Dutch semiconductor and AI infrastructure startup Euclyd has raised more than €200 million to build a new kind of AI chip system. The funding round totals $231 million and will support hardware designed for AI inference, the stage where trained models produce answers, predictions, or other results.
Euclyd was founded in 2024, but it is already aiming at a major challenge in the AI industry: the energy and cost of running foundational models at data center scale. The company is developing both the processor and memory architecture for its system, with a design that differs from the GPU architecture used by Nvidia.
The round was co-led by Samsung, Somerset Capital Partners, the Scaleup Europe Fund managed by EQT, and Innovation Industries. Samsung brings more than investment to the project. It is one of the world’s biggest memory manufacturers and has engineering, systems, and supply-chain expertise that could matter as Euclyd moves from chip design toward physical products.
A different approach to AI inference
Euclyd’s system is built around inference rather than a general description of AI computing. The company is designing an AI chip system that includes the processor and memory architecture, creating a complete approach to how foundational models run inside data centers.
That focus matters because AI infrastructure does not end when a model has been trained. Once a model is available, data centers must keep processing requests and producing results, which creates ongoing demands for computing power, memory, energy, and physical infrastructure.
Euclyd says its silicon systems for foundational models will reduce the energy needs and costs of AI data center infrastructure. The company’s architecture differs from GPUs, giving it a separate path as AI chip makers compete to support the expanding use of foundational models.
Bernardo Kastrup, Euclyd’s CEO, described the infrastructure challenge in direct terms: “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.”
From funding to physical systems
Euclyd plans to begin rolling out its physical chip systems in 2028. That timeline puts product delivery ahead of the company’s larger customer goal: serving thousands of enterprise customers by 2030.
The plan gives Euclyd four years between its expected physical rollout and its 2030 enterprise target. The company will need its processor and memory architecture to move from design into systems that can support enterprise use, while its investors provide knowledge and connections across the semiconductor supply chain.
Samsung’s role may be especially relevant to that transition. Kastrup said, “Samsung can help us in more ways than money. 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.”
Euclyd is entering a market where Nvidia remains a competitor in AI chips, while Google, AWS, and Meta are developing their own chips. OpenAI also announced its first AI chip, the Jalapeño, in August, though the year was not specified in the available details.
That group of companies shows how many parts of the AI industry are working on custom hardware. Euclyd’s stated difference is its architecture for inference, including the relationship between its processor and memory system, along with its goal of reducing energy needs and infrastructure costs.
What comes next for Euclyd
The company’s immediate challenge is turning its funding into a working physical system. Euclyd has set 2028 for the start of its rollout, followed by a target of thousands of enterprise customers by 2030.
Those dates also give the startup a clear path from financing to deployment. The funding announced on September 15, 2026, at 5:05 AM EDT, after an announcement dated September 14, 2026, at 7:30 PM EDT, will support that path as Euclyd develops its silicon systems for foundational models.
For enterprise customers, the appeal will come down to the results Euclyd promises: lower energy needs and lower costs for AI data center infrastructure. For the wider chip market, the company represents another attempt to build hardware around the demands of AI inference rather than relying only on existing GPU designs.
Euclyd now has the capital, a group of co-lead investors, and a stated architecture for its next stage. Its next major test will arrive in 2028, when the Dutch startup aims to begin rolling out its physical chip systems.
Based on
