Euclyd Secures $231 Million Series A with Samsung Backing to Build Energy-Efficient AI Chips

Euclyd Secures Major Series A Funding Round

Euclyd, a Netherlands-based startup founded in 2024 by Bernardo Kastrup and Atul Sinha at High Tech Campus Eindhoven, has raised more than €200 million, or approximately $231 million, in Series A funding1

1

2

2

. Samsung Electronics, Somerset Capital Partners, and the Scaleup Europe Fund led the round, with participation from several institutional backers including imec.xpand, a fund associated with the Imec nanotechnology research center2

2

. The funding round also brought Peter Wennink, former CEO of ASML Holdings, onto Euclyd's board, signaling strong industry confidence in the company's vision2

2

.

Building AI Chips to Reduce Energy Consumption for AI

Euclyd is developing chip systems specifically designed to reduce the energy consumption and cost of running AI models, addressing one of the most pressing challenges in the AI hardware ecosystem1

1

. The company's roadmap targets AI inference—the stage at which a trained model responds to queries—rather than the more power-intensive model training workload1

1

. This focus on the inference stage positions Euclyd to serve the rapidly growing market for AI agents and real-time AI applications that require efficient, cost-effective computing infrastructure.

Introducing craftwerk: An AI-Specific Chip with Innovative Architecture

At the heart of Euclyd's technology is craftwerk, an AI-specific chip that will be shipped as part of a system called CWS, expected to power AI clusters with more than an exaflop of performance2

2

. The craftwerk computing module will be an application-specific integrated circuit, or ASIC design, featuring what Euclyd describes as an "innovative memory architecture"2

2

. The company plans to pursue processor-memory co-design, indicating that its RAM architecture will be specifically optimized to work with its ASIC, a strategy that could deliver significant performance and efficiency gains1

1

2

2

.

Tackling the Memory Wall Challenge

Euclyd's planned hardware combines programmable compute, processor-memory co-design, and system-level optimization to address critical bottlenecks in AI processing1

1

. The company specifically aims to tackle the memory wall challenge, an engineering obstacle that complicates efforts to speed up AI chips by limiting the rate at which data moves between a GPU's computing and memory modules2

2

. Multiple custom memory designs have emerged in the AI hardware ecosystem over recent years as alternatives to standard HBM memory. Nvidia recently debuted NVHBM, which can free up 30% of a GPU's surface area, while startup d-Matrix has developed SRAM-based memory architecture for inference calculations2

2

.

Samsung Backing Strengthens Technical Capabilities

The Samsung backing in this Series A funding round could prove particularly valuable for Euclyd's engineering efforts2

2

. Samsung Electronics is one of the world's top HBM suppliers and recently debuted technology that makes it possible to place HBM memory directly atop a GPU's computing circuits, rather than implementing memory dies and computing modules side-by-side2

2

. This partnership positions Euclyd to leverage cutting-edge memory technologies as it develops craftwerk and CWS.

Dual Business Model Targets Multiple Markets

On the commercial side, Euclyd plans to pursue two distinct lines of business: direct sales of rack-based hardware to enterprises running AI inference on their own infrastructure, and licensing arrangements with chip developers who want to build on Euclyd's underlying technology1

1

. According to CNBC, Euclyd plans to launch its silicon in 2028 and hopes to gain thousands of enterprise customers by 20302

2

. The company envisions a future in which advanced AI is no longer constrained by infrastructure cost, power availability, or geography2

2

. Watch for how Euclyd's exaflop-scale performance claims translate into real-world deployments and whether its dual business model can effectively compete against established GPU manufacturers and emerging AI chip startups in an increasingly crowded market.