European chip startup develops alternative technology Nvidia’s Graphics processing units (GPUs) are attracting large funding rounds as they look to gain scale amid the AI boom.
Dutch company Euclid backed by former CEO of semiconductor manufacturing equipment giant ASMLis currently in talks with investors for a round of at least 100 million euros ($118 million), founder Bernardo Kastrup told CNBC in an exclusive interview.
Elsewhere, British startup Optalysys plans to raise more than $100 million later this year, while British companies Fractile and France’s Arago are reportedly raising nine-figure sums. Mr. Fractile declined to comment, and Mr. Arago did not respond to requests for comment. So far in 2026, investors have already poured more than $200 million into the Netherlands’ Axela and the UK’s Oryx.
Nvidia quickly became the world’s most valuable company as GPUs originally designed for gaming were repurposed to train AI models, but the focus is now on how to most efficiently use these models, known as AI inference.
U.S. semiconductor giants are also developing semiconductor systems for that purpose, but a growing number of European startups are emerging that say the technology they’re building can do it more efficiently.
Patrick Schneider-Sikorsky, director of the Nato Innovation Fund (NIF), which is investing in Fractile, told CNBC that “inference is mainstream right now, and existing GPU architectures are not built for inference in the way that matters most at scale.”
“There are clear geopolitical tailwinds, such as U.S. export restrictions and concentration risks (for semiconductor manufacturers).” TSMC And all true European sovereign computing imperatives are directing capital towards domestic silicon. ”
ASML Alumni
Euclyd is developing AI chips that run on systems that are 100 times more power efficient for inference compared to Nvidia’s latest generation Vera Rubin chips. Nvidia did not respond to a request for comment from CNBC.
Founded in 2024 by former ASML director Kastrup and counting former ASML CEO Peter Wennink as an advisor and investor, the Dutch startup has already raised a seed round of less than €10 million and is currently looking for new funding to scale up its technology and start supplying its first customers.
Euclyd is building a chip system to replace GPUs, but the architecture is different, Kastrup said. Unlike GPUs, which spend time and energy moving data through the memory stack, Euclyd’s chip processes data in multiple locations, making AI inference more efficient, Kastrup said.
The company’s foundational model silicon system reduces the energy, cost and footprint of AI data center infrastructure, it added. But unlike Nvidia’s chip, Euclyd’s system has not yet been proven in large-scale deployments by commercial partners.
Euclyd’s prototype system. Credit: Euclyd.
Euclyd is working on it. The company has already developed chips for AI inference and is currently developing a multi-chiplet system that will process faster than the current iteration of its product, aiming for production by 2028. Kastrup said he is in talks with four potential customers, two of whom he hopes to begin supplying next year and two the year after.
Olix, which is developing photonics-based processors for AI, is also currently in the research and development stage, but is targeting early customers next year, Taavet Hinrikus, a partner at Plural, which has invested in the company, told CNBC.
Photonic processors are chip systems that use light to move data and sometimes perform calculations.
Hinrichs said the startup will target any customer that needs inference services, including hyperscalers and governments. Orix did not respond to a request for comment.
The electronic architecture of chips, including GPUs, is truly “at its limits” in terms of how small they can be, Hinrichs said. Chip manufacturers are trying to shrink processors to fit more components on a wafer and improve the economics of running systems on the wafer.
“The heat generated by[current chips]is a big issue. We strongly believe that photonics platforms will be the next paradigm,” he added.
Nvidia is also working hard to stay ahead of the pack. The semiconductor giant spent more than $18 billion on research and development in its most recent fiscal year ending January 2026. In December, it acquired assets from AI inference startup Groq for $20 billion, and in March it announced a $4 billion investment in two companies developing photonics technology.
Challenges remain for European startups
European startups face hurdles.
“Chip development timescales are long and the path from tape-out to volume production deployment is challenging. The European foundry ecosystem still needs to mature,” said NIF’s Schneider Sikorsky.
Axela CEO Fabrizio Del Maffeo told CNBC that European governments remain “conservative” in investing in startups’ products and do not have an equivalent to DARPA, the U.S. Department of Defense research agency that funds startups and other technology projects.
He added that Europe lacks mechanisms to encourage the consumption of locally manufactured products, and fragmented labor laws across borders make it difficult to recruit European talent.
European AI chip startups have been slow to raise funding, with $800 million raised by 2026, compared to $4.7 billion for U.S. startups, according to Deal Room.
In the US, Cerebras Systems raised $1 billion in February, and this year saw $500 million in rounds for MatX, Ayar Labs, and Etched.
Nevertheless, a European startup developing AI inference chips to rival Nvidia’s is attracting increasing interest from investors.
“We’re seeing it in the deal flow and the conversations we’re having with founders in this space,” Carlos Espinal, managing partner at Seedcamp, which backs semiconductor startup Vere Computing, told CNBC. “This is no longer a niche bet; it’s becoming a core part of how people think about AI infrastructure.”
