AI IPO valuations face fresh risk after Trump's order retreat

Donald Trump’s decision to kill a draft AI executive order has reopened a question bankers and late-stage investors had started to narrow: how much regulatory discount should apply to the richest AI companies as they move toward public markets. For OpenAI and Anthropic, whose latest private marks sit at about $852 billion and a targeted $900 billion respectively, the vanished policy path matters less because it would have created a 90-day review window than because it removes one of the few visible federal frames investors could model.

Politico’s publication of the unsigned order showed how narrow the planned oversight channel really was. Frontier-model developers could have volunteered systems for up to 90 days of federal testing, while the text explicitly ruled out mandatory licensing or pre-clearance. Trump later told The Hill that he killed the plan because he disliked parts of it and thought it could have been a blocker. Markets now have to work with a messier signal: the White House has shown it can retreat even from soft-touch oversight, but it has not supplied a durable alternative.

From an analyst’s vantage, the pricing problem starts there. An academic study of IPO pricing and policy uncertainty found that political alignment and local policy uncertainty can affect how deals are priced, with an average discount of 5.39 per cent and $1.58 million left on the table for each one-standard-deviation increase in alignment. Frontier AI is its own asset class, not a copy of the paper’s sample. Even so, the lesson transfers cleanly enough. When buyers cannot map the policy path, they ask for room. Yes, the rollback may remove one near-term compliance cost, yet it also strips out a draft rulebook that could have narrowed the spread in valuation assumptions before roadshows begin.

Trump made the market case for abandoning the order in blunt terms when he described it as something that might have stood in the sector’s way.

“I really thought it could’ve been a blocker.”

— Donald Trump, via The Hill

From the policy side, though, the abandoned order was never obviously pro-competition oversight. Its own text said the review window could not be used to create “a mandatory governmental licensing, preclearance, or permitting requirement.” In practice, a voluntary testing regime often helps the biggest labs most, because they can absorb red-team cycles, lawyers and Washington process in a way smaller rivals cannot. So the policy counterpoint is not that Trump walked away from a robust brake on AI. It is that Washington nearly created a light-touch moat, then withdrew even that half-formed structure.

“Nothing in this section shall be construed to authorize the creation of a mandatory governmental licensing, preclearance, or permitting requirement.”

— Draft executive order, via Politico

By cancelling the order, the administration also hands the oversight narrative back to the frontier labs themselves. That may suit executives who prefer self-attestation to even voluntary federal testing. Investors, meanwhile, are left relying more heavily on company-defined safety language just as deal sizes are getting larger.

Why public markets may care more than private rounds

Public markets feel the sharpest valuation consequences because a company has to clear outside money every morning. OpenAI’s reported plan to file confidentially for an IPO and Anthropic’s revenue surge and fundraising target ask future buyers to underwrite explosive growth while margin durability, compute intensity and safety governance are still being argued in real time. That burden was already heavy. A visible White House retreat from even voluntary model reviews adds another layer of uncertainty around what investors should demand in disclosures.

Those numbers explain why the policy shift matters. OpenAI’s latest reported valuation of $852 billion and Anthropic’s $900 billion fundraising target are not ordinary software marks; they assume the eventual public market will accept extraordinary growth, tolerate huge compute bills and still grant strategic scarcity a premium. The missing White House framework does not break that thesis on its own. It does make the journey from private mark to public clearing price more fragile.

Inside the deal room, the picture is less dramatic than the political debate suggests. Bankers and founders are unlikely to halt filing calendars because a draft order died. The more likely response is over-preparation: fatter risk sections, harder questions on safety testing, more detail on compute concentration and a sharper focus on who controls board decisions when capital needs keep climbing. Public investors can tolerate aggressive growth stories. They tend to punish governance ambiguity faster than private backers do, especially when the valuation on offer already assumes category dominance.

For investors, that is why the rollback may enlarge the spread between private marks and public clearing prices rather than close it. A late-stage financing can still be sold on scarcity, access and the belief that regulation will stay permissive long enough for scale to arrive. An IPO cannot rely on patience in the same way. The market gets to ask, day after day, whether a company priced on trillion-dollar rhetoric also has a framework for safety incidents, policy reversals and the cost of staying ahead in model training.

A recent CNBC analysis of cheaper AI competition already argued that falling model prices and pressure on margins could complicate the listing case for both OpenAI and Anthropic. Washington’s reversal does not replace that concern. It sits on top of it. A company can grow at extraordinary speed and still see buyers widen the discount if the future rulebook looks contingent on internal White House politics rather than a stable federal approach.

Why softer oversight can still weigh on valuations

At first glance, less oversight should mean higher multiples. For some private holders, that will be the first instinct. A looser federal stance lowers the odds of an immediate compliance drag and gives management teams more freedom to frame safety as self-governance. It also leaves the hardest due-diligence questions unanswered, which is where public-market scepticism tends to harden.

Underwriters and portfolio managers are now more likely to push on the details a draft federal channel would at least have organised: how model evaluations are described, when incidents are disclosed, how dependent a lab is on a narrow set of compute suppliers and how much spending must stay elevated to preserve a lead. None of those questions vanish because the order did. They simply move back into prospectus language, non-deal roadshows and discount-rate arguments.

Seen that way, a voluntary review regime could have been both a regulatory moat and an investor comfort blanket. Smaller challengers would have struggled with the process costs. Large incumbents, by contrast, could have showcased participation as a badge of seriousness on the roadshow circuit. Scrapping the regime removes that potential moat, but it also removes a talking point that bankers could have used to argue the biggest labs were operating within a recognised federal channel. One source of friction disappears. So does one source of reassurance.

Beyond the next filing, the pattern is familiar. The AI listing wave has already drifted into a governance debate, not just a revenue debate, as investors weigh complex ownership structures, safety claims and the possibility that policy intervention arrives after private marks have surged but before public expectations settle. The result is a familiar late-cycle pattern: private capital rewards speed, while public capital asks whether the risk factors are finally catching up with the story.

For Washington, the retreat may look like a tactical win for the industry’s growth camp. For markets, it makes regulation a live valuation variable again. If OpenAI, Anthropic and the rest of the AI IPO queue come to market under a softer but less predictable federal stance, the discount is unlikely to arrive as a headline penalty. It will show up in quieter places: lower conversion from private mark to IPO price, deeper demands for governance disclosure, and a wider gap between what insiders think these companies are worth and what outside investors will actually pay.