▲ Samsung Logo
Samsung has reportedly participated as a co-investor in "Euclidean," an AI semiconductor design company with an architecture entirely different from Nvidia's graphics processing units (GPUs).
In an interview with U.S. business channel CNBC on the 15th (local time), Bernardo Castrup, CEO of Euclidean, stated that Samsung was included as a co-investor in a Series A funding round worth $231 million (approximately 315 billion won), which also saw participation from Somerset Capital Partners, Innovation Industries, and the Scaleup Europe Fund.
Established in 2024, Euclidean is a Dutch company that designs AI semiconductor systems for inference based on an architecture completely distinct from conventional GPUs.
While Nvidia has dominated the advanced AI semiconductor market by utilizing GPUs originally designed for gaming for AI training and inference, big tech companies and startups have recently been jumping into the development of their own proprietary AI chips one after another.
OpenAI announced its "Jalapeno" AI chip last month, boasting speed and efficiency set to lead the industry, while Google, Amazon Web Services (AWS), and Meta are also accelerating the upgrades of their custom AI semiconductors.
Although Euclidean has yet to undergo large-scale validation, it claimed that the silicon system for its foundation model can dramatically reduce AI data center infrastructure costs and power demand.
CEO Castrup noted, "Samsung can offer us much more than just financial support," adding, "As one of the world's largest memory semiconductor manufacturers, they possess exceptional engineering expertise, a deep understanding of systems, a supply chain, and a vast global network."
Castrup added that Euclidean aims to release its physical semiconductor system in 2028 and provide services to thousands of companies by 2030.
※
Copying, redistribution, and unauthorized use in AI training are strictly prohibited.