Private AI queues for the exits: SpaceX, OpenAI and Anthropic line up to go public
A week ruled by capital. OpenAI CFO Sarah Friar, Coatue's Thomas Laffont and two live All-In panels all circled the same turn: the decade of staying private is ending, and SpaceX, OpenAI and Anthropic are heading for public markets Laffont calls 'the great antiseptic.' Compute scarcity was the other refrain — and away from the money, Dara Khosrowshahi, Dan Loeb and John Lennox each marked a line the machines can't cross.
Jun 1 – Jun 7, 20269 episodes4 shows203 claims on the record39 new predictions
01
Private AI queues for the public markets
The week's loudest signal was an exit door swinging open. OpenAI CFO Sarah Friar played down the IPO 'race' with rivals — an IPO is a milestone, not a destination — even as Anthropic's confidential S-1 hit the same week. Coatue's Thomas Laffont was blunter: SpaceX, OpenAI and Anthropic will all go public inside a year, and the public market will be the reckoning that tests them against short-sellers. On the All-In panels, David Sacks saw 'the pendulum swinging back' from stay-private-forever, while Brad Gerstner showed secondary volume at double its 2021 peak, now trading at a premium to net asset value. His worry wasn't valuation but legitimacy: you can't build trillions in private wealth while most of the country feels locked out.
“In the end, an IPO, I say this to the team all the time, it's a milestone. It is not a destination.”
“We all think that we need to go public sooner. The reason I think, it is destabilizing when you're creating trillions of dollars in private value and 80% of America think it's a scam where they're left out and left behind.”
Underneath the liquidity story ran a sharper argument about information. Gavin Baker, who has invested on both sides, put it flatly: there is no good reason to stay private longer. The rot is sycophancy — private investors, he said, tell management teams 'what they need to hear' to keep their spot in the next round, and he cited Mark Zuckerberg's own regret that rigorous public-market scrutiny would have killed Facebook's HTML5 mistake faster. Jason Calacanis agreed the distortion is structural: private CEOs stop getting clean information because everyone fears losing access. On the IPO panel, Friedberg made the constructive case — public-market scrutiny sharpens execution rather than dulling it. The through-line across both panels: staying private buys founders freedom at the cost of the truth.
“I don't think there is actually a good reason to stay private longer.”
Compute is the binding constraint — on Earth and off it
The other constant was scarcity of the physical layer. Friar was stark: if you want to buy more compute in 2026, good luck — she doesn't know where else to find it, and defended buying years ahead of demand. The crunch surfaced everywhere. Uber's Dara Khosrowshahi said the company blew through a full year's AI budget in a single quarter; Laffont noted the hyperscalers are now bankrolling the very labs disrupting them. That pressure is pushing the frontier off-planet and off-GPU. Planet's Will Marshall argued that once launch costs hit $200-300 a kilogram it will be simply cheaper to put data centers in orbit, while Cerebras's Andrew Feldman claimed his non-GPU wafer runs OpenAI inference 15-18x faster than a GPU. The bottleneck, not the model, is the story.
“in 26, if you want to buy more compute, good luck to you. Tell me, because I don't know where else to find it.”
As the money raced to price the machines, three guests on three different shows drew the boundary. On Diary of a CEO, Oxford mathematician John Lennox insisted the systems only mime us — they are not conscious, they merely simulate intelligence — and told the transhumanists chasing digital immortality they were too late, the problem of death already answered at the resurrection. Investor Dan Loeb, on All-In, bet the human stays in the loop: AI agents will never look a management team in the eye and read them the way a person can. And Uber's Dara Khosrowshahi named social consent as the real limit on autonomy — the industry has to move at the pace society will accept, or provoke a backlash. Different rooms, one instinct: the machine's ceiling is human.
“The machines are not conscious, but they simulate intelligence.”
Stanford's Dr. Nolan Williams reframes depression as a fixable brain circuit rather than a serotonin deficit — his five-day SAINT/SNT protocol packs seven and a half months of TMS into 50 hours, with 60-90% remission. HUBERMAN LAB →
Bill Ackman says the best businesses now trade at the lowest multiples: Amazon, Meta and Microsoft dismissed as 'old-fashioned' in the OpenAI era — and he wants to grow Howard Hughes into a Berkshire-style compounder over 50 years. ALL-IN →
Dan Loeb calls NVIDIA the market's 'safe short' yet undervalued on the next two to three years of earnings, and warns valuation-only shorts get squeezed when Reddit piles in. ALL-IN →
Coatue's Thomas Laffont floats a '10X paradox': a unicorn has roughly an 8% shot at a 10x, but a $100B-plus centicorn has about 31% — at the top, scale begets scale. ALL-IN →
The Forge-Schwab tie-up would open private names like SpaceX to 46 million retail accounts via interval funds with $500 minimums — roughly 60 companies in all. ALL-IN →
Cerebras's Andrew Feldman says a GPU-shaped chip has 'approximately zero' odds of beating NVIDIA; his dinner-plate wafer runs OpenAI's inference 15-18x faster than a GPU. ALL-IN →
The record
Every episode this week
The issue above is written from these structured records — nothing else. Each card is the full breakdown: claims, predictions, quotes, all timestamped.
Tokenization and blockchain will eventually make secondary fund trading more efficient, though the bigger near-term need is name-level liquidity rather than fund-level trading.
Retail investors who YOLO into peak-valued late-stage privates at the top of the market will feel disappointed when returns do not go up 3X post-IPO.
Gavin Baker · ALL-IN · within a few years · source 23:22
Hundreds of billions of dollars of new late-stage demand from long-only mutual funds will enter public markets once major private companies go public and lockups expire.
Many companies in the panelists' portfolios will go public at $1-5B valuations as the IPO market reopens.
First mentions
New to the graph this week
Subjects no indexed episode had discussed before this week.
Uber×17secondary markets×16artificial intelligence×11IPOs×9God×7Christianity×7Compute scarcity×6Depression×6Planet Labs×6space-based data centers×6Pershing Square×5Cost of capital×5leadership×5Magnificent 8×5Transcranial magnetic stimulation×5ChatGPT×4
How it's built
Written at ingest, never at query time
The week's episodes are diarized into structured records; one editorial pass writes this issue from those records when the week's ingest lands. A grounding gate greps every quote above against its episode record — a quote that isn't verbatim fails the build.
the data behind this issue
# this issue, as structured data — stories, citations, receipts
curl 'https://aipod-two.vercel.app/digest/2026-w23.json'# the records it was written from
curl 'https://aipod-two.vercel.app/api/episode/<id>'# any id in episode_ids