argues that the software-versus-services distinction is a relic used by people who are anxious about change. He says a services business with 99% gross margins and 70% free cash flow margins would beat product companies. In a world where AI disrupts everything, investors should value numerical characteristics such as gross margin and free cash flow conversion. Companies without those metrics will be decimated regardless of what they do.

    “people are gonna stop caring about what kind of business you are from a business model perspective and care a lot more about what kind of business you are from a numerical perspective.”at 6:33

Earnings & corporate profits

Where everyone stands on Earnings & corporate profits

  • Chamath Palihapitiya · Social Capital

    Buybacks signal management lacks ideas; markets reward acquisitions that defend moats during disruption.

    Chamath argues the era of buybacks boosting EPS and steady stock gains is over. Companies that buy back stock are admitting they do not know what to build or buy. In high-disruption periods, acquisitions that defend existing moats and build new ones get rewarded by the market, as with Palo Alto's $4.5B deal. Companies with large free cash flow that avoid AI model and data-center investment will be punished through selling pressure.

    “If you take your free cash flow and you spend it on buybacks ... what you're basically saying is I don't have a clue.”at 10:17

IPOs, M&A & deal activity

Where everyone stands on IPOs, M&A & deal activity

  • Jason Calacanis · LAUNCH

    M&A is open under Trump, but a window may close before any 2028 Democratic administration.

    Calacanis says big M&A was off the table under Biden but is back under Trump. He thinks there may be only about a six-month window before it closes again if investors assume Democrats win in 2028. He sees this as the reason companies should play the game on the field and make acquisitions now.

    “The market wasn't allowed to do big m and a. That was off the table during Biden. Now it's back on the table for Trump. We got maybe six month window before it closes again”at 13:00

US rates & the Fed

Where everyone stands on US rates & the Fed

  • Chamath Palihapitiya · Social Capital

    Fed under Kevin Warsh has little room to cut and may hike because inflation is sticky.

    Chamath argues that inflation is sticky, which limits the Fed's ability to cut rates. He says that if Kevin Warsh wanted to cut, he does not have much room, and if anything he is going to hike. He ties this to his view that rates and inflation are bad for equities.

    “To the extent [Kevin Warsh] wanted to cut, I don't think he's got much room, and if anything, he's going to hike.”at 23:48

Antitrust & tech regulation

Where everyone stands on Antitrust & tech regulation

  • Jason Calacanis · LAUNCH

    Political turnover may bring breakups of big tech if megacorp consolidation continues.

    Calacanis says the people coming into power hate AI, big tech and capitalism. He argues that if Chamath's megacorp thesis plays out, a future conversation in two or three years could feature populist figures demanding that big tech companies be broken up. He sees regulation, banning or breakups as the real headwinds to AI adoption.

    “They're gonna be in the breakup big tech. We could be having a conversation in two or three years about”at 24:58

US Treasuries & yields

Where everyone stands on US Treasuries & yields

  • Chamath Palihapitiya · Social Capital

    10-year Treasury yield more likely at 5.5-6% than 4.5%, a headwind for all equities.

    Chamath responds to Jason's pushback by saying he thinks the 10-year is more likely to be at 5.5 or 6 percent than 4.5 percent. He says that has a mathematical implication for all equities because higher discount rates compress valuations. This is part of his case that the stock market has not bottomed.

    “it's more likely that the ten year is at five and a half or 6% than it is at four and a half percent, that just has a mathematical implication to all equities.”at 28:20

AI regulation & safety

Where everyone stands on AI regulation & safety

  • Chamath Palihapitiya · Social Capital

    A disclosure rule for open-source model use is a meaningless trial balloon that would hurt American companies.

    A website disclaimer that a company uses an open-source model changes nothing. Markets reward companies that are smart, budget-friendly, ROI-oriented and own their technology. Forcing firms to disclose cheaper open-source models, or to avoid them, would cost them their edge against competitors. Chamath thinks Washington lacks technically literate people to see these implications, so the proposal is a trial balloon that should be reconsidered.

    “To not say that you'd be using cheaper open source alternatives to retain your edge is more destructive to American companies. Is that what we actually want?”at 40:17

Venture capital & startups

Where everyone stands on Venture capital & startups

  • David Friedberg · The Production Board

    VC money has favoured closed frontier labs over open-source AI startups, distorting incentives.

    Over the last couple of years it was much easier to raise capital as a private frontier model company than as an open-source AI company. Open-source plans have no clear revenue model, while closed frontier labs can pitch the standard venture business model. Friedberg argues this incentive pushed talent and capital toward closed labs and left open-source AI underfunded, unlike China, where the game theory differs.

    “I think it was a lot easier over the last couple of years to raise capital as a private frontier model company than go out and say, I'm gonna build an open source AI company.”at 51:38

Frontier AI labs & model race

Where everyone stands on Frontier AI labs & model race

  • Jason Calacanis · LAUNCH

    Frontier labs' huge pay packages drew top talent away from open-source efforts, which can't match them.

    Top computer scientists were offered extreme pay packages by OpenAI and Anthropic. Jason gives an example of a recent undergraduate offered about $10 million a year in equity over four years, and of a $40 million package. Such pay has never existed before in the industry. Open-source organisations lack the dollars and the market cap to match these offers, so talent jumped to proprietary labs and open-source competition suffered.

    “These people got offered pay packages... This is a really good computer science undergraduate who got a $40,000,000 package. Like, this has never existed in the history of of our industry.”at 56:40

Across the shows

What other episodes Chorus covered said about the same companies in the 30 days up to this one.

  • NVIDIA179 speakers · 186 bullish, 184 bearish calls
  • Anthropic154 speakers · 270 bullish, 136 bearish calls
  • Palo Alto Networks10 speakers · 9 bullish, 1 bearish calls

Ask your AI who they are and what they said

Topics

  • 10-year yield
  • AI acquisition by security firm
  • AI adoption rates
  • AI job displacement
  • AI megacorp vertical integration
  • AI talent flow to frontier labs
  • AI usage and token consumption growth
  • Fed rate path
  • M&A window under political cycle
  • Nvidia leadership and governance
  • Nvidia open-source model quality
  • Nvidia valuation vs earnings
  • US stock market outlook
  • US vs Chinese open-weight model gap
  • VC funding for open-source AI labs
  • big tech breakup risk
  • buybacks vs acquisitions
  • chip demand durability
  • chip diversion to China
  • enterprise product lead and growth runway
  • frontier vs incremental AI use
  • market direction and risk-off positioning
  • open-source AI restrictions
  • open-source acquisition bet
  • open-source model disclosure rule
  • open-source model security
  • open-source model strategy for incumbents
  • open-source proliferation vs AI oligopoly
  • open-weight model auditability
  • post-IPO valuation
  • software vs services business model

More from All-In Podcast

All All-In Podcast episodes

Ask your AI about this episode

Chorus gives Claude, ChatGPT and other assistants every view from this episode and every other show Chorus follows. Ask who agrees, who changed their mind, and what else was said about the same companies.

A weekly email with the calls that mattered from the most important voices in investing. Unsubscribe any time. Privacy Policy

Chorus reports what speakers said. It isn't investment advice. How Chorus works · About Chorus