Does OpenAI Need a Bailout? Mamdani Wins, Socialism Rising, Filibuster Nuclear Option
Speakers
Chamath Palihapitiya
— CEO, Social Capital
David Sacks
— General Partner, Craft Ventures
Brad Gerstner
— CEO, Altimeter Capital
Jason Calacanis
— Angel Investor / Founder, LAUNCH
David Friedberg
— CEO, The Production Board
Summary
Brad Gerstner joins the All-In hosts days after his BG2 question to Sam Altman was blamed for cracking the AI trade. The first half debates whether OpenAI's $1.4 trillion of commitments are financeable, whether its CFO's federal backstop remark implied a bailout, and why Jensen Huang says China is going to win the AI race. The second half turns to markets, covering a risk-off year-end rebalancing phase, a cracking low-end consumer, still-restrictive real rates and growing dispersion inside the Mag 7. It closes on Zohran Mamdani's New York win, the rise of socialism among indebted young voters and whether Republicans should end the Senate filibuster.
- Gerstner argues OpenAI's $1.4 trillion of commitments is a misleading aggregate: roughly half sits with partners, it spans five to six years, and the deals can be recut to match revenue.
- David Sacks says there will be no federal bailout for AI and none is needed, because five frontier labs compete and any failure is simply replaced.
- Sacks pushes for a single federal AI framework to preempt a 50-state patchwork he says would let four blue states write the rules for the whole country.
- Chamath Palihapitiya calls a two-to-three-month risk-off phase driven by AI-capex ROI uncertainty and year-end tax-loss harvesting pulled forward to mid-November, with risk-on returning in February.
- Gerstner cuts Altimeter from extra-large to medium or medium-small positioning on consumer weakness, 2009-level credit card delinquencies, wobbling regional banks and wider credit spreads, while staying long compute.
- Gerstner reads the 10-year TIPS as showing highly restrictive policy and expects three to four rate cuts, GDP reacceleration and inflation rolling back under 3%.
- Chamath flags Bitcoin breaking below $100,000 with another 5% to 10% of downside, and growing dispersion inside the Mag 7 after Q3 earnings.
- Jason Calacanis and Sacks clash over whether spiking youth unemployment and record October layoffs are caused by AI or by post-ZIRP corporate efficiency.
- The panel discusses Mamdani's narrow NYC win, student-debt-driven support for socialism, and the case for ending the filibuster if the shutdown continues.
Ideas
Risk-off now, risk-on by February.
Palihapitiya says the market has entered a risk-off, rebalancing phase for two reasons. First, investors are trying to digest the enormous AI capex, build models that predict the ROI of that spend, and work out how it hits future earnings. Second, year-end behaviour has been pulled forward: consolidating, booking wins and tax-loss harvesting that used to start in mid-December now starts in mid-November, so the price action arrives early. He expects the risk-off phase to last at least two or three months and the market to be back firmly in risk-on mode by February, and he warns that in the meantime people will overblow every random little headline, such as the OpenAI backstop story.
Mag 7 dispersion favors Google.
Palihapitiya sees real dispersion opening up inside the Mag 7 after this quarter of earnings rather than one uniform AI trade. Google's earnings were phenomenal and its AI numbers were blazing hot; Meta (Facebook) was terrible; and Apple is in the strange position of appearing to cede its AI business to Google and pay it billions of dollars a year, mirroring the billions Google already pays Apple for search distribution. His conclusion is that the market now has to sort out which of the large stalwarts actually converts the AI capex cycle into future earnings, and that investors should stop treating the group as a single position.
Mag 7 dispersion favors Google.
Palihapitiya sees real dispersion opening up inside the Mag 7 after this quarter of earnings rather than one uniform AI trade. Google's earnings were phenomenal and its AI numbers were blazing hot; Meta (Facebook) was terrible; and Apple is in the strange position of appearing to cede its AI business to Google and pay it billions of dollars a year, mirroring the billions Google already pays Apple for search distribution. His conclusion is that the market now has to sort out which of the large stalwarts actually converts the AI capex cycle into future earnings, and that investors should stop treating the group as a single position.
Mag 7 dispersion favors Google.
Palihapitiya sees real dispersion opening up inside the Mag 7 after this quarter of earnings rather than one uniform AI trade. Google's earnings were phenomenal and its AI numbers were blazing hot; Meta (Facebook) was terrible; and Apple is in the strange position of appearing to cede its AI business to Google and pay it billions of dollars a year, mirroring the billions Google already pays Apple for search distribution. His conclusion is that the market now has to sort out which of the large stalwarts actually converts the AI capex cycle into future earnings, and that investors should stop treating the group as a single position.
