EUROPA

Europe holds both ends of the hardware stack. The middle is missing.

Europe holds both ends of the hardware stack and is missing the middle. At one end: ASML in Veldhoven builds the EUV lithography machines the whole world's leading-edge chips depend on, and Europe runs real fabs for non-cutting-edge chips (power, analog, automotive, sensors) at GlobalFoundries Dresden, Bosch Dresden, Infineon, STMicroelectronics, NXP, X-FAB. At the other end: European software is genuinely strong (SAP, enterprise and developer tooling, a deep research base).

What's missing is the layer in between — turning chips and components into finished, manufacturable hardware products at speed and at the volumes a real market needs. That layer is what the Pearl River Delta (Shenzhen, Dongguan, Shenzhen's hinterland) built over 30 years: dense contract manufacturing, fast PCB turnaround, component distribution deep enough to buy anything same-day, and thousands of small shops that know how to take a prototype to a production line in weeks, not quarters. Europe has none of that at comparable density, so European hardware founders — even ones building on European chips — still fly to Shenzhen to get a product made.

This page lays out why that gap exists, why it matters strategically, and what a plan to close it actually needs to contain.

What Europe already has

• Lithography and fab equipment. ASML (NL) is the sole source of EUV machines. Zeiss SMT and Trumpf (DE) supply the optics and lasers ASML depends on. ASM International and BESI (NL) supply deposition and assembly equipment. This is the strongest link in the whole European chain — genuine, unmatched leverage.
• Fabs, at the trailing edge. GlobalFoundries Dresden and Bosch's Dresden fab (opened 2021) make power/analog and automotive chips. Infineon (DE, the world's largest automotive-chip maker), STMicroelectronics (FR/IT), NXP (NL) and X-FAB (DE/Belgium/France) run fabs for exactly the chip classes most hardware products actually need — not AI accelerators, but the sensors, power management and microcontrollers inside almost everything physical.
• Software. SAP, a deep base of developer tooling and research, and increasingly capable European AI labs.
• The EU Chips Act (2023). €43bn aimed at fabs and R&D. It targets capacity and sovereignty at the silicon layer — it does not fund the manufacturing layer this plan is about.

The pattern: Europe's public money and industrial strength cluster at the capital-intensive, scientifically hard ends of the stack. The middle — turning a chip into a shipped product — has had no comparable attention or investment.

The three tiers

What's missing

The Pearl River Delta isn't one factory, it's a system. Five parts of it don't exist in Europe at comparable density:

• PCB fabrication at speed. Shenzhen-area board houses turn a 2-layer prototype in 24 hours and an 8-layer board in days. Europe has real advanced-substrate capability (AT&S in Austria is world-class), but it's built for large, planned production runs — not a founder who needs 10 boards by Friday to keep iterating.
• Contract manufacturing (EMS) at small-to-medium batch, economically. European EMS exists — Kitron (Nordics), Zollner (DE), Scanfil (FI), Variosystems (CH), Flex and Jabil's Central European plants — but it's priced and structured for large committed volumes. A founder needing 500–5,000 units at a workable price and a fast changeover has very few European options, and often none that will even quote a run that small.
• Component distribution depth. Huaqiangbei (Shenzhen's electronics market) and its wholesale networks mean almost any component, including obsolete or hard-to-source ones, is buyable same-day, in any quantity, often for cash. Europe's distribution is reliable for standard parts at normal lead times, but has none of that improvisational depth for the parts that don't show up in a normal BOM.
• Design-for-manufacturing expertise, embedded and cheap. In Shenzhen, DFM review, mold design, tooling and fixture-making are a commodity service — fast, inexpensive, and done by people who've seen thousands of products go through the same steps. In Europe this expertise exists but sits inside a handful of expensive engineering consultancies, not as a cheap, abundant service any founder can tap.
• The informal network effect. The single hardest thing to replicate: in the PRD, a factory that can't do something knows exactly which factory three streets over can, and makes the introduction for free. That density of knowledge, built over 30 years and millions of transactions, is what actually makes the system fast. No amount of capital builds this quickly — it's the real moat, and the real multi-decade challenge for any European plan.

Europe doesn't lack engineers, capital or chips. It lacks this layer — and it's the layer that turns the other two into a shipped product.

Who's already building the pieces

None of this is hypothetical. Pieces of exactly this stack already exist or are actively being built, mostly without a shared plan connecting them:

What's missing isn't any one piece — it's a plan that deliberately connects them into one stack, rather than each solving its own corner independently.

