Demand has outrun the factory
On 14 October 2026 ASML reported third-quarter net sales of €11.7bn and said it expects €44–45bn for the full year, with Q4 sales guided to €14.2–15.2bn. It sold more than 120 lithography machines in the quarter, against just over 70 a year earlier (Silicon Republic). CEO Christophe Fouquet said supply as currently planned for 2027 and 2028 (a 30% capacity increase in 2027, with a further 30% being studied for 2028) will not fully meet demand. Some customers are already trying to place orders for 2029 and sometimes 2030, driven by their views on the size and duration of AI-related investment.
That last point matters for policy. Export controls are usually discussed as a security filter: which tools may go where. When a single company is the only maker of the most advanced lithography equipment and its order book runs years ahead of supply, every licence decision also affects who gets scarce capacity and when. The Dutch government did not design its regime for that role, but it now plays it.
What the Dutch regime does
The Netherlands has been tightening its own national controls in steps. Minister Reinette Schreinemacher told parliament of planned additional measures in a Kamerbrief dated 8 March 2023. A national authorisation requirement for advanced deep ultraviolet lithography took effect on 1 September 2023. On 7 September 2024 it was widened to more equipment types, through a ministerial order published in the Government Gazette. Then, from 1 April 2025, it was extended again to a limited set of other manufacturing-step technologies, including certain measuring and inspection equipment.
Two features of this design are worth noting. Licences apply to exports outside the EU and are assessed case by case; the government stresses this is a licensing regime, not a ban. And the stated rationale is security: Minister Reinette Klever said the Netherlands has "a unique, leading position" and cited "increased security risks associated with the uncontrolled export of this specific equipment."
The strongest case for tight control
The case for restraint is serious and should be stated fairly. These tools underpin both frontier AI chips and advanced military electronics. A supplier with no substitute has unusual leverage, and a country that holds it can reasonably decide that some end-users should not receive it. Case-by-case review also lets officials respond to new intelligence without rewriting the law. The government's own justification, that technologies from several manufacturing steps can be combined with other countries' technology to make advanced chips for military use, is a coherent reading of how the supply chain works.
Nothing in ASML's results undermines that logic. If anything, scarcity strengthens the argument that the highest-capability tools should not flow to actors that pose security risks.
Where proportionality comes in
The problem is not that the Netherlands controls these tools. It is that scarcity changes what an opaque licence process does. When supply is slack, a slow or unpredictable licence mostly delays a sale. When customers are queuing for 2029 and 2030 slots, delay and uncertainty decide who is served, and they do so without the public criteria a capacity-allocation decision would normally require. Three consequences follow.
- Investment planning suffers. Chipmakers commit multi-billion-euro fabs against tool delivery schedules. Licence timing that cannot be forecast raises the cost of capital for exactly the AI buildout the Dutch economy benefits from.
- Security and allocation get mixed up. If a licence is slow because of commercial queueing rather than a security finding, the public cannot tell the difference. That erodes trust in the legitimate security function.
- Pressure to widen controls grows. Tight supply makes it tempting to treat every unfilled order as a reason to restrict more, when the policy question should stay narrow: does this specific end-use create a security risk?
The government itself says it wants to avoid unnecessarily disrupting the chip industry and has kept expansions narrow. That is the right instinct. It needs institutional form.
What a proportionate response looks like
First, publish decision criteria and typical processing times. Denial and delay reasons can stay confidential case by case, but the categories of concern should be public. Second, keep security review separate from supply allocation. Allocation is ASML's commercial job; the state should not become a hidden queue-manager. Third, report aggregate licensing statistics to the Tweede Kamer, such as the number of applications, approvals, conditions and denials, so parliament can test whether the system is proportionate. The government already publishes an annual dual-use export report; semiconductor equipment deserves its own section.
Fourth, keep pursuing alignment with partners. The 2023 approach was framed as complementing multilateral action, and unilateral Dutch rules that diverge from allies' rules both burden ASML and invite workarounds. Coordination is the only way controls bind without simply relocating the tool trade.
The bottom line
ASML's quarter shows a company that cannot build fast enough for customers who want AI compute. Dutch export policy was built to answer "who must not get this tool?" It now also shapes, however indirectly, "who gets it first?" A pro-innovation approach keeps the security question sharp, narrow and evidence-based, and makes the process predictable enough that investors can plan around it. Opacity is tolerable when the queue is short. In a sold-out market it becomes industrial policy by default.