The European Commission wants to spend up to €10 billion in public funds on seven gigafactories, data centers full of cutting-edge AI chips designed to help European researchers, startups, public services, and industrial firms catch up in the AI race.
It’s a good idea to accelerate access to AI compute. But the gigafactory agenda risks reviving a damaging European instinct: equating sovereignty with self-sufficiency and treating imported technology as evidence of strategic failure.
Gigafactories will be consumers of chips, not manufacturers of them. European gigafactories will use imported GPUs, not create a European GPU, foundry, packaging ecosystem, or AI software stack. The “gigafactories” label blurs compute infrastructure with semiconductor production and creates false expectations.
Semiconductor supply chains are too specialized, capital-intensive, and geographically distributed for any country to reproduce the entire chain at home. As OECD semiconductor analyst Jan-Peter Kleinhans has emphasized, resilience means managing critical interdependencies, not pursuing the impossible goal of self-sufficiency. Europe’s goal should be to become indispensable in the global semiconductor supply chain and AI compute system, not to wall itself off from the US, Taiwan, Japan, South Korea, Singapore, or the UK.
The European Commission’s own financing model recognizes that public money cannot, and should not, carry the whole gigafactory burden. It aims to use the public money to encourage private investors to invest $20 billion.
That’s good. The unanswered question is whether Europe can attract enough private demand to ensure gigafactories become economically and strategically useful. European taxpayers risk subsidizing US and Asian tech. A subsidized AI cluster without paying users will become an expensive and wasteful public customer of US chip and software vendors. The 100,000 advanced AI processors per gigafactory can only come from
US firms such as NVIDIA and AMD, manufactured mainly by TSMC in Taiwan and Samsung in South Korea.
The fact that Europe imports these technologies is not, by itself, a problem. It reflects a supply chain based on specialization: US design, Dutch lithography, German chemicals and optics, Belgian research, Taiwanese and Korean manufacturing, Japanese materials, and Southeast Asian assembly and testing. Policy should seek redundancy, and secure access and mutual dependence. It should not attempt to make Europe self-sufficient in every layer.
Gigafactories should avoiding deepening reliance on a single US cloud, GPU, or software provider. Openness will help — the software connecting chips and the software running the AI models, should be, as much as possible, open and interoperable, not proprietary. Open products allow customers, not suppliers, to keep control of their outputs.
Europe should concentrate on bolstering its already substantial strengths: Dutch ASML’s lithographic chipmakers; German Zeiss optics; Belgian IMEC’s semiconductor research; specialty chemicals and materials; power semiconductors; automotive chips; sensors; industrial automation; photonics; and embedded systems. The correct policy question to ask is not “is the asset European?” but “does it make Europe harder to bypass?”
NanoIC, a pilot line at IMEC Leuven opened in February with €2.5 billion in total investment, represents a good example of strategic industrial policy — rather than an imagined and impossible-to-achieve goal of generating a European NVIDIA. Unlike a commercial mega-fab, NanoIC is a shared pilot line, allowing companies and researchers to develop and validate advanced technologies before committing to industrial-scale production. The priority should be to keep NanoIC and related facilities accessible to startups, SMEs, universities, and trusted international partners — not allow them to become closed assets of a few incumbent firms.
These design and materials strengths require nurturing a network of specialist funds such as Cloudberry VC, the continent’s first dedicated semiconductor venture-capital outfit. This €30 million fund invests €400,000-€1 million in pre-seed and seed-stage startups. Its goal is to create the next Arm, the UK-founded design firm behind almost all the chips in mobile phones.
Arm’s success shows why the UK needs to be included in any European chip strategy. Cambridge’s chip-design cluster, compound semiconductors in South Wales, and strong university research make it indispensable. A serious EU strategy would create an EU–UK semiconductor and AI compute partnership: reciprocal access to pilot lines; joint R&D calls; cross-border startup funds; coordinated standards; common skills programs; and aligned export-control and investment-screening rules.
If the gigafactories are to become a miracle and not a mirage, a few key rules should be followed. Rather than build them where politically convenient (think Germany), they should be built in locations with affordable low-carbon energy, available grid capacity, rapid permitting, skilled operators, research institutions, network connectivity, and identifiable demand (think Nordics or Eastern Europe).
Gigafactories should require genuine private risk-sharing. A defined share of compute should be reserved for public-interest and strategic use: universities, startups, and small industrial firms.
Foreign participation should not be treated as a strategic defeat. A European-owned data center using a single proprietary software stack may be less resilient than a multi-vendor, interoperable service operated with trusted partners. The choice is not “European sovereign hardware,” a brittle, politically attractive isolation. It is instead building a diversified, competitive, open AI sector that emphasizes European control.
Christopher Cytera CEng MIET is a senior fellow with the Tech Policy Program at the Center for European Policy Analysis and a technology business executive with over 30 years of experience in semiconductors, electronics, communications, video, and imaging.
Bandwidth is CEPA’s online journal dedicated to advancing transatlantic cooperation on tech policy. All opinions expressed on Bandwidth are those of the author alone and may not represent those of the institutions they represent or the Center for European Policy Analysis. CEPA maintains a strict intellectual independence policy across all its projects and publications.
Steel and Silicon: Allied Innovation for the Next Century
2026 CEPA Tech & Security Conference
