Logo

Why ASML is investing $1.5B in Mistral - and why it makes sense

With Nvidia as linchpin and Mistral as the vehicle, ASML is buying control over the AI layer that will help determine lithography's future.

Published on September 8, 2025

ASML Campus Veldhoven

ASML Campus Veldhoven

Bart, co-founder of Media52 and Professor of Journalism oversees IO+, events, and Laio. A journalist at heart, he keeps writing as many stories as possible.

ASML plans to invest approximately €1.3 billion ($1.5 billion) in Mistral AI, reportedly becoming the largest shareholder, including a seat on the board of directors. The round would value Mistral at approximately €10 billion, making the French startup Europe's most valuable AI company. The stakes are higher than a single deal: this is industrial policy via the market.

The news of this investment is striking for several reasons. We list the most important ones below, in the hope of assessing the logic behind the decision.

asml mistral logo

Mix of logos, by © IO+

1 - “Put your money where your mouth is”: strategic autonomy for Europe

Europe has been talking about digital sovereignty for years. By financing a European champion in generative AI – and doing so from the European center of gravity in chip production – ASML is turning words into action. Reuters explicitly states that the deal is intended to reduce dependence on American and Chinese AI models, which is precisely the core of the European autonomy agenda. Although this investment is a corporate decision, it seems inconceivable that it was not thoroughly discussed with the European Commission beforehand.

2 - A response to Trump's protectionism

In Washington, the protectionist script is back. The Trump administration announced import tariffs on semiconductors for companies that do not bring production to the US; at the same time, a broader, lower limit on import tariffs applies, affecting the entire tech chain. For AI players that purchase large quantities of advanced chips, this increases the political and cost pressure on US infrastructure. More European AI capacity (and capital) is therefore a rational hedge.

3 - “Hardware meets software”: how unusual is this for ASML?

ASML rarely makes large, direct equity deals outside its own core business. An investment in SMART Photonics two years ago is the exception. More familiar is participation in deep-tech funds (such as DeepTechXL) or targeted acquisitions within the lithography chain. A billion-dollar stake in a pure software/AI player is therefore exceptional—and precisely why it is interesting. It fits in with ASML's holistic strategy, in which more and more value lies in software and data-driven optimization (computational lithography, metrology, service).

But could it nevertheless make sense? Mistral's models and tooling can be directly integrated into ASML's development and production chains, enabling data-driven yield improvement, faster root-cause analysis, predictive maintenance, and improved EDA interfaces. In short: more AI means more uptime, higher throughput, and faster iterations at the limits of physics, which is exactly where ASML competes.

asml mistral logo

ASML invests $1.5 billion in European AI leader Mistral

ASML's investment in Mistral is seen as an important step in strengthening European technological sovereignty.

4 - The Nvidia triangle: ASML × Nvidia × Mistral

ASML has been working closely with Nvidia for years on GPU-accelerated computational lithography (cuLitho) – achieving a 40-60x acceleration on the most compute-intensive step of chip production. At the same time, Mistral is intensifying its collaboration with Nvidia for “sovereign” HPC infrastructure in France. This creates a strategic triangle: Nvidia provides the computing layers, Mistral the AI models, and ASML the application in tooling and production. This alignment increases the likelihood that Mistral's stack will have a rapid industrial impact.

5 - A harbinger of a broadening of ASML's focus?

Yes and no. No, because ASML is not suddenly becoming a consumer-oriented AI player; its core business remains lithography for chip manufacturers. Yes, because the boundary between “machines” and “software” is becoming even more blurred. The unique performance of EUV/High-NA increasingly depends on algorithms, data pipelines, and model-driven optimization. A governance position at Mistral accelerates access to talent, IP, and integration points, while securing European AI capacity close to ASML's R&D axis.

6 - Did the French CEO play a role?

Christophe Fouquet

Christophe Fouquet, CEO of ASML since April 2024, is French. This makes the outlook for Paris and Brussels favorable and could accelerate the forging of a “European coalition” around AI. There is no evidence that his nationality was decisive, but it does reduce political and cultural friction. In short, it is certainly not the main reason, but it does help.

7 - Link with US tariffs and data centers in the EU

If the US taxes imports of chips not made in the US, AI capacity in Europe will become relatively more attractive, especially for players that consume a lot of TSMC chips. Mistral is building (in collaboration with Nvidia and French partners) a sovereign HPC infrastructure in France, which shifts computing demand and chip absorption to the EU, thereby avoiding the reach of US tariffs and compliance risks. This is doubly beneficial for ASML: it stabilizes European demand for advanced nodes and dampens geopolitical volatility.

8 - Risks and side effects

  • Antitrust/collusion: customers may be suspicious if ASML takes stakes in tooling suppliers or AI players that also work with competitors. Clear firewalls and non-discrimination are crucial. (Reuters explicitly reported that the rationale is “tool efficiency” and European sovereignty.)
  • Political headwinds: tariff and export regimes are fluid and are being challenged in the US. A sudden policy change could shift value cases. The US (or China) could also view ASML's action as hostile, requiring a countermeasure.
  • Execution: value will only be realized if Mistral's models demonstrably improve yield/throughput and if the European HPC plans achieve scale on time.

It's about control

The deal is less strange than it seems at first glance. There are obviously risks and uncertainties, but the opportunities are great enough to see the move as strategically interesting. ASML is moving from physics-driven mechanical engineering to AI-accelerated system optimization, embedded in a European autonomy strategy and hedged against American protectionism. With Nvidia as the linchpin and a French AI champion as the vehicle, ASML is not only buying equity—it is buying control over the AI layer that will help determine the next generation of lithography.