AI companies are increasingly spending heavily on specialised hardware as the cost of running larger models pushes power use, memory demand, and infrastructure budgets higher. And this pressure is creating room for chip startups that want to reduce dependence on Nvidia, especially in AI inference, where trained models process requests and generate responses.
Dutch startup Euclyd is entering that race after securing more than €200 million in Series A financing co-led by Samsung Electronics.
The round, announced in mid-September, was also co-led by Somerset Capital Partners, the EQT-managed Scaleup Europe Fund and Innovation Industries. EIFO, imec.xpand, the Brabant Development Agency and Quadri also participated. In addition, Former ASML President and CEO Peter Wennink joined Euclyd as chairman of its board.
What Euclyd Is Building
Founded in 2024 by Bernardo Kastrup and Atul Sinha in Eindhoven Gerard Egelmeers, alongside co-founders Ingolf Held and Harm Peters, Euclyd is developing AI infrastructure focused on inference rather than primarily training large models. Its platform centres on a chip called craftwerk and a data centre system called CWS 32.
Euclyd says the platform combines programmable ASIC computing, a new processor and memory design, as well as system-level optimisation. In simpler terms, the company is trying to redesign how computing and memory work together so AI models can run with less power and at a lower cost. But the architecture is different from the GPU-based systems that have made Nvidia the leading supplier of AI computing hardware.
The company has also described craftwerk as agentic AI silicon and CWS 32 as an exascale AI system, but those performance and efficiency claims have not yet been proven in large-scale commercial deployments.
Why Samsung Matters
Samsung brings more than funding to the deal, as the company is one of the world’s largest memory chip manufacturers, giving it deep experience in a part of AI hardware that Euclyd is trying to redesign.
“Samsung can help us in more ways than money,” Kastrup told CNBC, adding that Samsung also brings engineering expertise, supply chain knowledge, and a large global network.
This support could be important as Euclyd moves from chip design to physical systems, especially as Samsung Senior Vice President Dede Goldschmidt said the next phase of AI will depend on making infrastructure more efficient and easier to scale.
A 2028 Test Against Nvidia
Euclyd plans to begin rolling out physical chip systems in 2028 and aims to serve thousands of enterprise customers by 2030. Its business model includes selling hardware and rack systems to companies that want to run AI inference on their own infrastructure, while also licensing its technology to other chipmakers.
But for now, the company remains at an early stage. Its technology has not been commercially validated at scale, while Nvidia already has a mature hardware and software ecosystem around its GPUs.
The €200 million-plus funding round gives Euclyd capital to expand its engineering team, speed up its silicon and systems roadmap, strengthen partnerships, and prepare for commercial deployment. And the bigger test will come in 2028, when Euclyd is expected to show whether its alternative architecture can deliver the efficiency gains it is promising in real customer environments.
