ASML Makes Its Move

The Chipmaker's AI

Ordinarily, the company that makes the printing presses does not buy a stake in the author. In the neat, layered world of technology, hardware companies make the physical stuff, and software companies make the virtual stuff that runs on it. This week, that tidy separation was shattered. ASML, the Dutch monopolist that makes the machines that make the world's most advanced chips, just invested €1.3 billion to become the largest shareholder in Mistral AI, France's answer to OpenAI.

The deal is, on its face, a strange one. ASML operates at the deepest, most physical layer of the tech stack. Mistral operates at the highest, most abstract layer. The logic, it turns out, is deeply practical. ASML isn't just investing in the abstract idea of AI; it is buying a toolkit to make its own products better. The company plans to embed Mistral's AI models directly into its multi-million-dollar lithography machines. This creates a competitive advantage: smarter machines that can help its customers, like TSMC, Intel and Samsung, manufacture chips with higher yields and at a faster pace. As ASML's CFO Roger Dassen explained to The New York Times, the goal is to "better understand and influence the technology's development."

This is, in theory, a powerful combination, and a cornerstone of Europe's attempt to build a homegrown AI champion. The alliance pairs the continent's most critical hardware company with its most credible software contender. Here is ING analyst Jan Frederik Slijkerman on the industrial rationale, via Reuters:

"There is an industrial rational to develop products together," [Slijkerman said]. "For ASML it is probably easier to develop AI based products through a partnership then to do this in house."

The problem, of course, is one of scale. The deal values Mistral at a formidable €11.7 billion, making it Europe's most valuable AI startup. But its American rival OpenAI is reportedly seeking a valuation of around $500 billion. This is Europe's most coherent attempt yet to build a homegrown AI champion. The only question is whether it's a few years too late.

OpenAI's Original Sin

OpenAI's original sin was its nonprofit status. It was founded in 2015 on a noble, tax-advantaged mission to ensure artificial general intelligence "benefits all of humanity." But it turns out that building AGI is spectacularly expensive, which led the company to create a unique and convoluted "capped-profit" subsidiary to raise tens of billions from investors. For a while, this worked. But now that strange structure has become a crisis, trapping the world's most valuable startup between the high-minded ideals of its past and the brutal, capital-intensive demands of its future.

The company's plan to restructure into a traditional for-profit corporation is now under investigation by the attorneys general of California and Delaware, imperiling its future fundraising and a potential public listing. The Wall Street Journal reported on the mounting pressure:

OpenAI executives are growing concerned that mounting political scrutiny in California could stymie their efforts to become a for-profit company and have discussed a last-ditch option of moving out of the state. ... Failing to restructure could be catastrophic for the world's most valuable startup... OpenAI's financial backers have conditioned roughly $19 billion in funding—almost half of the startup's total in the past year—on receiving shares in the new for-profit company. If the restructure doesn't happen, they could pull their money.

The legal weapon being wielded against OpenAI is California's charitable trust law. Because OpenAI's parent is a nonprofit, the state's attorney general has a duty to "protect assets held in charitable trust." The core argument from a growing coalition of opponents—including nonprofits, labor unions, and rival AI companies—is that OpenAI used its tax-exempt status to develop its incredibly valuable AI models. Those models, they claim, are a public asset built for the benefit of humanity, not a private one to be transferred into a for-profit company to enrich shareholders.

This is not just a philosophical debate; it is a corporate knife fight. Rivals like Meta and Elon Musk's xAI have reportedly urged regulators to block the conversion, seeing a strategic opportunity to kneecap their primary competitor by weaponizing its own founding charter against it.

It is a delightful and expensive mess. OpenAI needs the capital and structure of a for-profit entity to fund its $115 billion cash burn and win the AI arms race. But it is legally and morally bound by the nonprofit promises of its past. The battle will test whether it's possible to "save the world" while also delivering venture-scale returns, especially when your competitors are happy to hand the authorities a copy of your mission statement to use as evidence.

The Scoreboard

• Semiconductor: Nvidia’s ‘Wow’ Factor Is Fading (WSJ)
• Cloud Infra: Oracle stock soars as CEO says AI-fueled cloud revenue set to soar to $144 billion (Yahoo Finance)
• EV: Chinese EV maker Xpeng eyes global launch of mass-market Mona series in 2026 (CNBC)