Sriram Krishnan is leaving the White House AI job to build an outside policy institution

Sriram Krishnan is leaving the White House. The senior policy advisor for AI told officials he’ll step down at the end of June 2026, TechCrunch first reported, after 18 months turning Silicon Valley’s AI agenda into federal policy. His next plan: an outside institution to keep steering it.

That matters because Krishnan wasn’t a background staffer. He co-wrote the document that defines the Trump administration’s whole approach to AI, and he ran the talks that decide which chips ship to which countries. His exit pulls one of the two or three most consequential AI-policy operators out of government at the exact moment the rules he helped write are being tested. If you build, sell, or staff AI systems in the US, the open question is whether his replacement keeps the same line or quietly redraws it.

Who Sriram Krishnan is

Before the White House, Krishnan spent two decades as a product executive. He started at Microsoft in 2007 on Visual Studio, built the Facebook Audience Network at Meta, led product at Twitter, and did stints at Snap and Yahoo, per his Wikipedia profile. Then in February 2021 he joined Andreessen Horowitz as a general partner, running the firm’s London office and co-hosting a popular tech podcast with his wife.

He left a16z in late November 2024. Weeks later, on December 22, 2024, then-president-elect Trump named him senior policy advisor for AI. He started January 20, 2025. The appointment was a clean example of the venture-capital-to-government pipeline that has defined this administration’s tech bench, and he reported into David Sacks, the former PayPal executive serving as the White House AI and crypto czar.

In his farewell statement, Krishnan didn’t hide what the job meant to him. “It is hard to express how big a privilege it has been to serve the American people and how grateful I am to have had the opportunity to do so,” he wrote. That’s not a man leaving in frustration. It reads as someone who got what he came for and wants a different lever.

What the role actually covered

The title was vague. The portfolio wasn’t. Krishnan’s brief ran across four pieces of the AI agenda, and each one is live right now.

First, the AI Action Plan.

The AI Action Plan is the administration’s July 2025 strategy document, co-authored by Krishnan, Sacks, and Michael Kratsios. It prioritizes building data centers and spreading US AI models abroad over writing safety rules at home. Krishnan was one of three credited authors, so the deregulatory, build-first posture is partly his design.

Second, export controls. Export controls are the federal rules that limit which advanced technologies, like high-end AI chips, US companies can sell to which foreign buyers. Krishnan ran a lot of the diplomacy around them, including the politically charged question of which Nvidia accelerators can go to China and which AI deals get cut with Gulf states like Saudi Arabia. He framed the stakes bluntly, calling it “an existential race with China” measured in market share, according to his Wikipedia profile.

Third, immigration. Krishnan was a vocal advocate for pulling high-skilled AI talent into the US, a position that put him crosswise with parts of the administration’s base over H-1B visas. And fourth, the broader policy coordination job: keeping the executive orders, the model-neutrality rules, and the council of science advisors pointed in the same direction.

All of this sat inside the OSTP, the White House office that advises the president on science and technology and houses the federal AI policy shop.

Why the departure matters

A staff change isn’t usually news. This one is, for two reasons.

The timing is the first. Krishnan is leaving days after the administration narrowed its own AI oversight order. We covered that move in our piece on the scaled-back AI executive order: the mandatory model-review requirement got cut after Silicon Valley pushed back. Krishnan helped shape the lighter-touch posture that produced that retreat. Losing the architect right after the architecture shifts leaves the policy without its main internal defender.

The second reason is concentration. With Krishnan out, more of the AI brief funnels back to Sacks, who already wears the crypto hat too. White House spokesman Kush Desai called Krishnan’s role “vital” and said the office looks forward to working with him “in his new venture,” The Washington Post reported. Praise on the way out is standard. What’s missing is a named replacement, and that gap is the actual story.

There’s also a revolving-door angle worth naming plainly. Krishnan came from a16z, wrote rules that benefit the AI build-out his old firm invests in, and is now leaving to run an “outside institution” that, by his own description, will keep influencing tech policy. That’s legal and common in Washington. It’s also exactly the pattern that critics of the VC-to-government pipeline point to, and reporters will be watching where his funding comes from.

Who might replace him, and what it signals

No name has surfaced yet. The likeliest outcomes split three ways.

Sacks absorbs the portfolio directly, which centralizes AI and crypto policy under one person and tilts further toward the industry-friendly line. Or Kratsios, the third AI Action Plan co-author and a returning Trump tech hand, takes a larger share. Or the administration recruits another operator from the venture or frontier-lab world, keeping the talent pipeline intact but resetting the relationships Krishnan spent 18 months building.

Whichever way it breaks, the policy text doesn’t change overnight. The AI Action Plan stands. The export-control framework stands. What changes is execution: who picks up the phone on a chip deal, who decides whether the next executive order tightens or loosens, and whether the immigration push survives without its loudest internal champion. For an industry that just spent a year learning to lobby this specific office, relationships reset to zero.

Why you’re hearing about this now

The trigger is Krishnan’s own statement, posted June 6 and picked up by TechCrunch, The Washington Post, and The Information within hours. It lands in a stretch where the government’s posture toward AI is unusually fluid: the executive order just got narrowed, federal money is moving into strategic tech through unusual structures like the equity stakes Commerce is taking in quantum companies, and the line between national-security policy and commercial advantage keeps blurring, the same blur we traced in how adversaries buy US troop-location data. Krishnan sat at the center of that blur. His seat is now empty.

What this means for you

If your work touches AI policy, here’s the practical read. Don’t expect the rules to flip. The deregulatory direction Krishnan helped set is durable, baked into the AI Action Plan and the narrowed oversight order, and it won’t reverse because one advisor left. What you should expect is a relationship reset and a few months of uncertainty about who actually decides.

If you’re at a frontier lab or a chipmaker, your export-control contact may change; line up the next one early. If you’re hiring AI talent from abroad, the immigration tailwind just lost its strongest backer inside the building, so plan for it to stall rather than accelerate. And if you’re tracking where US AI policy heads next, watch Krishnan’s new institution. A former VC building an outside body to influence the same policy he just wrote is a clearer signal of direction than any press release the White House will put out about his replacement.