Samsung is betting on a two-year-old Dutch startup to chip away at Nvidia’s grip on AI silicon. As reported by SamMobile, Samsung co-led a $230 million funding round for Euclyd, a company designing AI inference chips that deliberately move away from the GPU-heavy architecture that Nvidia has built its empire on. The exact size of Samsung’s check hasn’t been disclosed, but co-leading a round of this size says plenty on its own.
Euclyd is building inference chips with a different approach to both processor and memory architecture. That’s a meaningful distinction. Nvidia’s dominance is largely a training story, but inference, running models after they’ve already been trained, is where the real volume and recurring cost lives for most AI deployments. That’s the market Euclyd is targeting, and it’s a smart place to plant a flag. The other co-investors include Somerset Capital Partners, the Scaleup Europe Fund, and Innovation Industries.
Why does Samsung care? Because it currently sits in an awkward position in the AI chip supply chain. It supplies critical memory components, including HBM, to companies like Nvidia, but that puts Samsung in a supplier role rather than a strategic one. Investing in companies that could rival Nvidia gives Samsung a seat at a different table. Euclyd CEO Bernardo Kastrup has been direct about this, saying Samsung brings more than capital. Their engineering depth, supply chain knowledge, and industry network are all things a two-year-old chip startup can’t easily buy elsewhere.
The broader context matters here. Hyperscalers are actively looking for alternatives to Nvidia hardware, partly for cost reasons and partly to reduce dependency on a single supplier. Google, Amazon AWS, Meta, and OpenAI are all developing their own inference silicon. Euclyd is positioning itself as an option for companies that want a third-party alternative without building entirely in-house. That’s a real gap in the market, and $230 million gives the startup a credible runway to actually fill it.
Still, skepticism is warranted. The chip industry has a long history of well-funded Nvidia challengers that never reached meaningful scale. Competing on architecture alone isn’t enough. Software ecosystems, developer tooling, and production reliability are where challengers historically lose ground. Euclyd is early, and the distance between a funded startup and a chip that ships at scale is significant. But Samsung’s involvement adds a supply chain dimension that most pure-play challengers haven’t had. That’s not nothing.