AI sector healthy, bubble narrative false.
Sacks argues there will be no federal bailout for AI and that none is needed. There are five major frontier model companies and new ones are being formed all the time, so if one gets over its skis and goes bankrupt the chips fall where they may and the others replace it. In his view that makes AI maybe the healthiest, meaning most ruthlessly competitive, sector of the entire American economy. He separately rejects both dominant market narratives, that AI is a giant bubble and that superintelligence is about to replace everyone, as astroturfed by doomer-funded organizations and logically contradictory, since the sector cannot be simultaneously fake and powerful enough to justify the spend. He says he believes neither, which makes him a buyer of neither the bust story nor the mania.
Betting on the AI compute supercycle.
Gerstner says he is betting on the supercycle, which he calls the biggest of all of our lives, and that investors do not have to make the call today about which single company wins. He is an investor across OpenAI, Anthropic, Google, Microsoft and Nvidia, and Altimeter has owned all of compute for three years. He frames the roughly $4 trillion of AI infrastructure Jensen Huang expects to be built over five years as ten times the size of the Manhattan Project, but privately funded, with power as the real gating issue. He believes the end market is as big or bigger than current estimates, but insists it will not be a straight line up and to the right: there are recurring moments of risk-off panic as there were with internet, mobile and cloud, and there is a massive conviction tax paid by investors who lack conviction and sell when these things are down.
Google wins with Gemini and Apple.
Within his AI book Gerstner is specifically positive on Google. He likes that Sundar Pichai is coming off the mat swinging, expects Gemini 3 to be great and thinks Google may even make it free, and he welcomes Apple paying Google to make Siri better because it is both a great consumer experience and a second distribution annuity on top of the search payments. The same free, ad-funded distribution is the structural headwind he concedes for paid consumer AI subscriptions priced at $20 a month, which is why he prefers owning the platform that can give the product away rather than betting only on subscription revenue.
Trimming equity exposure after huge rally.
Gerstner is cutting gross exposure after a very large move. The Nasdaq was down about 20% year to date on an intraday basis in April and is now up 20%, a roughly 40% swing in a few months, with the S&P making a similar round trip. Altimeter went from small positioning early in the year when tariffs were the worry, to extra large from May once trade and the tax bill looked resolved, and is now back to medium or medium-small. The reasons for trimming: signs the consumer is pulling back, seen in Chipotle, Cava and JetBlue commentary, a two-tier economy where the low-end consumer is faltering, US credit card delinquencies back at 2009 levels, regional banks rolling over and credit spreads blowing out, plus multiples that have expanded a lot. He is not exiting the AI trade, he is simply making position size smaller for a pause over the next couple of months.
Restrictive real rates mean rate cuts.
Gerstner points at the 10-year TIPS to argue that policy is still in highly restrictive territory: the Fed is keeping the market tight because it sees indices at all-time highs and AI stocks ripping, even though under the surface there is real weakness in credit, the low-end consumer and the labor market. He treats that restrictiveness as remaining firepower rather than a threat. Looking into next year he expects three to four rate cuts, a reacceleration of GDP and inflation continuing to roll over on one-time effects, and he takes the under against Jason Calacanis on inflation staying above 3% next November. He also argues the number one affordability issue for the administration starts with getting interest rates lower.
Bitcoin breaks 100k with more downside.
Palihapitiya points to Bitcoin as the other market that is breaking in this de-risking phase. It is about to break through $100,000 to the downside, and because that level is a psychological barrier he thinks the break probably has another five or ten percent more to run on the downside. He frames it as part of the same broad risk-off, rebalancing regime he expects to last two or three months into February, in which everything has to get sorted out.
Up Next
This All-In Podcast video, published November 07, 2025, features Chamath Palihapitiya, David Sacks, Brad Gerstner discussing SPY, GOOG, META, AAPL, AI-SECTOR, MSFT, NVDA, TIP, BTC. 10 trade ideas extracted by AI with direction and confidence scoring.
Speakers: Chamath Palihapitiya, David Sacks, Brad Gerstner · Tickers: SPY, GOOG, META, AAPL, AI-SECTOR, MSFT, NVDA, TIP, BTC