The Inverted Presumption
For three decades, the presumption ran one way. Capital was welcome unless there was specific reason to refuse it, and the burden fell on the state to justify saying no. That presumption has inverted. Across the advanced economies, foreign ownership of strategically significant assets now requires justification, and the burden has shifted onto the investor to explain why control should pass.
The inversion rests on a judgement about what markets settle and what they do not. Capital allocated by price finds its most productive use, and on that measure the open regime performed as advertised. It was never designed, however, to weigh who ends up holding a position of strategic consequence, because that is not a question returns can answer. For as long as ownership was assumed to carry no consequence beyond the balance sheet, the omission cost nothing. States have concluded it now does.
What they have rediscovered is what ownership functions as. A controlling stake is not simply a claim on returns; it is the position from which control over an asset is exercised, and capital is the instrument through which that position is acquired. Whatever the asset - and the forms are more various than the ports and terminals that dominate public argument - the leverage runs through the same point. Which raises a question worth putting directly: when a state screens a transaction, is it protecting the asset, or the position the asset confers?
What Screening Selects For
The turn is measurable, if narrow. In 2024, for the first time in six years, the Organisation for Economic Co-operation and Development (OECD) recorded a slight increase in average restrictiveness on foreign direct investment (FDI), reversing decades of liberalisation. The OECD attributes the rise mostly to domestic priorities rather than geoeconomic ones. What the aggregate index does not capture, the screening data does.
EU Member states notified 477 transactions to the European Union's screening cooperation mechanism in 2024, 15% more than in its first full year. Manufacturing accounted for half of the European Commission's in-depth reviews. Within those cases, the factors that pulled a transaction into deeper assessment were critical technologies (49%), critical infrastructure (26%), and critical inputs (20%), with sensitive data trailing at 9%. None of these statutes uses the word 'chokepoint'. The criteria describe one precisely: technologies, infrastructure, and inputs whose control confers leverage over everything that passes through them.
Screening also tracs where ownership has already gone. The states now scrutinising most closely are those whose firms built few positions beyond their own borders while the networks were assembled; the states whose investments they scrutinise built many. Two Chinese state-owned operators, answerable to the same agency, together handle 11% of global container throughput, and Gulf sovereign capital has concentrated on transport systems, maritime facilities, and supply-chain corridors across fragile states. The largest American terminal operator ranks tenth worldwide, at an eighth of the leader's volume, with a portfolio anchored in North America. Screening is what a state reaches for when it does not hold the position.
The effects are visible even when nothing is blocked. The International Monetary Fund found foreign direct investment increasingly concentrated among geopolitically aligned countries, particularly in strategic sectors, and Federal Reserve research shows American outward investment shifting away from China and Hong Kong towards Mexico, India, and Vietnam. Most notified transactions are cleared; the geography of investment is shifting regardless, which suggests screening deters more than its enforcement record shows.
Two Kinds of Chokepoint, One Instrument
Critical infrastructure is a place. A berth, a canal, a substation: goods pass through it physically, or they do not arrive. Critical technologies and inputs are positioned in a network. A process node, a data flow, a refining stage: production passes through them functionally, wherever the plant stands. Henry Farrell and Abraham Newman named the second kind in 2019, applying to networks a logic that straits and harbours had carried for a century.
The three Dispatches in this section take one case each. Daniel examines port ownership, contested berth by berth. Neville examines the divestiture of TikTok, an assetthat occupies no ground. Meylinda examines Mexico, where the subject is a screening regime and the flows that pass through it rather than any single transaction. The assets or scenarios have (almost) nothing in common. The instrument applied to them is the same, and it operates on the transfer of control rather than on the asset.
Not all these interventions were screening. A dormant statute, an act of Congress, a challenge to a concession: where ownership had already passed, states reached for whatever was to hand. The Dutch order was suspended within weeks, after talks with Beijing, and the American divestiture closed a year past its statutory deadline. Screening is what the same reflex looks like once it has a standing regime behind it, and it is the instrument to watch rather than the only one in play.
Screening adds a step to the older account. Farrell and Newman described states exercising leverage through nodes they already held. Screening reaches the prior question of who may acquire the node and answers it with a veto over the transaction. State ownership is not required; the power to withhold permission is sufficient.
Where the veto proves insufficient, states take the position. Washington holds a golden share in US Steel: a non-economic stake carrying consent rights over closures, relocations, and material acquisitions.
Sovereign capital arrives from the other side. Between 2015 and 2025 the United Arab Emirates deployed $449bn across 138 countries, concentrating on transport systems, maritime facilities, and supply-chain corridors in fragile states, and on semiconductors and cloud infrastructure in advanced economies. Dalia Aita describes the Emirati pattern as “an evolving mode of statecraft”. The instrument varies by jurisdiction; the object does not.
From Overture to Ownership
Four developments have been described here, but really they are one development. Restrictiveness turned upward in 2024 for the first time in six years. The criteria that trigger scrutiny select for technologies, infrastructure, and inputs whose control confers leverage. State-linked capital already holds the nodes that the screening states do not. And where the veto has proved insufficient, governments have taken the position directly.
What connects them is not the ports, or the networks, or the algorithms. Those are assets. The chokepoint is the ownership of them - the position from which passage is granted or refused - and it is the same position whatever the asset. A berth and a recommendation engine have nothing in common except that someone holds them, and whoever holds them decides who passes. Screening is how states took that decision back.
This leaves a question the statutes do not answer. If the case for scrutiny rests on the leverage that ownership confers, no asset is excluded in principle, because ownership confers leverage everywhere. The criteria mark out technologies, infrastructure, and inputs today. Nothing internal to the logic explains why they should stop there.
For three decades, states asked whether foreign investment was good for growth. They now ask, before anything else, whom it makes powerful.