Component distribution: Avnet, Arrow, and the WPG signal

Avnet and Arrow Electronics are the two largest global distributors with deep European operations — Avnet through Avnet Silica, Avnet Abacus and Farnell/element14, and Arrow through Arrow Central Europe and a genuine value-added design-services business. Alongside them, strong European-headquartered independents matter just as much: Rutronik and Würth Elektronik (both German), RS Group (UK), TTI, Future Electronics, and the catalog players Mouser and Digi-Key. This layer is good at authorized stock with no counterfeit risk and reliable lead times on known BOMs, but it still lacks the improvisational, cash-market depth for an obsolete or 50-unit-prototype part.

The WPG signal: WPG Holdings, one of Asia's largest semiconductor distributors and the primary franchised gateway into a long list of Chinese fabless chipmakers, began a greenfield EMEA operation in 2021 — led by Nigel Watts as President of WPG EMEA, holding inventory in Europe via a Netherlands distribution center, and signing supplier lines such as Alliance Memory and Gowin FPGAs. It's a distribution-layer bridge to the same PRD-connected chip supply that feeds gongban reference designs, now reachable from inside the EU — a leading indicator worth tracking, not yet proof the depth gap is closed.

Other gaps this plan hasn't addressed yet

• Tooling and injection molding at prototype economics. Europe's mold shops are built for automotive-scale volumes with multi-month lead times and high NRE. Shenzhen's low-cost, week-scale tooling for enclosures has no European equivalent here yet, and it's often the single slowest step in hardware NPI.
• Battery and cell supply chain — actually the clearest precedent for this whole plan. Europe's one major attempt at building this from scratch, Northvolt, went insolvent in 2024–2025. But the capacity that's actually landing in Europe is coming the same way this plan proposes for hardware manufacturing: Asian producers setting up on European soil. LG Energy Solution runs one of the world's largest battery plants near Wrocław, Poland. CATL now runs two European gigafactories — Arnstadt, Germany and Debrecen, Hungary (first phase already fully booked by BMW, Mercedes-Benz, Stellantis and Volkswagen) — plus a joint plant with Stellantis in Zaragoza, Spain. Greenway runs a European office in Belgium and has signaled 2026 European production. This is the pattern this plan is proposing for the manufacturing layer, already working for batteries.
• Certification and test-lab speed. CE marking is a genuine European strength — one regime across the whole single market — but EMC/RED test-lab turnaround and cost for a small-batch startup is slow and expensive next to Shenzhen's co-located test houses.
• This is explicitly a technology-transfer initiative, and that's the point. Kingbrother, SMT Machine Line, and the gongbans design-reuse model aren't an accident to explain away — bringing PRD manufacturing knowledge, equipment and partnerships into Europe is the method, stated plainly rather than left implicit. Resilience here doesn't mean decoupling from Asia; it means geographic proximity, shorter logistics, and a European option that doesn't exist today — built by transferring expertise that already exists in the PRD, not by reinventing it from zero. The partnership is the near-term mechanism, not the end state. The PRD itself built three decades of density and expertise the same way — by doing the work, not by starting independent. The plan's actual bet is that transferred knowledge compounds into real European capability and reduced China dependency over time.
• No collective voice. Sectors that got sustained EU attention and funding — automotive, semiconductors — have an organized lobby. This layer has none yet.
• The talent pipeline, made concrete. Europe's vocational training (Germany's Ausbildung system is the strongest example) is not currently oriented toward SMT operation, hand-rework, or DFM at consumer-hardware speed. It would need deliberate redirection.

The stakes, plainly

• Value capture. A European-designed, European-chipped product that gets manufactured in Shenzhen still sends the manufacturing margin, the jobs and the know-how abroad.
• Resilience and sovereignty. The EU Chips Act addresses the fabs. It doesn't address what happens if the PRD becomes harder to reach — export controls, a Taiwan contingency, tariffs, or rising Chinese labor costs. Right now there is no European fallback for turning a chip into a product at scale.
• Defense and dual-use hardware. European defense and dual-use programs increasingly cannot use Chinese manufacturing for sourcing-security reasons, yet have nowhere domestic to go for fast-turn, small-batch production.
• Climate and logistics. Air-freighting prototypes and small production runs between Europe and Shenzhen, repeatedly, has a real carbon and time cost a local manufacturing layer would remove.
• Founder gravity. Hardware founders go where the manufacturing is easy. If that stays Shenzhen, European hardware startups keep orienting their supply chains — and increasingly their teams — toward Asia by default, not by choice.

