EU AI rules are becoming a boardroom risk for Europe’s own technology champions. Christophe Fouquet, ASML president and chief executive, warned in Antwerp on May 20 that companies like the European Union market but fear the amount and complexity of regulation, a complaint that matters because ASML sits at the choke point of the global chip boom.
His warning lands after seven European technology chiefs asked Brussels to simplify digital rules and weeks before the European Commission, the EU executive, presents its next sovereignty package. The sharper question is whether Europe can turn artificial intelligence (AI) into factory output while the rulebook, the compute buildout and the chip supply chain move at different speeds.
Fouquet Put the Bottleneck in Brussels
ASML, the Dutch lithography supplier whose machines help chipmakers print advanced circuits, rarely sounds like a normal corporate lobbyist. Its tools are used by Taiwan Semiconductor Manufacturing Company, Samsung Electronics, SK Hynix, Micron and Intel, which makes Fouquet’s complaint less like a narrow grievance and more like a warning from a toll booth everyone has to pass.
Demand on AI is coming so strongly that we will be in a supply-limited market for quite a while
Fouquet said that in the Reuters interview from Antwerp, where he linked AI demand with satellites, robots and new factory plans. The point for Brussels is simple: when supply is tight, the factory phase of AI rewards the places that can approve, test, build and buy fastest.
Europe has long argued that its rules export trust. That argument still has force. AI systems used for hiring, schooling, public services and border decisions can damage people at scale if they fail or discriminate. But industrial AI adds another clock. A factory automation model waiting on legal guidance, data permissions, chip allocation and power contracts is not waiting in a vacuum. It is losing a purchasing window to someone else.
That is why Fouquet’s choice of target matters. He did not only point at US export controls or chip supply. He put Brussels in the same sentence as execution risk, which is harder for the Commission to dismiss when its own competitiveness agenda says speed is now part of sovereignty.
Seven CEOs Turned One Complaint Into an Industrial Ask
Fouquet’s warning was not a solo act. On May 4, the heads of Airbus, ASML, Ericsson, Mistral AI, Nokia, SAP and Siemens published a joint statement after meeting Ursula von der Leyen, European Commission president. Their message had a harder edge than the usual call for innovation: Europe has the companies, patents and talent, but fails to turn them into scale.
The group is broad enough to matter. Guillaume Faury runs Airbus, the aerospace company. Arthur Mensch co-founded Mistral AI, the French AI developer. Christian Klein leads SAP, the German software group. Roland Busch leads Siemens, the industrial technology company. Together with Fouquet, Nokia’s Justin Hotard and Ericsson’s Börje Ekholm, they represent the industries Brussels says it wants to anchor at home.
| Signal | Number or Claim | Why It Matters |
|---|---|---|
| ASML warning | AI demand is running ahead of available chip supply across data centers, satellites and robotics. | Regulatory delay becomes a time cost when tools and wafers are scarce. |
| Joint CEO statement | The seven firms cited €417 billion in revenue, almost €1.1 trillion in market value, more than 957,000 high-tech jobs, more than €40 billion in annual research spending and 213,000 patents. | The complaint comes from Europe’s own industrial stack, not just US platforms. |
| Commission response | The Digital Omnibus political agreement sets some high-risk AI rules for 2 December 2027 and product-embedded rules for 2 August 2028. | Brussels is buying more time while trying to keep safety promises intact. |
The Numbers Favor Builders, Not Committees
ASML’s financials show why this debate has moved beyond white papers. In April, ASML’s latest quarterly results showed first-quarter net sales of €8.8 billion and net income of €2.8 billion. The company said demand for chips was outpacing supply and raised its full-year sales outlook to between €36 billion and €40 billion.
- €8.8 billion: ASML’s first-quarter net sales, a reminder that AI infrastructure demand is already showing up in the machinery layer.
- €200 billion: The Commission’s AI Continent Action Plan investment ambition, including €20 billion for up to five AI gigafactories.
- 20.0%: The share of EU enterprises with at least 10 employees using AI technologies in 2025, according to Eurostat’s enterprise AI data.
