OpenAI wants to be worth $1 trillion. The government wants to slow-walk its most powerful model. And the CFO wants to wait until 2027 to go public. Something’s gotta give.
Let’s start with the elephant in the room: Sam Altman has reportedly refused to lower OpenAI’s $1 trillion valuation target. Not a penny less. That’s like walking into a car dealership, pointing at the priciest vehicle, and insisting you’ll only pay sticker — even when the mechanic says the engine’s overheating.
The engine? Massive infrastructure commitments that started at $600 billion in May and ballooned to $1.15 trillion by late June. Meanwhile, CFO Sarah Friar is pushing for a delay, citing missed growth targets and spending that’s outpacing revenue. Altman wants a 2026 IPO. Friar wants 2027. Prediction markets (Kalshi) give only a one-in-three chance of a 2026 listing, but a high likelihood by June 2027. Ouch.
Then there’s Washington. On June 25, the U.S. government asked OpenAI to delay the full public rollout of GPT-5.6. OpenAI agreed to phase it, limiting access to a small group of vetted partners. This isn’t just a request — it’s a new normal. Federal oversight is now shaping the trajectory of frontier AI models, and the parallel clearance of Anthropic’s Mythos 5 for vetted organizations shows the government is playing favorites. Or at least playing gatekeeper.
Who wins here? Anthropic, obviously. It not only overtook OpenAI at a $965 billion private valuation and filed its own S-1, but now its flagship model gets a green light while OpenAI’s gets phased. Who loses? Altman’s timeline, OpenAI’s internal harmony, and maybe investors who expected a quick billion-dollar exit.
The real question: Will Altman bend on the valuation or accept a 2027 delay? If he holds firm, he risks alienating Friar and triggering a CFO exit. If he caves, he loses face. Either way, the message is clear: frontier AI is no longer just a tech race — it’s a regulatory chess match, and OpenAI just got checkmated into a slower game.