€8.7 Billion Promised to Chip Plants Since 2018 — Almost €5 Billion Paid With No Strings Attached

Seven years, eight point seven billion euros of public support, and not a single binding performance obligation imposed on the manufacturers. This is the finding of the Court of Auditors in a report published on 21 April 2026 about the public support to France’s semiconductor sector. Public support to the microelectronics supply chain, largely made up of subsidies paid by the State, amounts to 8.7 billion euros of programmed assistance, including 5 billion euros disbursed during 2018-2025, without any comprehensive consolidation ever carried out until this report. Of this amount already paid out, 5 billion was disbursed during that period, of which 4.3 billion by the State. The remainder? European funds and local authorities, added over time without any visible coordination.

A maze of funding windows rather than a strategy

Public money comes from everywhere, and that is the core problem. The State shoulders the bulk, with 7.7 billion euros programmed, supplemented by European funds (715 million euros) and regional/local authorities (219.5 million euros). In addition to these subsidies, there is another envelope, often overlooked in debates: 3.6 billion euros more invested directly in company equity, primarily by Bpifrance, the State’s public investment arm. Two instruments structure the whole: the European IPCEI microelectronics projects, rolled out in two successive waves, and the France 2030 plan, meant to coordinate the national industrial strategy since 2021.

However, the promised centralization never truly occurred. Despite the consolidation of steering under the General Secretariat for Investment as part of France 2030, the budgetary architecture remains complex, and the State lacks a consolidated view of aid paid by local authorities and by the European Union. As a result, until this report, no one had ever added up all these aids to draw an overall balance sheet; money flowed without anyone really keeping the books. Two players concentrate the bulk of financing: STMicroelectronics and the CEA-Leti, the Grenoble microelectronics research laboratory.

Crolles, a showcase of ambitions and fragilities

The Crolles site in Isère embodies the paradoxes of this policy on its own. The Liberty project, launched in 2022, brings the French-Italian foundry together with American GlobalFoundries to extend the production capacity at the historic Crolles site, near Grenoble, with a facility capable of producing 300 mm wafers. The stated ambition is clear: to double France’s chip production capacity by 2028, for a total investment announced at 7.5 billion euros. France itself is putting on the table 2.9 billion, i.e., 1.8 billion for GlobalFoundries and 1.1 billion for STMicroelectronics.

The snag is that commitments do not keep pace with disbursements. By the end of 2025, only 574 million euros had been paid to STMicroelectronics, and nothing at all to the American foundry, which has not yet invested in the project. Public money flows from one side while the industrial commitment purported to justify it remains, for now, on the other side: a promise. The Court of Auditors does not miss this: it cites the Liberty project, pairing ST and GlobalFoundries, as a key example illustrating both the ambitions and the fragilities of the public-aid framework.

Cheques with no repayment conditions

Here lies the core problem highlighted by the institution on Rue Cambon. In France, the money goes out, but nothing legally obliges beneficiaries to keep their promises. These aids show a low level of conditionality: unlike practices in the United States or Japan, contracts contain few requirements regarding domestic production or jobs created. Worse, the standard budgetary safeguards that exist in other sectors are barely present in this sector: aid rarely takes the form of repayable advances, unlike in other fields, and clawback clauses—provisions that would allow the State to recover its investment in case of failure—were only recently introduced, under European impetus, for beneficiaries of more than 50 million euros.

Concretely, if a company receives hundreds of millions of euros and never delivers the promised plant, or later fires a large number of workers, the State has almost no legal lever to compel repayment. The Court goes further on employment tracking: the impact on employment is not monitored in a consolidated way, despite a sector employing roughly 53,600 people. And on industrial sovereignty—argued as the primary justification for these billions—there is a similarly troubling gap: the objective of industrial sovereignty remains without a true measurement instrument, and no indicator shows what share of French needs is covered by domestic production, nor whether capacities are advancing in strategic segments.

Real results, but impossible to link them to the subsidies

Not everything is bleak in this picture. The raw figures for the sector are progressing, and fairly quickly. The turnover of the French sector rose by 46% between 2018 and 2024, compared with 35% for the global industry. But it is impossible to determine what portion of this growth is truly attributable to the billions of public injections, and what portion would have occurred anyway due to global demand for chips. This is precisely what the Court criticizes: financing without a clear capacity to measure its impact.

The European context does not help to downplay urgency. The European Chips Act, which came into effect in September 2023, provides 43 billion euros in public investments to attempt to raise Europe’s share of world semiconductor production, which today remains marginal compared with Asia and the United States. The Court of Auditors offered seven recommendations to tighten the screws: precise mapping of needs by chip type, stronger conditionalities, and a single interministerial steering body. It remains to be seen whether the ministries of Finance (Bercy) and the Crolles industry players, used for eight years to an open grant window with no counterparty, will agree to change the rules of the game mid‑play.

Sources: program-evaluation.ccomptes.fr | assemblee-nationale.fr | data.consilium.europa.eu