Trump Floats Federal Equity in AI as White House Adviser Krishnan Exits
On 6 June 2026, President Donald Trump told reporters the White House was weighing equity stakes in American AI companies, with OpenAI cited as the most likely first candidate. Within ninety minutes, two independent accounts carried a related story: Sriram Krishnan, the administration's AI policy adviser, would leave his post at the end of the month and was preparing a new institution to keep him close to the president's AI agenda from outside government. The two stories, taken together, point to a phase of US industrial policy in which the federal government is behaving less like a regulator and more like a portfolio manager.
The Trump administration came to office promising that America would "win" the AI race against China, and it has spent its first months tearing down guardrails its predecessor erected. The next phase looks different: an explicit willingness to fold the federal balance sheet into the private companies building frontier systems. That direction has been visible in pieces — the expansion of the CHIPS Act, the conditional federal support attached to data-centre buildouts, the courtship of Nvidia and TSMC — but the language Trump used on 6 June is the clearest articulation yet. The equity-stake frame reframes the public-private relationship as a co-investment, with the American taxpayer positioned as a residual claimant on AI upside.
Krishnan's exit and the institution question
Sriram Krishnan's reported departure was carried on 6 June by TechCrunch and corroborated within hours by the prediction-market account Polymarket and the markets-tracking account Unusual Whales. According to those accounts, Krishnan will leave his White House post at the end of June and is preparing to launch a new institution intended to keep him close to the president's AI thinking once he is technically back in the private sector.
Krishnan, a former venture partner at Andreessen Horowitz and a product leader at Microsoft, Twitter and Snap, joined the White House in 2024 as an adviser on AI policy. His tenure coincided with the release of the administration's AI Action Plan and the rollback of the Biden-era executive order on model evaluation and compute reporting. By all accounts he was a key architect of the administration's permissive approach to frontier-AI development, and the question of what he does next will be read closely inside the industry.
The "institution" framing is the part worth pausing on. A clean revolving-door exit — back to venture, back to a lab — would be unremarkable for a former Big Tech operator. Starting a new vehicle explicitly to maintain policy influence is a different signal. It implies Krishnan views the next two years as a window in which decisions made at the White House will determine which companies get to define the AI stack for the decade after, and that he wants to keep a hand on the tiller when he is no longer drawing a government salary.
The equity-stake doctrine
The other half of the day's news is the more unusual claim. President Trump told reporters on 6 June that his administration was considering taking equity stakes in American AI companies, and pointed specifically to OpenAI as a likely target. The comment, carried by Reuters and reposted by the Unusual Whales account, used unusually direct language: the goal, Trump said, was to ensure "the American people can benefit from the success of AI."
That phrasing does several things at once. It positions frontier AI as a national resource rather than a purely private commodity. It raises — without answering — the question of what federal equity would actually buy: preferential compute access, export-control cooperation, locational commitments, or simply a financial claim on returns. And it sets up an obvious tension with the existing federal posture. The United States has historically regulated the technology sector through antitrust and disclosure rules, not through portfolio ownership.
The OpenAI connection is the most politically charged element. OpenAI remains a capped-profit company with a non-profit parent; its governance has been a source of friction since the 2023 board restructuring, and the company has spent most of the past year signing large cloud commitments with Microsoft, Oracle and others. A federal stake would land in the middle of an already-complicated capital structure, and it would put the US government in the unusual position of holding a financial interest in a specific frontier lab while supposedly regulating the field as a whole.
The structural frame
Read together, the two stories describe a coherent industrial-policy posture. The United States is treating frontier AI the way it treated the railroads in the nineteenth century, the oil sector in the twentieth, and semiconductors in the early 2020s: a strategic industry in which the federal government is willing to take an ownership position in exchange for the right to set the rules of the road.
The historical comparison is imperfect. The rail and oil eras pre-dated modern antitrust law, and the equity components of the CHIPS Act remain legally distinct from a direct federal stake in a private AI lab. But the direction of travel is the same. Washington is moving from a procurement-and-subsidy model to a co-investment model, and it is doing so at the precise moment that frontier AI is being framed, in official discourse, as a matter of national security on par with nuclear weapons or semiconductor fabs.
This is also where the China variable enters directly. The administration's stated reason for both moves is to keep the United States ahead of Chinese AI development. If federal equity is the price of pulling more compute onto American soil, and if staying ahead of Beijing requires Washington to behave more like the Chinese state in its industrial-policy toolkit, the policy logic of the past four decades begins to invert.
That inversion is the most analytically interesting part of the story, and the part least discussed in the early coverage. The dominant framing — a free-market White House finally getting serious about Beijing — is accurate as far as it goes, but it leaves out the structural shift in the relationship between Washington and Silicon Valley. Federal equity in OpenAI would not be a return to the New Deal. It would be a new arrangement, in which the government positions itself as a counterparty to the labs it is supposed to police.
Counterpoint and the unknowns
Two counter-reads are worth naming before drawing a conclusion. The first is that the equity talk is positioning, not policy. Presidents have floated direct stakes in private companies before — most recently in the airline and bank rescues of 2008-09 and the auto bailouts of 2009 — and many of those proposals did not survive contact with the Justice Department's antitrust division or the Government Accountability Office. Federal equity in a private technology company raises legal questions the administration has not answered: whether such a stake would be lawful under existing appropriations law, how it would interact with the Federal Acquisition Regulation, and what disclosure obligations would attach to a federal position in a single, large private firm.
The second counter-read is that the China framing is doing more rhetorical work than is warranted. The argument that the United States must out-build China on AI to preserve national security is now ubiquitous in Washington, but the operational meaning is contested. Some in the administration define the gap as a compute gap; others as a model-capability gap; others as a deployment gap. Federal equity in OpenAI would not by itself close any of them, and it would not change the export-control regime that remains the binding constraint on Chinese access to advanced compute. The equity-stake language may be a way of signalling seriousness without specifying a strategy.
What remains unknown is whether Krishnan's new institution will be a think tank, a lobbying vehicle, a venture fund, or a hybrid. The sources do not specify. They also do not specify which other AI firms the administration has approached, whether the federal stake would sit in OpenAI's for-profit subsidiary or in the non-profit parent, or whether Treasury or Commerce would hold the position. The day's news opens more questions than it closes, and the answers will determine whether 6 June 2026 is remembered as the date US AI policy turned, or as a day on which it merely drifted a little further from where it had been heading.
Monexus reads the two stories as a single signal — Washington moving from subsidy to co-investment in frontier AI, with the institutional scaffolding being built to make that shift durable — whereas the wire coverage has so far treated the personnel change and the OpenAI question as separate beats.
Wire provenance
This editorial synthesis draws on the following public wire/social posts:
- https://en.wikipedia.org/wiki/Sriram_Krishnan
- https://en.wikipedia.org/wiki/OpenAI
- https://en.wikipedia.org/wiki/Office_of_Science_and_Technology_Policy
- https://en.wikipedia.org/wiki/CHIPS_and_Science_Act
- https://en.wikipedia.org/wiki/Executive_Order_14110
- https://en.wikipedia.org/wiki/Andreessen_Horowitz