Why Samsung Is Betting $231 Million on Dutch AI Inference Chip Startup Euclyd to Directly Rival Nvidia

EINDHOVEN, Netherlands — Samsung Electronics has co-led a $231 million funding round for Euclyd, a two-year-old Dutch startup building specialized chips for AI inference, marking the South Korean electronics giant's latest move to establish a foothold in the fast-growing market for alternatives to Nvidia's dominant AI processors. Here is why Samsung made the bet, based on public statements from the companies involved.

Euclyd, founded in 2024 by Bernardo Kastrup and Atul Sinha at High Tech Campus Eindhoven, raised more than 200 million euros in its Series A round, with the financing co-led by Samsung alongside Somerset Capital Partners, the Scaleup Europe Fund managed by EQT, and Innovation Industries. The exact size of Samsung's individual contribution has not been disclosed.

The first and most fundamental reason behind Samsung's interest is where Euclyd has chosen to compete. Rather than building chips for the resource-intensive process of training AI models, where Nvidia's dominance is closest to a monopoly, Euclyd is focused specifically on inference, the stage at which an already-trained AI model responds to live user queries. Nvidia's share of the inference chip segment has been estimated at between 60% and 75% as of 2025, a commanding position but one industry analysts view as considerably more contestable than Nvidia's grip on training hardware, giving challengers like Euclyd a more realistic opening.

A second reason ties directly to Samsung's own core business. Euclyd's chip architecture is built around what the company describes as processor-memory co-design, an approach that treats memory bandwidth and system-level architecture as just as central to performance as the processor itself. Because Samsung is one of the world's largest memory chip manufacturers, an inference architecture built around tighter memory integration plays directly to Samsung's existing manufacturing and engineering strengths, rather than requiring the company to compete purely on general-purpose processor design against far more established rivals.

Euclyd's own founder pointed to that alignment as a central part of what makes Samsung's involvement valuable beyond the capital itself. "Samsung can help us in more ways than money," Kastrup said. "They are one of the biggest memory manufacturers in the world. They do a lot of engineering, they know a lot about systems, they know the supply chain, they have a huge network." That combination of manufacturing scale, systems engineering expertise and established supply chain relationships gives Euclyd resources a typical venture-backed startup would struggle to access on its own, while giving Samsung early visibility and influence over an architecture built specifically to leverage its own memory technology.

A third reason reflects broader industry dynamics around the rising cost of AI infrastructure. Kastrup has framed Euclyd's mission explicitly around addressing those cost and energy pressures. "AI is becoming a foundation of economic growth, scientific discovery, and national competitiveness, but its potential will remain constrained unless we fundamentally change the infrastructure beneath it," Kastrup said. For Samsung, backing a company aimed at reducing the energy consumption and cost of running AI models aligns with the broader industry push to make large-scale AI deployment more economically sustainable, a trend that, if successful, could expand overall demand for the memory and systems components Samsung supplies regardless of whether Euclyd's own chips achieve widespread adoption.

A fourth factor is diversification against overreliance on any single AI hardware supplier. Samsung's investment comes amid a broader wave of spending by major technology companies on alternatives to Nvidia's GPUs, with Google, Amazon Web Services, Meta and OpenAI all separately developing custom AI silicon. OpenAI disclosed its first custom chip, code-named Jalapeño, in August, claiming industry-leading efficiency gains over general-purpose GPUs. By backing Euclyd, Samsung positions itself on the investor side of that broader shift away from GPU dependence, rather than solely as a component supplier reacting to decisions made by chip designers elsewhere.

A fifth reason involves access to Europe's semiconductor ecosystem and talent base. Euclyd's board is chaired by Peter Wennink, the former president and chief executive officer of Dutch lithography equipment maker ASML, whose machines are essential to advanced chip manufacturing worldwide. Wennink has said Europe's underlying semiconductor capabilities position Euclyd to compete on a global scale, a statement that reflects the deep pool of chip engineering expertise concentrated around the Eindhoven region, home to ASML and a dense cluster of semiconductor suppliers and research institutions.

A sixth reason relates to timing and stage of investment. Backing Euclyd at the Series A stage gives Samsung an early position, and likely board-level influence, in a company still years away from commercial deployment, rather than competing to invest at a later, more expensive stage once the startup's technology and market fit are more proven. Euclyd has said it aims to deploy its first physical AI chip systems in 2028 and to serve thousands of enterprise customers by 2030, giving Samsung a multi-year runway to help shape the company's direction as it approaches commercialization.

A seventh reason concerns Euclyd's dual business model, which combines direct hardware and rack system sales to enterprise customers seeking on-premises AI inference capacity with a separate revenue stream from licensing its underlying chip architecture to other manufacturers. That licensing path offers Samsung a potential future avenue to incorporate Euclyd's inference architecture into its own broader product and manufacturing ecosystem, beyond the direct equity stake itself.

An eighth reason is straightforward market growth. As AI adoption shifts from an initial phase dominated by model training toward sustained, large-scale deployment, inference workloads are expected to represent a rapidly expanding share of total AI compute demand over the coming years, giving early investors in credible inference-focused hardware companies exposure to that anticipated shift.

A ninth reason is risk-sharing. By co-investing alongside established venture and growth-equity firms including Somerset Capital Partners, EQT's Scaleup Europe Fund and Innovation Industries, Samsung shares both the financial risk and the due-diligence work involved in backing an unproven, pre-commercial hardware startup, rather than bearing that risk entirely on its own balance sheet.

A tenth and final reason is optionality. Euclyd remains years from any commercial deployment, and Kastrup himself has acknowledged the company's technology is unproven at scale. For Samsung, the investment functions as a relatively contained financial bet on a plausible challenger to Nvidia's dominance in a segment of the AI chip market Samsung has a direct interest in seeing diversify, without requiring Samsung to abandon or compete directly against its own existing chip and memory manufacturing relationships with other AI hardware developers in the meantime.