Europe Should Skip the 3nm Chip Race and Go All-In on 2nm

The semiconductor race is the new space race, and Europe is falling behind. While Asia and the U.S. are pushing hard into 3nm and even 2nm chip production, Europe is cautiously betting on incremental progress. But time is not on our side. Instead of playing catch-up at 3nm, Europe should make a bold leap directly into the 2nm era, just like Japan is doing with the Rapidus project.

The EU Chips Act is a good start, but it lacks urgency. Its target of 20% global market share by 2030 sounds ambitious, but without a clear push into leading-edge nodes, it's just a number. Most EU-funded fabs are still focused on mature or mid-range nodes. Meanwhile, Rapidus, backed by the Japanese government and major industrial players, is building a 2nm fab with a go-live target of 2027. They have IBM’s 2nm tech, ASML EUV machines, and are training engineers at IMEC. It’s a national moonshot, and they're moving fast: Japan's first EUV tool is installed, pilot lines are underway, and prototype chips are expected within two years.

Europe already has the assets to follow suit. We have ASML; the only company in the world that produces EUV lithography machines. We have IMEC and CEA-Leti, world-class research institutions already collaborating on sub-3nm process technologies. We have industrial champions like STMicroelectronics, Infineon, and NXP, who could act as anchor customers. And we have the automotive, aerospace, and defense sectors eager for sovereign access to advanced chips.

What we lack is a focused, Apollo-style mission. The EU Chips Act is a sprawling, multi-track effort, but lacks a single, ambitious objective. It spreads funds across dozens of smaller initiatives instead of concentrating firepower where it matters most. A Europe-wide 2nm program would rally talent, capital, and supply chains toward a strategic goal: building the most advanced logic chips on our own soil.

What will these 2nm chips actually power? They are not just about thinner smartphones. These chips will be foundational to the next generation of high-performance computing, AI acceleration, advanced defense systems, autonomous vehicles, 6G communications, and quantum-enabling architectures. Their ultra-dense, energy-efficient designs are critical to training massive AI models, running edge inference on low-power devices, and enabling real-time decision-making in safety-critical environments like medical robotics or aerospace. In other words, they will sit at the heart of tomorrow’s global infrastructure.

Rapidus is also crafting a business strategy deeply aligned with its technical model, an approach Europe would do well to study and adapt:

1. Single-Wafer Processing for Agile Prototyping: Technically, Rapidus focuses on single-wafer processing rather than batch production. Commercially, this allows them to serve customers needing fast design cycles: AI startups, aerospace firms, or national defense labs who value turnaround speed over volume. In other words: no need for big orders to to start producing. Smaller orders becomes possible.
2. Dynamic Wafer Routing for Resilience: Their modular fab design enables rerouting wafers around failed or idle tools. The business payoff? Higher reliability and uptime, making them a dependable partner for mission-critical supply chains.
3. Embedded R&D and Pilot Lines: With R&D and pilot manufacturing co-located, Rapidus shortens feedback loops. This creates a business model based on co-development partnerships, especially attractive to cutting-edge chip designers and early adopters.
4. Compact, Vertical Integration: A smaller footprint and close partner integration enables fast deployment and better cost control. Rapidus can appeal to domestic clients seeking regional, secure chip sourcing without needing massive economies of scale.
5. Did I say they want to test Realtime the wafers through AI?

Europe can and should learn from these strategies, especially as it builds around its own industrial strengths and geopolitical objectives.

Importantly, any 2nm effort must be truly pan-European. The investment required, likely tens of billions of euros, cannot be shouldered by a single nation or even a single company. It must be a joint initiative, with coordinated funding and governance. The returns, too, should be shared: revenues, jobs, and innovation spillovers should benefit multiple countries. The European Commission and its institutions must play a central role, not only to provide stability and oversight, but to ensure alignment across borders. At the same time, the private sector must be fully engaged, industry players from across Europe should co-invest and co-own the project (car, defense manufacturers, Cloud providers, universities, telcos etc). This blend of public coordination and private commitment is the only way to make such an ambitious program both credible and resilient.

Yes, the risks are high. Rapidus could fail. A European equivalent could burn billions. There may be yield issues, customer skepticism, or timeline slips. But the rewards? Strategic autonomy, reduced supply chain exposure, stronger negotiating power in alliances, and a reinvigorated tech sector. Even a partial success would elevate Europe’s semiconductor standing significantly.

From a geopolitical lens, this is about relevance. The U.S. has the CHIPS Act. Japan has Rapidus. Korea has Samsung. China is pouring state capital into its own efforts. If Europe doesn’t push to the frontier, it risks becoming dependent on others not just for chips, but for everything those chips power: AI, quantum, defense, advanced manufacturing.

And critically, aiming for 2nm forces us to integrate our fragmented semiconductor ecosystem. ASML in the Netherlands. IMEC in Belgium. Research in France. Industrial customers in Germany. A 2nm program would be a forcing function for pan-European cooperation at the highest level.

If we aim for 3nm, we’ll always be behind. But if we leap for 2nm, even if we fall short, we’ll land far ahead of where we are today.

We must stop playing defense.

‍