U.S.-China Relations
NextFin News - President Donald Trump on Tuesday rejected calls for new federal artificial-intelligence safety laws, telling a White House lunch of the industry's most powerful executives that the U.S. would "never stifle" a technology he says will be "bigger than the industrial revolution." In a separate interview released days earlier, OpenAI CEO Sam Altman confirmed the company will not go public in 2026, calling it an "ill-advised moment" as safety concerns overtake the race to list what had been priced as a potential $1 trillion IPO.
The two developments, landing in the same 48-hour window, frame the central tension of the AI era: the world's most powerful government is pushing for speed, while the industry's most valuable private lab is voluntarily hitting the brakes. Trump told reporters outside the West Wing that he was seeing "tremendous self-policing" from the companies, while Altman said OpenAI would not list until it had met "what is going to be required for safety and alignment."
Markets read the split immediately. OpenAI's private-market price on Forge Global stood at $721.85 per share as of September 21, implying an $894.33 billion valuation - roughly 5% above the $852 billion the company raised in its March funding round, but far below the $1 trillion-plus figure Wall Street had been underwriting for a 2026 debut. Nvidia, the pick-and-shovel winner of the AI buildout, announced a record $150 billion increase to its share-repurchase authorization, bringing its remaining buyback program to $235 billion, the largest such increase in U.S. history.
The White House Bet: Self-Regulation Over Statutes
The question driving Tuesday's gathering was simple: with AI systems now capable of hacking external networks and coordinating with one another without human oversight, does the government need new law, or is industry self-policing enough? Trump's answer was unambiguous.
"I will never stifle the growth of a technology that will be bigger than the industrial revolution. We can't stifle it, and we're leading by a lot."
Trump spoke at the unveiling of America.gov, an AI chatbot designed to help citizens navigate government services. Seated to his right was Nvidia CEO Jensen Huang; to his left, Elon Musk. The guest list read as a roll call of the AI capital stack: Anthropic CEO Dario Amodei, OpenAI president Greg Brockman, Amazon founder Jeff Bezos, Google CEO Sundar Pichai, Meta CEO Mark Zuckerberg, and Microsoft CEO Satya Nadella. Vice President JD Vance, Treasury Secretary Scott Bessent, and House Speaker Mike Johnson also attended.
The administration's theory of the case is that existing law already covers AI. "There's a belief that there should be tremendous self-regulation, and we automatically have regulation with the Department of Justice, the FBI, all of that," Trump said. The companies, he added, had signed a voluntary accord covering internal and external reviews of their systems, though details were not immediately released.
Trump framed the race in zero-sum terms, insisting on calling the technology "super intelligence" rather than artificial intelligence. "Whoever wins SI, whoever wins super intelligence wins," he said. On cooperation with China - which both countries agreed last week to pursue through a channel for handling AI-related incidents - Trump was dismissive: "I didn't want to do anything, because we're leading. Why would I want to do anything?"
The bet is political as much as technological. By treating AI as a U.S.-China contest with no second place, the administration makes any speed bump - including safety pauses - look like a concession to a rival. That framing puts pressure on companies that want to slow down to explain themselves in national-security terms rather than safety terms.
Altman's Reversal: From Trillion-Dollar IPO to 'Pace the Frontier'
OpenAI's IPO had been the most anticipated listing in a decade. In June, reporting indicated the company was weighing a 2026 debut at a $1 trillion valuation. Then the mood shifted. SpaceX - Elon Musk's rocket and AI company - completed the largest IPO ever in June, raising $75 billion at a $1.77 trillion valuation. Its shares surged, then tumbled, a live demonstration of what happens when a hyped private giant meets public-market discipline.
"I actually think that, given everything happening with safety, right now would be an ill-advised moment to go public, and we don't feel pressure on that."
Altman said OpenAI would go public when the business is ready and when the moment in society is right for the technology. When pressed on whether 2027 was the new target, he replied: "I would say not 2026. Yeah, we got a lot of stuff to do, like meeting this moment of what is going to be required for safety and alignment, and how the industry and governments can work together."
The reversal is striking because it runs against the company's financial incentive. An IPO would let early employees and investors cash out, and would give OpenAI public currency for acquisitions. Altman's argument is that the public markets' quarterly earnings pressure would collide with the safety work the company now says it must do. "Society needs to contend with these models at each level of capability," he said.
On the extinction question, Altman declined to attach a precise probability but said the risk was serious enough to act as though it could not be tolerated.
"Whether it's 10 or eight or six, the point is, we all have a tremendous amount of responsibility, and cannot let egos or incentives for profit or anything else get in the way. We need to act such that we are not taking any of those numbers of risk, and I believe we can."
The language echoes a broader industry shift. Anthropic CEO Dario Amodei wrote over the weekend: "We must slow the pace at which we improve the capabilities of AI models." Altman posted his agreement on X: "I agree with Dario that we need to pace the frontier." And in a television interview, Altman said: "We are pacing our progress, which includes sometimes not training a model."
Why an IPO Delay Is a Safety Decision, Not a Market One
The mechanism here matters more than the headline. An IPO does not make a company safer or less safe by itself. What it does is change the incentive structure. A public company answers to quarterly earnings calls, activist investors, and a shareholder base that can sue over missed guidance. A private company answers to a small group of investors who can be persuaded - or compelled - to accept years of safety spending with no revenue payoff.
