Dutch Semiconductor Startup EUCLYD Secures Over 200 Million Euros to Build Advanced AI Chips
Eindhoven, Wednesday, 16 September 2026.
Eindhoven-based EUCLYD raised over €200 million in Series A funding to develop efficient AI chips, appointing former ASML Chief Executive Officer Peter Wennink as its new chairman.
A Strategic Leap in AI Semiconductor Infrastructure
The global artificial intelligence landscape is undergoing a critical transition where software capabilities are outstripping the physical limits of hardware [2][4]. Addressing this bottleneck, Dutch semiconductor startup EUCLYD has secured a landmark Series A funding round of more than €200 million to build next-generation hardware designed specifically for foundation AI workloads [2][4]. Founded in 2024 by Bernardo Kastrup and Atul Sinha, the company is headquartered at the High Tech Campus Eindhoven in the Netherlands [2][4]. This massive injection of capital represents one of the largest Series A funding rounds ever raised by a European technology firm, positioning the Eindhoven-based startup as a formidable challenger in the global AI computing market [2][6].
Industry Alignment and Key Backers
This milestone transaction operates firmly within the semiconductor industry, specifically targeting specialized AI silicon and custom processor architectures [2][4]. The funding round, announced on September 15, 2026, was co-led by major industry players and venture firms, including Samsung Electronics, Somerset Capital Partners, the Scaleup Europe Fund (managed by EQT), and Innovation Industries [4][7]. Additional backing came from a strong consortium of deep-tech investors, including the Export and Investment Fund of Denmark (EIFO), imec.xpand, the Brabant Development Agency (BOM), and Quadri [2][4]. In a major validation of EUCLYD’s technical vision, Peter Wennink, the former President and Chief Executive Officer of photolithography giant ASML, has formally joined the company as Chairman of the Board [2][4].
Rethinking the Silicon Stack: How the Technology Works
To overcome the physical limits of current AI hardware, EUCLYD is shifting away from general-purpose graphics processing units (GPUs) toward highly specialized, inference-specific architectures [5][6]. The startup’s technical approach relies on processor-memory co-design and programmable Application-Specific Integrated Circuit (ASIC) compute [4][5]. Rather than optimizing isolated components, EUCLYD is rethinking the entire computing stack from the logic gate up, designing how processors, memory, and datacenter systems work in tandem to eliminate data transfer bottlenecks [3][6]. This architecture is realized through two core offerings: “craftwerk,” which is an agentic AI silicon platform, and “craftwerk station CWS,” designed as an exascale AI factory [2][4].
Overcoming the AI Efficiency Wall
The primary benefits of EUCLYD’s innovation center on solving the severe power, cost, and memory bandwidth constraints that currently plague AI datacenters [2][4]. Traditional GPU stacks consume vast amounts of electricity and require complex cooling systems, limiting where and how AI can be deployed [3][5]. By utilizing a dedicated processor-memory co-design, EUCLYD’s platform is engineered to deliver dramatically lower power consumption and a reduced physical footprint [5][6]. For enterprises and hyperscalers, this translates directly to a lower cost per token, making the execution of advanced AI models economically sustainable and geographically flexible [2][4]. Ultimately, this specialized architecture aims to democratize access to advanced AI by removing the infrastructure barriers to what the company terms “Abundant Intelligence” [2][4].
Scaling Up for Global Commercialization
With the newly acquired €200 million, EUCLYD plans to rapidly expand its engineering team across the full compute stack, from silicon design and physical implementation to compilers, firmware, and runtime software [3][4]. The company is actively preparing its technology for commercial deployment, targeting enterprise, sovereign, and hyperscale AI markets [4]. While the immediate focus remains on accelerating its silicon and systems roadmaps, the company’s long-term commercial timeline is already taking shape, with plans to ship its first physical systems to customers in 2028 [2][5]. This strategic trajectory highlights Europe’s growing capacity to leverage its deep engineering and semiconductor heritage to build globally competitive AI infrastructure [4].