This article takes a close look at Europe’s growing push into AI inference hardware. It highlights startups working on new silicon, photonics, and multi-chiplet designs, all hoping to challenge Nvidia’s lead. You’ll find out who’s raising money, what tech they’re betting on, and the big-picture forces—geopolitical and market—that are fueling European investment in homegrown AI chips. The piece also touches on the hurdles these startups face.
Europe’s AI chip startups set sights on challenging Nvidia
Across the Netherlands, the UK, and continental Europe, founders and investors are rallying around the idea that Europe can build serious AI inference processors. Funding rounds are getting bigger, with several teams hoping to move fast and deliver production-ready hardware in just a few years.
Everyone’s talking about power efficiency, data movement, and the hope of breaking up a supply chain that’s gotten a little too concentrated on just a few players.
Key players and their technological bets
European challengers aren’t sticking to one approach. Some are going for chiplets, others for photonics, all trying to get ahead of current AI inference workloads. Here’s who’s making moves:
- Euclyd (Netherlands) — backed by ex-ASML leadership and Peter Wennink as advisor. They’re negotiating a round “in excess of €100 million” after a much smaller seed, and claim their inference architecture could be up to 100x more power-efficient than Nvidia’s Vera Rubin chips. Euclyd has built an inference chip and is working on a multi-chiplet system they hope to have ready for production by 2028. They’re in talks with four potential customers and planning first deliveries in the next two years.
- Optalysys (UK) — lining up a $100 million+ fundraising round later this year to ramp up optically accelerated AI processing.
- Fractile (UK) — reportedly chasing a nine-figure funding round to push their hardware roadmap forward.
- Arago (France) — also said to be seeking a nine-figure round to advance its platform.
Investors have already put more than $200 million into EU chip companies Axelera and Olix in 2026. Olix is working on photonics-based processors that use light to move data, hoping to cut down bottlenecks and maybe even unlock new ways to compute.
Why Europe now? The drivers and the funding trajectory
Backers and policymakers point to a mix of geopolitical and supply-chain factors shaping capital flows into European silicon:
- Geopolitical resilience — Europe wants to avoid relying on a single upstream supplier for AI silicon and build up its own compute capabilities.
- Supply-chain risks — There’s real worry about export controls and depending on wafer fabrication that’s concentrated in just a few regions.
- Foundry maturity gaps — Critics say Europe doesn’t yet have the scale or reliability of U.S. or Taiwanese chip ecosystems, which makes it tough to speed up hardware development.
- Public procurement and policy support — Compared to U.S. programs like DARPA, Europe’s government backing and procurement systems still lag in several key areas.
Nvidia’s dominance and the European challenge
Nvidia still sets the pace for AI inference. The company is pouring huge sums into R&D—reportedly over $18 billion in its latest fiscal year—and just picked up Groq’s assets for around $20 billion. They’re investing in photonics and related tech too.
This sheer scale makes things tough for new players who want to compete on performance and cost per inference. Still, investors say the mood’s shifting: European inference hardware isn’t just a niche thing anymore—it’s quickly becoming a real part of AI infrastructure investment.
Outlook and risks for European AI chip startups
There’s definitely a surge in interest, but plenty of risks still hang over the scene. European challengers face long development cycles and an immature foundry ecosystem.
The talent market feels fragmented, and public procurement tools aren’t exactly robust. To scale up, these startups need more than just clever architectures—they’ve got to nail their go-to-market plans, secure reliable fabrication partners, and actually win over customers.
In the near term, more funding rounds seem almost inevitable for European inference hardware. Power efficiency is a hot topic, as is data movement, and photonics-based approaches are starting to catch some attention.
Nvidia still dominates, no question about that. But honestly, it looks like Europe’s investors and engineers are hoping—maybe even believing—that the continent can carve out its own space with competitive, sovereign silicon for AI workloads.
Here is the source article for this story: Nvidia rival tells CNBC it’s seeking at least $100 million in funding as European AI chip market booms