Nvidia rival tells CNBC it's seeking at least $100 million in funding as AI chip market booms

As the AI chip market surges, European startups focusing on alternatives to Nvidia's graphics processing units (GPUs) are actively seeking substantial investments to scale their operations. Dutch company Euclyd, supported by the former CEO of chip manufacturing giant ASML, is currently negotiating for at least 100 million euros (about $118 million) in funding, according to founder Bernardo Kastrup in an exclusive conversation. In the U.K., Optalysys is also planning to raise over $100 million later this year, while British startup Fractile and France's Arago are reportedly pursuing nine-figure funding rounds. Fractile chose not to comment, and Arago did not respond to inquiries. So far in 2026, over $200 million has already been invested in companies like Axelera from the Netherlands and Olix in the U.K. Nvidia has swiftly ascended to become the world’s most valuable company as its GPUs, originally intended for gaming, are now being utilized for training AI models. However, there is a growing focus on improving the efficiency of AI inference, which is becoming increasingly vital. While Nvidia is also developing semiconductor systems for this purpose, a host of new European startups are emerging, claiming their technology can deliver superior efficiency. "Inference is now dominant, and existing GPU architecture isn't optimized for scale in the ways that matter most," said Patrick Schneider-Sikorsky, director at the NATO Innovation Fund (NIF), which has invested in Fractile. He noted that geopolitical factors, including U.S. export controls and the concentration risk around chipmaker TSMC, are driving capital toward homegrown solutions in Europe. Euclyd is crafting AI chips that promise 100 times greater power efficiency for inference compared to Nvidia's newest Vera Rubin chips. Founded in 2024 by Kastrup, who is advised by ex-ASML CEO Peter Wennink, Euclyd previously raised a seed round of less than 10 million euros and is now seeking additional funding to scale its technology and begin servicing its initial clients. The startup’s chips are designed to replace GPUs but utilize a different architecture that processes data in multiple locations, potentially increasing AI inference efficiency. Kastrup claims that this innovation will significantly reduce the energy consumption, costs, and physical footprint of AI data center infrastructures. However, Euclyd's systems have yet to be validated through large-scale commercial deployment. The company is actively developing a multi-chiplet system aimed for production by 2028 and is currently negotiating with four potential customers, with hopes to begin supply to two next year and the other two the following year. Similarly, Olix, which is working on photonics-based processors for AI, is also targeting its first clients within the next year, although it remains in a research and development phase. Photonic processors leverage light for data movement and computation, and Olix aims to serve various customers, including hyperscalers and government entities. Experts highlight that traditional chip architectures, including GPUs, are reaching their physical limits in miniaturization, complicating efforts to fit more components onto silicon wafers. "The heat generated by current chips is becoming a significant challenge. We believe photonics will lead the next paradigm shift in chip development," Hinrikus added. While Nvidia continues to invest heavily in R&D, with more than $18 billion spent in its latest financial year, European startups face significant challenges. Schneider-Sikorsky emphasized that chip development timelines are lengthy, and the distance from initial design to mass production remains difficult. He also noted that Europe lacks a research funding entity comparable to the U.S. Department of Defense’s DARPA, which supports startups in technology development. As of now, European AI chip startups have raised only $800 million in 2026, a stark contrast to the $4.7 billion amassed by their U.S. counterparts. Despite the hurdles, interest in European startups developing AI inference chips is growing, signaling a shift in investment dynamics. "We are witnessing an uptick in deal flow and discussions with founders in the sector. This is no longer a niche investment; it’s becoming a fundamental aspect of AI infrastructure," noted Carlos Espinal, managing partner at Seedcamp.