In semicon, sovereignty begins with demand, not with production

At the launch of the Casimir Institute in Eindhoven, Paul de Bot, Head of TSMC Europe, warned that Europe’s semiconductor ambitions will falter unless the continent builds demand and scale alongside factories. “Production follows demand; it’s never the other way around.”

The opening of the Casimir Institute at Eindhoven University of Technology was more than a ceremonial ribbon-cutting. It marked Europe’s determination to claim a stronger position in the global semiconductor industry, one of the most strategic and contested sectors of the 21st century. Speakers from inside and outside (like Lars Reger, Jo de Boeck, Pierre Chastanet, Vincent Karremans, and Richard Kemkers) underlined that message.

On stage, Paul de Bot, a Dutchman who now heads TSMC Europe, reflected on how the world has changed since his own days as a student in Eindhoven. In 1987, he designed and tested chips in a small university fab, working with second-hand Philips equipment. “At that point in time, if you wanted to work on chips, you had to produce them yourself,” he recalled. But that same year, halfway across the globe, Morris Chang founded Taiwan Semiconductor Manufacturing Company (TSMC). “By doing that, he unleashed the foundry model; and with it, a whole new industry of fabless companies.”

From fabs to fabless

That model, De Bot explained, separated design from manufacturing. Instead of each integrated device manufacturer (IDM) carrying the massive burden of both transistor R&D and wafer production, TSMC shouldered the heavy lifting. “Those two pieces - the development of transistors and wafer manufacturing - required enormous investment and created huge entry barriers,” he said. “By taking them on, TSMC made it possible for a whole new category of companies to emerge: fabless chip designers.”

The results speak for themselves. Today, half of the global semiconductor industry is controlled by fabless companies. They are also capturing most of the value creation. “If you want to participate in the growth of the semiconductor industry, focus on fabless innovation,” De Bot urged. “Don’t get obsessed with manufacturing.”

TSMC itself has grown into the world’s largest contract chipmaker, producing the most advanced chips used in smartphones, data centers, and increasingly, artificial intelligence. Its strategic importance has put Taiwan at the center of geopolitics, an uncomfortable reality that underpins Europe’s push for greater “strategic autonomy.”

Chips Act ambitions

That push has taken shape in the European Chips Act, announced in 2022 with the goal of doubling Europe’s share of global chip production to 20% by 2030. Billions of euros in subsidies are now flowing to build new fabs in Germany, France, and Italy, with Intel and TSMC both announcing major investments.

But De Bot cautioned that factories alone will not deliver autonomy. “Production typically follows demand. If there is demand, production will follow. It’s not the other way around,” he said. “Some people say: if you build it, they will come. That doesn’t work in the semiconductor industry. The most costly, loss-making asset is an under-loaded chip factory.”

His message cut to the heart of Europe’s dilemma: how to ensure that new fabs won’t stand idle while most of the demand is driven by American and Asian tech giants?

Demand starts with software

According to De Bot, Europe needs to consider the entire value chain. “You have to look at this picture from left to right", De Bot says while pointing to the slides behind him. "It starts with software, with workloads. Major software companies are currently driving the majority of chip demand. They enable data centers, cloud providers, and, in turn, chip designers,” he explained.

Without strong European software players, the continent risks remaining a supplier rather than a shaper of demand. “If Europe wants to play, it has to enable European winners from left to right,” De Bot said. “That means starting with software and building strengths across the chain.”

Here, scale is everything. The European Union has pledged €200 billion for AI development; it's an impressive figure, until you compare it to the spending power of U.S. tech giants. “Each of them spends similar amounts,” De Bot noted. “If you want to create a big European winner, it’s important that choices are made. One or two champions must emerge. If the €200 billion is split into 27 smaller buckets, it won’t be enough.”

Moore’s Law lives on

De Bot also stressed that innovation at the cutting edge remains Europe’s best bet for relevance. Despite frequent predictions of its demise, Moore’s Law endures, with transistors continuing to shrink and energy efficiency improving. “Moore’s Law is there at least until 2042, if not beyond,” he said.

This relentless progress enables high-performance computing, the fastest-growing market segment, fueled by AI. “AI workloads are driving stellar growth in data centers and cloud infrastructure,” De Bot explained. “That’s where demand is, and that’s where Europe must decide whether to participate.”

The implication is clear: if Europe fails to invest in advanced design and system integration, its new factories will serve others’ innovation rather than its own.

A rat race

In closing, De Bot reminded his audience of the intense competition in the semiconductor industry. “This is a rat race,” he said. “Companies all over the world are trying to get the same business. You need to innovate fast. We cannot sit back and believe that we can build on old strengths. We have to build new strengths.”

For the newly launched Casimir Institute, dedicated to advancing semiconductor research and collaboration in Eindhoven, the words were both a challenge and a call to action. Strategic autonomy, De Bot argued, cannot be achieved solely through subsidies. It will require Europe to cultivate demand, foster software and design champions, and make hard choices about where to concentrate its resources.

Factories may be symbols of sovereignty, but in the semiconductor industry, sovereignty begins with demand.