Friend-Shoring's Fiction: A Chinese Supply Chain in an American Coat

The Inflation Reduction Act of 2022 was the largest industrial-policy bet in modern American history, and its critical-minerals provisions had an unambiguous aim: rebuild a domestic and allied battery and clean-technology chain and end China's leverage over the inputs. Three years on the subsidies have certainly moved capital. Whether they moved the dependence is a different question, and it turns on a distinction the policy never quite confronts. A subsidy can buy the geography of value-added, where the last step is performed, but it cannot buy the origin of capability, the accumulated ability to perform the hard steps at all. The trouble is that the last step, cell assembly, is the cheapest part of the chain to relocate, while the capability that constitutes the dependence lives upstream, in refining, precursor and cathode. A rule that rewards assembly location is aimed at the one link that was never the problem.

A rule about ownership cannot see a problem of capability

The exclusion mechanism is the Foreign Entity of Concern rule, finalised in May 2024: from 2024 a battery component made by a FEOC voids the consumer credit, from 2025 the same for critical minerals. Its hinge is a bright line, an entity is a FEOC if a covered-nation government holds 25 percent or more of equity, board seats or votes. A bright-line ownership test invites the obvious response, which is to restructure the ownership. A Chinese majority owner dilutes below 25 percent, or recasts itself as a technology licensor rather than a shareholder, and the output qualifies. The rule measures who owns the plant and where the plant sits. It does not, and cannot, measure where the capability came from, which is the only variable that determines dependence.

The arbitrage is not hypothetical; firms describe it on the record. In Morocco, which has a US free-trade agreement, CNGR and a Moroccan fund are building precursor capacity and LG Chem and Huayou are building cathode and lithium plants for the North American market, with LG stating openly that it will adjust ownership shares as needed to stay under the FEOC threshold. Onshore, Ford's plant in Marshall, Michigan, licenses lithium-iron-phosphate technology from CATL of China while Ford owns the building, a structure that survived after Congress stripped a proposed ban on credits for Chinese-licensed batteries. In both cases Chinese capability re-enters the subsidised chain wearing a permitted coat, because the rule was written to inspect the coat.

The allied chain runs on Chinese intermediates

Underneath the assembly plants, the midstream the subsidy barely reaches is where Chinese dominance is near-total. China refines well over 90 percent of battery-grade graphite and anode material, holds about 87 percent of cathode-active-material capacity, makes roughly 96 percent of precursor and 98 percent of lithium-iron-phosphate cathode, and processes around 60 percent of the world's lithium and cobalt. A gigafactory in Georgia or Tennessee performs the last step on inputs that are overwhelmingly Chinese in origin even when they arrive through a third country.

South Korea, the friend-shoring partner held up as the template, makes the point against itself. Chinese suppliers provide about 96.6 percent of Korea's precursor cathode material, 93.7 percent of its synthetic graphite and 80.4 percent of its lithium hydroxide. Korean cell makers are roughly a fifth of global output and sit squarely downstream of Chinese chemistry. Routing the chain through Seoul satisfies an allied-sourcing rule on paper while leaving the Chinese midstream it was meant to replace entirely intact.

A subsidy can certify where the last step happens. It cannot certify where the capability came from, and the dependence lives in the capability.

Subsidising the visible step can widen the invisible gap

There is a second-order effect that makes the policy, in places, worse than merely ineffective. Subsidies aimed at the visible final step inflate that step's capacity, cells and packs, while doing little for the invisible upstream, so the ratio of US assembly capacity to US input capacity can widen, deepening reliance on Chinese intermediates in absolute terms even as the ribbon-cuttings multiply. And where Washington has blocked Chinese material head-on, the result has been cost, not supply: the Section 301 graphite tariff was finalised in 2024 but delayed to 2026 because no domestic substitute exists, and stacked anti-dumping and countervailing duties pushed effective rates on Chinese anode material well above 100 percent. The leverage, meanwhile, ran the other way when China banned gallium, germanium and antimony to the United States in December 2024, sending antimony from the low tens of thousands of dollars a tonne toward 60,000 and reminding Washington which end of the chain it actually held.

The price floor is the confession

The clearest admission is the flagship rare-earth deal. When the Department of Defense took roughly a 15 percent stake in MP Materials in July 2025 and guaranteed a floor of 110 dollars a kilogram for its main oxide, about double the market, for ten years, it was not standing up a competitive industry; it was guaranteeing a price because the market will not clear at the level China sets. MP's near-term magnet target is around a thousand tonnes, under one percent of Chinese output, and China still controls roughly 90 percent of rare-earth separation. The honest counter-case is Lynas, which produced heavy rare earths outside China for the first time in 2025, at its plant in Malaysia, not the United States, while its Texas project remains uncertain. The one genuine break in the monopoly is happening abroad, which only underlines that America is subsidising demand, not building the chokepoint.

What would change this view

The claim is falsifiable, and the disconfirming evidence is specific. It would be wrong if US and allied midstream capacity, precursor, cathode, anode and magnets, reached a meaningful share of cell capacity at unsubsidised cost; if a non-China anode or magnet supplier cleared automotive volume without a government price floor; or if Korea's roughly 96 percent precursor dependence on China fell materially rather than holding. The leading indicators have not turned. Capacity is rising at the assembly step and crawling at the steps that matter, which is the pattern the model predicts.

What it reveals

Friend-shoring is not a fiction because nothing is built; the gigafactories are real. It is a fiction because what is built is the part that was never scarce. The credits certify final assembly in a friendly place while refining, precursor, cathode and magnet, the steps where China's advantage actually lives, stay Chinese or stay unbuilt. The policy moved where value is booked and where ribbons are cut. It has not yet moved the capability, and the capability is the whole game. Read against the China note in this series, it is the same lesson from the buyer's side: whoever owns the step with the highest capability barrier owns the chain, and a subsidy aimed at the last step cannot buy its way past that.