The Dutch startup Euclyd is seeking at least €100 million ($118 million) in funding to scale up its development of AI chips, which it claims will offer 100 times higher power efficiency for AI inference compared to Nvidia’s latest products. Euclyd, founded in 2024 by former ASML director Bernardo Kastrup, is backed by ASML’s former CEO, Peter Wennink. As European startups increasingly pursue alternative chip designs to improve AI workloads, the challenges they face include a lack of local investment structures and a less developed semiconductor manufacturing sector compared to their U.S. counterparts, which have raised substantially more funding.

Euclyd: Euclyd is a Dutch startup based in Eindhoven developing ultra-efficient silicon systems and chips for AI inference in foundation models, using a novel architecture that processes data across multiple locations to minimize energy use. Backed by former ASML CEO Peter Wennink as advisor and investor, the company has unveiled its initial inference chip and is advancing multi-chiplet designs for data center deployment. In the news, Euclyd is seeking major funding to scale production and begin supplying customers as Europe’s AI chip sector heats up.
Nvidia: Nvidia is the leading designer of graphics processing units repurposed as the backbone for AI model training and increasingly inference workloads globally. The company is pushing advancements in inference-optimized hardware, software, and complementary technologies like photonics following its GTC conference emphasis on the inference era. European inference-focused startups like Euclyd claim architectural advantages over Nvidia’s GPUs for large-scale efficiency.
Bernardo Kastrup: Bernardo Kastrup is the founder of Euclyd and a former director at ASML, leveraging semiconductor industry experience to innovate in AI hardware. He oversees development of Euclyd’s power-efficient inference chips targeting data center infrastructure. In a recent CNBC interview, he discussed the startup’s funding discussions and differentiation from Nvidia’s architecture.
Patrick Schneider-Sikorsky: Patrick Schneider-Sikorsky serves as director at the NATO Innovation Fund, a multi-sovereign venture fund investing in dual-use technologies such as AI chips and defense tech. The fund supports European startups like Fractile developing inference solutions. He recently noted the shift to inference dominance and geopolitical drivers boosting investment in homegrown European silicon.

`json
{
“Inference Shift”: “AI workloads are increasingly focused on inference rather than training, demanding architectures optimized for efficiency at scale.”,
“European Innovation”: “Startups across Europe are developing alternative chip designs such as photonics and custom silicon to improve AI inference efficiency beyond traditional GPU capabilities.”,
“Geopolitical Tailwinds”: “U.S. export restrictions and the push for European sovereign compute capabilities are driving investor interest in European AI chip initiatives.”
}
`