Those numbers do not point in one direction. ASML says demand is hot. The Commission says money is coming. Eurostat says most European companies still do not use AI in their business. That gap is where the current fight sits.
Brussels wants Europe to adopt AI in healthcare, cars, science, public services and manufacturing. Fouquet and the other chiefs are saying that the adoption target will not be met by grants alone. The firms that must embed AI into machines, networks, factories and maintenance systems need legal clarity before procurement teams can sign contracts at scale.
Brussels Is Buying Time, Not Quiet
The Commission has already moved. On May 7, it welcomed a political agreement between the European Parliament and the Council of the EU to simplify parts of the AI rulebook. Henna Virkkunen, European Commission executive vice-president for Tech Sovereignty, Security and Democracy, presented the deal as a way to keep safety while making implementation easier for business.
The agreement matters because it admits something companies have said for months: the first version of the timetable was running ahead of the tools needed to comply. The delay is a bet on standards, not an abandonment of regulation.
- Sensitive high-risk uses, including biometrics, critical infrastructure, education, employment, migration, asylum and border control, move to 2 December 2027.
- AI systems embedded in products such as lifts or toys move to 2 August 2028.
- Small and medium-sized firms, plus small mid-cap companies, get broader simplification benefits and more access to regulatory sandboxes.
That compromise will not end the fight. European lawmakers are tightening other digital obligations at the same time, including pressure on platform executives over child safety, a theme Mind Cron covered in its report on EU tech boss liability under the Digital Services Act. For industry, the signal is mixed: Brussels can slow one track while hardening another.
Chip Supply Turns Policy Drift Into a Cost
The constraint Fouquet knows best is physical. Lithography systems take time to build, advanced packaging capacity is stretched and memory suppliers are racing to feed AI servers. TSMC, the Taiwanese foundry group, Samsung, SK Hynix and Micron are all trying to expand while cloud buyers and chip designers fight for the same capacity.
That changes how policy delay feels. In a loose market, a European manufacturer can spend six months waiting for guidance and still find chips, power and suppliers. In a tight market, scarce chips punish late movers. A buyer that waits for a compliance answer may discover that the next delivery slot, the best engineering team or the cheapest power contract has already gone elsewhere.
The same squeeze is visible in energy. AI data centers are forcing power developers to move faster, especially in the United States, where Mind Cron has reported on the NextEra and Google data center power expansion. Europe’s AI strategy also depends on power, permitting, grid access and cooling, not only chips and code.
China adds another layer. Fouquet warned that tighter export restrictions on older deep ultraviolet lithography tools, often called DUV tools, could push China to build more of its own equipment. The Dutch government has objected to US pressure on ASML’s exports. For Europe, the trade-off is uncomfortable: export controls may slow China, but they also cut into one of Europe’s few world-leading hardware firms while Beijing spends to replace it.
Europe’s Factory AI Test Starts With Permits
The next milestone is not a speech. On May 19, the Commission opened feedback on draft high-risk AI classification guidelines, with comments due by 23 June 2026. Those guidelines are meant to help providers, deployers and market surveillance authorities decide whether a system falls into the high-risk bucket.
That sounds technical. For an industrial buyer, it is commercial. A company putting AI into a warehouse robot, a predictive maintenance tool, a hiring workflow or a medical screening process needs to know whether the system triggers conformity checks, documentation duties, human oversight demands and monitoring obligations. Until that answer is clear, the safest procurement option is often delay.
The political problem is that Europe wants speed and consent at the same time. It wants AI factories, gigafactories, trusted data spaces, startup access, industrial adoption and rights protection. All are defensible goals. Together, they require a state machine that can move faster than the market it regulates.
By putting regulatory complexity beside chip scarcity, Fouquet has forced Brussels to measure sovereignty in months, not slogans. Execution has become the test. If the answers arrive before the first big industrial AI deployments are ready, his warning will look like pressure that worked. If they do not, the companies building Europe’s machines will buy more of their intelligence somewhere else.
