CFIUS and the American Screening Machine

The Committee on Foreign Investment in the United States is one of the most quietly powerful bodies in economic statecraft. An interagency committee chaired by the Treasury and populated by defence, state, commerce, energy and intelligence officials, it reviews foreign acquisitions of American businesses for national-security risk and can recommend that the President block or unwind them. For most of its history it operated in obscurity, waved through most deals and blocked only a handful. That era is over. CFIUS has become a central instrument in the contest over who may own the physical and corporate substrate of artificial intelligence.

The turning point was the Foreign Investment Risk Review Modernization Act of 2018, known as FIRRMA. It widened the committee's reach in ways precisely tuned to the technology contest. Where CFIUS had historically focused on deals that gave a foreigner control of an American firm, FIRRMA extended jurisdiction to certain non-controlling minority investments in businesses dealing with critical technologies, critical infrastructure and sensitive personal data, the trio often abbreviated as TID. It also made filings mandatory in defined categories rather than purely voluntary, and it lengthened review timelines. The intent was to catch the subtle acquisitions, the minority stake that buys a board seat and a window into a sensitive laboratory, that the old regime would have missed.

For the AI economy this expansion is decisive. Semiconductor firms, chip-design houses, quantum and advanced-computing startups, firms holding vast troves of Americans' personal data, and increasingly the operators of critical infrastructure such as ports, telecoms and energy all fall within the widened net. When a foreign, and especially a Chinese, investor seeks a stake in such a company, the deal now runs a gauntlet. The most cited early signal was the blocking of attempts by Chinese-linked capital to acquire American chipmakers and, famously, the scuttling of a large semiconductor acquisition on national-security grounds. The message to Chinese capital was that the door to the American technology base was closing.

The critical-minerals dimension has grown in parallel. As Washington awoke to its dependence on Chinese mineral processing, CFIUS and related authorities began scrutinising foreign investment in domestic mining and refining projects, wary of an adversary buying up the very capacity the United States was trying to rebuild. Data centres, too, have drawn attention, because the physical siting of compute near sensitive facilities, and the ownership of the entities that run it, raise questions about surveillance, sabotage and control. The committee's logic has expanded from protecting discrete technologies to protecting the whole ecosystem that produces and houses compute.

CFIUS also acquired a forward-looking cousin. In 2023 the United States moved to establish an outbound investment screening regime, an executive order restricting American capital from flowing into Chinese firms in semiconductors, quantum and artificial intelligence. This was a genuine novelty. For decades the assumption was that inbound investment might threaten security but outbound investment was simply commerce. The new logic holds that American money, expertise and prestige flowing into an adversary's advanced-technology sector can itself accelerate a rival's capabilities. Screening thus became bidirectional: guarding not only who buys into America but where Americans may invest abroad.

Faraday's read is that CFIUS embodies the ownership dimension of the compute contest. Export controls and CFIUS are complementary halves of a single strategy. Controls prevent the adversary from buying the technology outright; screening prevents the adversary from buying the companies, the assets and the data that would deliver the technology by another route. A rival blocked from purchasing advanced chips might instead try to purchase the firm that designs them, or the mine that feeds them, or the data centre that runs them. Screening closes that flank. The two tools together define a perimeter around the American compute base that is far harder to breach than either alone.

The cost is not zero, and honest analysis must acknowledge it. A screening regime that is too aggressive or too unpredictable can deter benign foreign capital, raise the cost of financing for American startups, and invite retaliation against American investors abroad. Allied investors chafe at being treated with the same suspicion as adversaries, which is why the regime has carved out streamlined treatment for trusted partners. The art of the system lies in calibration: tight enough to keep the adversary out, open enough to keep allied and neutral capital flowing in. Get it wrong in the restrictive direction and you starve your own innovation base. Get it wrong in the permissive direction and you hand the adversary a side door. CFIUS is where that balance is struck, deal by deal, largely out of public view.