This is the missing chapter of EuroStack

EuroStack is the closest thing Europe already has to a coordinated movement on exactly this kind of dependency. Led by Francesca Bria and backed by over 500 industry signatories, it's an industrial-policy push to reduce Europe's reliance on mostly US-owned digital infrastructure — connectivity, cloud, AI, and digital platforms. The underlying blueprint nominally covers semiconductors too, but the public campaign is almost entirely about software and cloud; it has no chapter on what happens physically between a chip and a shipped product.

That's exactly this plan's subject. EuroStack argues Europe shouldn't depend on US-owned cloud and AI platforms; this plan argues Europe shouldn't depend on Shenzhen to turn a chip into a product. They're the same argument, aimed at two different dependencies, on two different layers of the same stack. This plan is the hardware-manufacturing chapter EuroStack's own semiconductor section gestures at but hasn't written.

A legal lever: EU Inc. plus a 51% rule

EU Inc. — the "28th regime" the European Commission proposed in March 2026 — lets a founder register one pan-EU legal entity online in 48 hours, instead of picking a single national jurisdiction. That's a real lever here: it's exactly why Europe's pieces of this stack are scattered today (CATL picked Hungary, LG picked Poland, Eilbek picked Hamburg) — a Chinese or Asian manufacturing partner could register once and operate across the whole single market from day one.

The sharper move: pair it with a 51% EU-ownership rule, borrowed directly from China's own playbook. For decades China required foreign entrants in strategic sectors to operate as joint ventures with a Chinese partner holding at least 51% of the equity and board control — the actual mechanism that forced real technology transfer into Chinese hands. Europe is earlier in the same cycle and currently holds the same leverage China once held. Applied here: any Asian manufacturing partner entering under this plan sets up as an EU Inc. joint venture with at least 51% EU ownership and governance control. EU Inc. is the carrot; the 51% rule is the stick, made palatable by that speed.

Why this layer is missing

• Labor cost, but less than people assume. What remains dominant in Shenzhen isn't cheap labor — it's density, speed and flexibility, which are organizational, not wage-driven.
• Fragmentation across 27 states. The PRD's advantage comes from everything sitting within a few hours' drive. Europe's equivalent capacity is scattered across countries with no shared logistics or supplier network binding it together.
• Risk aversion and capital patterns. European capital is structurally wary of capital-intensive, slower-return manufacturing bets.
• A slower iteration culture. European manufacturing culture is built around large committed runs and long qualification cycles — the opposite of what a startup needs.
• No one has had to build it yet. Flying a prototype to Shenzhen was simply cheaper and easier than fixing this at home. Geopolitics is what's now making it expensive to keep ignoring.

What it would take to build it

A plan that closes this gap runs in three gated phases, not a single leap.

Four levers, moving together

• Geography. Pick one region to concentrate the pilot in. A Dutch/German/Nordic border cluster near existing fabs and ports is a natural first bet.
• Policy. EU and national funds that currently only reach fabs and R&D need a parallel instrument for manufacturing-layer infrastructure.
• Capital. Patient, manufacturing-literate capital willing to underwrite a facility's ramp, not just equity rounds for the startups that would use it.
• Talent and partnership. Recruiting manufacturing engineers and DFM specialists who've actually worked in PRD-style environments, to seed the first facility's expertise.

Open questions and risks

• Does labor cost actually not matter anymore, or is this wishful? Worth stress-testing with real EMS quotes, not assumption.
• Can a single pilot region actually reach density, or does it need critical mass from day one? A small pilot might prove the concept but fail to reproduce the actual network effect.
• Who is the first flagship facility's customer base, concretely? That needs a real pipeline, not an estimate.
• Is this a public-infrastructure play, a company, or both? The levers pull differently depending on the answer, and this plan hasn't picked one yet.
• What kills this? Most likely: a pilot that proves technically sound but can't close the cost gap with Shenzhen for anything beyond the most sourcing-security-sensitive customers.

Next steps

1. Pick the pilot region candidate(s) and pressure-test the "near existing fabs and ports" assumption.
2. Get real EMS and PCB-fab quotes at small-to-medium batch from 3–5 European providers, benchmarked against Shenzhen quotes for the same spec.
3. Decide the operating model question before going further.
4. Talk to 5–10 hardware founders who currently manufacture in Shenzhen about what would actually make them switch.
5. Scope who in Europe — policy, capital, industrial — would need to be in the room for Phase 1 to be fundable.