EUCLYD Raises More Than €200 Million for Non-GPU AI Inference Systems

EUCLYD has raised more than €200 million for a bet on a different way to run powerful AI models. The Eindhoven-based startup is developing chips, memory architecture and data-center systems as one platform for AI inference, with Samsung among the investors backing the effort.

The Series A was co-led by Samsung, Somerset Capital Partners, EQT-managed Scaleup Europe Fund and Innovation Industries. EIFO, imec.xpand, Brabant Development Agency and Quadri also participated. EUCLYD also named former ASML president and chief executive Peter Wennink as chairman of its board.

A design aimed at inference, not GPU imitation

Inference is the task of generating an output from an AI model. EUCLYD is targeting that task with a system whose processor and memory architecture differs from GPUs, CNBC reported. The company’s approach is to design those components alongside the data-center system, rather than present a processor as a standalone product.

EUCLYD calls its chip architecture craftwerk and its planned data-center system craftwerk station CWS. It says the platform is intended to lower the energy and cost of AI infrastructure, but those outcomes are company aims rather than independently established results.

Capital for the system and its route to market

• Expand EUCLYD’s engineering organization and accelerate its silicon and systems roadmap.
• Strengthen ecosystem partnerships and prepare for commercial deployment across enterprise, sovereign and hyperscale AI markets.
• Sell hardware and physical racks for self-hosted enterprise inference, while licensing intellectual property to other chip developers.

Samsung’s role may extend beyond capital. Chief executive Bernardo Kastrup told CNBC that Samsung could offer memory-manufacturing, engineering, systems and supply-chain expertise—capabilities relevant to EUCLYD’s attempt to integrate several hardware layers.

The commercial test remains ahead

But EUCLYD’s systems have not been proven at commercial scale. Kastrup told CNBC the company aims to begin rolling out physical chip systems in 2028 and serve thousands of enterprise customers by 2030. Those targets make the first customer deployments the key test of whether its integrated design can meet its stated economic goals.