Your AI Startup's Cap Table Could Be a National Security Problem

For most startups, flipping to a Delaware C-Corp is paperwork. Updates for a new parent entity, new stock ledger, but the same company underneath. For AI startups engaged in fundraising in 2026, the entity structure decision isn’t this simple. It's a decision that can trigger a national security review at the highest levels of the federal government.
‍

The money chasing AI is increasingly sovereign, meaning it comes from government-owned investment vehicles rather than private funds. GIC, Singapore's sovereign wealth fund, co-led Shield AI's $1.5 billion Series G. The Qatar Investment Authority backed Anthropic.

Sovereign funds can write checks no traditional venture firm can match. The size of that backing is a great opportunity. The source of it is potential exposure.


Why AI ventures get a different level of scrutiny
‍

The Committee on Foreign Investment in the United States (CFIUS) runs the national security review and foreign investment screening process for foreign acquisitions of US businesses. A foreign investor doesn't need to hold a majority stake to bring a deal under CFIUS review.

Things such as access to sensitive technical details, board seats or observer rights, involvement in substantive decision-making, and access to sensitive personal data are now being reviewed. A single sovereign-linked investor with a board seat in an AI funding round can be enough to bring the deal into CFIUS's jurisdiction, even when every other investor is a standard US fund.

Sovereign Wealth Funds are treated as government entities under CFIUS rules, which triggers a 45-day review followed by a 45-day investigation if concerns aren't resolved in the first window. CFIUS has blocked or conditioned several sovereign-fund-led AI investments in the past eighteen months, most often due to sensitive data access, governance rights, or exposure to critical technical details.

Treasury has also introduced a two-tier review process this year, featuring faster, more predictable reviews for "known investors," and a slower, higher-scrutiny review for investors CFIUS considers unfamiliar or higher-risk, including anyone with sovereign or state-linked ownership.


How this fits into your expansion plan
‍

This decision used to sit further down the roadmap, something to sort out once fundraising was already in motion. That's no longer how the timeline works. Which investors you can take and how fast a deal can close are now shaped by decisions that need to happen alongside product positioning and go-to-market planning.

Getting ahead of the following questions will help keep your structure from working against you later.
‍

Who is actually on your cap table, and where does their capital originate? This goes a layer deeper than the fund name to who has an ownership stake in the fund and whether any government moneys are invested.
‍

Does any investor hold governance rights that go beyond passive ownership? Board seats, observer rights, or access to technical or data systems can shift a deal into federal review even when the check size is modest.
‍

Is your data or technology sensitive enough to draw attention on its own? AI ventures working with personal data at scale, defense-adjacent applications, or foundational model infrastructure start every review with a higher baseline of scrutiny regardless of who's investing.
‍

Can your structure absorb a mitigation agreement if one gets requested? CFIUS rarely blocks deals outright. More often it will negotiate restrictions, limited governance rights, or ring-fenced data access as a condition of clearance. Structuring for that possibility in advance is faster and cheaper than negotiating it under deal pressure.

None of this is to say sovereign capital is off the table for an AI startup. It does mean how you enter the US market, how you establish your US entity, and how you structure ownership rights have to be carefully considered in this new light.