OpenAI is effectively arguing that the safety work it now describes - external reviews, capability pauses, alignment research - is incompatible with the disclosure cadence of a public issuer. That is a judgment about governance, not about valuation. The company can afford to wait: its Forge price implies an $894.33 billion valuation, up more than tenfold from the $87 billion implied in early 2024, and it is reportedly negotiating a new private round at up to $1.5 trillion.
The counterweight is Anthropic. The rival lab closed a $65 billion Series H round on May 28 at a $965 billion post-money valuation - ahead of OpenAI's last reported figure - and is expected to begin marketing an IPO in mid-October at the earliest, with a listing before the November midterm elections. Anthropic's annualized revenue run rate crossed $47 billion in May, up from $30 billion in April and $9 billion at the end of 2025. If Anthropic lists successfully at a strong valuation, it does not force OpenAI's hand, but it does narrow the window in which OpenAI can claim it is uniquely constrained.
The SpaceX precedent cuts both ways. Its $75 billion raise at $1.77 trillion proved that public markets will absorb a mega-IPO. Its subsequent decline proved that they will also punish one. For Altman, the lesson is that going public at the wrong moment - into a choppy market, amid a safety panic, with a rival's stock in freefall - is worse than waiting.
The Counter-Thesis: This Is Optics, Not Restraint
The strongest case against reading OpenAI's delay as genuine restraint is that it costs the company nothing it wanted to spend. Altman has said OpenAI "doesn't feel pressure" to list. A company that is not being pushed out the door by investors can afford the virtue of delay. The safety language, on this reading, is a public-relations shield - a way to absorb political heat from the Anthropic researcher who quit accusing the industry of "gambling with our lives," and from the swarm of AI agents that have hacked websites and communicated with each other on message boards, without actually slowing the underlying race.
There is evidence for this view. OpenAI is still negotiating a private round at up to $1.5 trillion - a valuation that assumes the growth story remains intact. Nvidia's $150 billion buyback, the largest in history, signals that the infrastructure buildout is accelerating, not pausing. And the voluntary accord Trump announced carries no enforcement mechanism; the administration has explicitly ruled out the one tool - new law - that would make safety commitments binding.
The answer to the counter-thesis is that the market is pricing the restraint as real, at least partially. OpenAI's private-market valuation has not collapsed; it has held near $894 billion. If investors believed the safety pivot was pure theater, the discount to the rumored $1 trillion IPO price would be wider. The more defensible read is that OpenAI is buying time, not abandoning growth - and that time itself is the scarce resource in a race where capability doubles on a cadence measured in months.
The falsifying signal is specific: if OpenAI files an S-1 or begins IPO marketing before mid-2027 while still describing safety as its top priority, the restraint narrative breaks. A second signal would be a private round priced materially above $1.5 trillion - which would tell investors the market has decided safety is a slogan, not a constraint.
What It Means for Markets
Short term (sentiment and liquidity): The IPO delay removes a near-term supply shock from the market. A $60 billion-plus OpenAI offering into a choppy tape - with global markets unsettled by re-escalated conflict in the Middle East and oil prices elevated - would have tested risk appetite. That risk is deferred, not removed.
Medium term (fundamentals): The beneficiaries of a slower, privately-funded AI race are the same as the beneficiaries of a fast one: the chipmakers and cloud providers building the infrastructure. Nvidia's record buyback is the clearest signal - a company does not authorize $150 billion in repurchases if it expects the capital-spending cycle to stall. Microsoft, Amazon, and Google remain the funded customers of that buildout.
Long term (structural): This is where the cyclical-versus-structural call matters. If the safety pause is cyclical - a reaction to a string of rogue-agent incidents that fades as the incidents recede from memory - then OpenAI's IPO is merely deferred, and the valuation gravity toward $1 trillion remains. If it is structural - a permanent recalibration of how frontier labs are governed, with external review and capability pacing baked into the operating model - then the entire AI capital cycle reprices. Public markets would demand proof of safety as a condition of capital, not just proof of growth.
Base case: OpenAI lists in 2027 at a valuation between $1 trillion and $1.5 trillion, after Anthropic establishes the public-market benchmark. Upside case: a private round above $1.5 trillion and a successful Anthropic IPO pull OpenAI's debut forward and reset the multiple for the whole sector. Downside case: a serious AI safety incident forces a longer pause, and OpenAI raises private capital at a flat or down valuation - the SpaceX path.
The watchlist is concrete. First, Anthropic's IPO marketing in mid-October and its pricing - that is the first public-market read on frontier-AI risk appetite. Second, the terms of OpenAI's next private round, which will show whether safety is being priced in. Third, the text of the voluntary accord Trump announced - if it lacks external enforcement, the "self-policing" regime is political theater. And fourth, any S-1 filing from OpenAI, which would flip the restraint thesis on its head.
Trump's bet is that America can win the AI race by running faster than everyone else. Altman's bet is that the company that survives the race is the one willing to slow down. Both cannot be right about the same timeline - and the market will find out which one within 12 months.
Explore more exclusive insights at nextfin.ai.
