Samsung’s investment in Dutch startup Euclyd is a bet on how AI models will be run, not just how they will be built. Euclyd is developing hardware intended to lower the cost and power use of AI inference. At the same time, Samsung is working with Mistral AI to apply customized AI within its own semiconductor operations. Together, the moves put Samsung on both sides of AI deployment—though Euclyd has yet to prove its approach in commercial systems.
6
16
34
Who invested in Euclyd?
Euclyd announced on September 15, 2026, that it had signed a Series A round of more than €200 million, widely reported at approximately $231 million. Samsung co-led it with Somerset Capital Partners, the EQT-managed Scaleup Europe Fund and Innovation Industries. EIFO, imec.xpand, Brabant Development Agency (BOM) and Quadri also participated. The headline figure describes the whole round, not Samsung’s disclosed contribution.
16
6
Founded in 2024 by Bernardo Kastrup and Atul Sinha, the Eindhoven-based company is developing systems for AI inference: the work a trained model does when it processes a request and produces an output. Euclyd proposes a non-GPU approach that brings processor and memory design together, with the aim of reducing data-center energy use and operating costs.
2
5
19
Why Samsung matters beyond the funding
CEO Bernardo Kastrup has said Samsung could help Euclyd in more ways than money. He points to its memory expertise, engineering and systems knowledge, supply-chain experience and network—capabilities relevant to turning a processor-and-memory design into a deployable product.
30
9
19
The competitive target is clear: Nvidia’s GPUs are established AI infrastructure, while Euclyd is pursuing an alternative designed specifically for inference. But a proposed architecture is not evidence of lower costs in production. Euclyd’s planned physical-system rollout starts in 2028, and its ambition to serve thousands of enterprise customers by 2030 remains a target, not an existing customer base.
4
15
26
Euclyd also plans two routes to market: selling chips and physical rack systems to enterprises seeking self-hosted inference, and licensing its technology to companies developing processors. Those plans depend on the company delivering systems that customers can evaluate at scale.
30
1
How the Mistral deal completes the picture
The previous week, Samsung led Mistral AI’s €3 billion Series D and announced a strategic partnership with the French AI company. Samsung says the companies will develop customized, on-premises AI for its semiconductor engineering and manufacturing operations, using Mistral services including Mistral Large. Reported intended applications include data analysis, defect detection and process or equipment optimization.
33
34
35
The two relationships serve different purposes. Euclyd offers Samsung a stake in a potential new way to run AI models; Mistral offers a way to use AI in Samsung’s existing chip business. On-premises deployment is part of the announced Mistral plan, but the available information does not establish an absolute guarantee about where every piece of sensitive chip data will remain. Nor has Euclyd demonstrated that it can outperform established GPU systems at commercial scale. Those outcomes—not the size of either funding round—will determine how consequential the strategy becomes.
16
34
1