Europe Built the Plants. Now It Needs the Companies.

Europe has spent the last several years racing to subsidize new chip fabrication plants, pouring tens of billions of euros (tens of billions of US dollars at recent exchange rates) into a smaller number of headline projects. The harder, higher-value test is whether industrial policy can also grow the design-led companies that own customer relationships, intellectual property (IP in the chip-design sense), and product roadmaps. That is where modern semiconductor value actually lives.

The argument, as a recent EE Times analysis puts it, is constructive rather than critical. The fabs are necessary. The missing layer is champions.

Where the value actually sits, the picture is more familiar than the subsidy headlines suggest. The most successful semiconductor businesses in the world rarely own the largest fabs (chip fabrication plants). They own customer relationships, system expertise, IP libraries, architecture choices, and long-term supply contracts. The fab is a stage. The orchestration is the business.

A single chip product can involve dozens of organizations across multiple countries: research universities writing the underlying physics, EDA (electronic design automation) software vendors, semiconductor IP developers, chip architects, foundries, assembly and test providers, product companies, system integrators, and end customers. The largest share of profit typically goes to the company that pulls these pieces into a product a customer will pay for, year after year. That is the layer Europe's subsidies have underweighted.

Europe does not start from zero on the customer side. Its industrial base gives it an unusually deep set of verticals where performance, reliability, safety, and longevity matter more than raw transistor count: automotive electronics, where electrification and software-defined vehicles are turning every car into a rolling data center; industrial automation, where long product lifecycles favor customized silicon; aerospace and defense, with their own certification regimes and supply-chain rules; secure communications, including government and defense-grade cryptography; medical technology, where regulatory cycles run in decades; power electronics, which sit at the boundary between the silicon world and the grid; RF (radio frequency) and wireless systems, including satellite communications (satcoms), where Europe already has incumbents; and semiconductor equipment, where ASML alone sets the pace for the entire global industry.

In each of these, generic off-the-shelf chips do the job poorly. Buyers want application-optimized silicon tuned to their standards, their lifetimes, and their supply chains. That is a market designed for design-led companies, not for the fab that prints the wafers.

The shift in demand makes the policy question sharper. "The days of designing one chip to do everything are ending," as the EE Times analysis observes. AI accelerators, autonomous-vehicle compute, satcom payloads, industrial controllers, and security appliances each want different silicon: different memory hierarchies, different power budgets, different certification paths. They want, in other words, the kind of specialized designs that European system companies are already positioned to specify. That divergence plays to European strengths. The same trend that pressures a fab-centric strategy also creates a market for the customer-owned designs Europe is well placed to build.

Fabs are slow, capital-intensive, and visible, but they are the wrong place to look for the next layer of European competitive advantage. The next investment round has three plausible levers. Procurement can anchor long-term offtake contracts that let European design houses scale to production volumes without a domestic foundry. Capital can fund the IP-and-product stage from Series B onward, where US-style venture has produced Nvidia, AMD, and Marvell, rather than only the fab stage. IP policy can tighten European licensing frameworks so design houses can defend their architectures against commoditization. None of these are substitutes for the fabs already being built. They are the layer above.

The next twelve to twenty-four months will be informative. Intel's Magdeburg site, TSMC's Dresden expansion, and the broader European Chips Act pipeline will move from announcement to construction. If policy attention stays anchored on those ribbon-cuttings, the design-IP layer will stay underfunded. If a meaningful share of the next investment cycle flows toward design houses, IP developers, and product companies in automotive, industrial, aerospace, and RF, Europe will have a real shot at the layer where the margin is.

The plants are necessary. They are not sufficient. The companies that design what those plants make are the next test.