European semiconductor startups focused on developing alternative technologies to NVIDIA's graphics processing units (GPUs) are planning significant funding rounds to scale their operations, driven by the artificial intelligence boom.

Bernardo Kastrup, founder of the Dutch company Euclyd, stated in an interview that the firm is negotiating a new funding round of at least €100 million (approximately $118 million). The company has backing from the former CEO of chip manufacturing equipment giant ASML.

Additionally, UK-based startup Optalysys aims to secure over $100 million in funding later this year. Reports indicate that Britain's Fractile and France's Arago are also preparing funding rounds in the nine-figure range. Fractile declined to comment on the matter, while Arago did not respond to requests for comment. Since the beginning of 2026, investors have injected more than $200 million into Dutch firm Axelera and UK-based Olix.

NVIDIA, whose GPUs were initially designed for gaming but are now widely used for training AI models, has rapidly become one of the world's most valuable companies. Market attention is now shifting towards the most efficient methods for running these models, specifically AI inference.

Although NVIDIA is also developing semiconductor systems for this purpose, a wave of new European startups has emerged, claiming their technologies can achieve higher efficiency.

Patrick Schneider-Sikorsky, head of the NATO Innovation Fund (NIF), which has invested in Fractile, commented, "Inference now dominates the landscape, and current GPU architectures were not designed for key aspects of large-scale deployment." He added, "Geopolitical tailwinds are evident: US export controls, risks associated with industry concentration around TSMC, and Europe's practical need for computational autonomy are driving capital towards local chip firms."

Euclyd is developing AI chips with systems that are reportedly 100 times more energy-efficient than NVIDIA's latest Vera Rubin chips for inference tasks. NVIDIA did not respond to requests for comment.

The Dutch startup was founded in 2024 by former ASML executive Kastrup, with former ASML CEO Peter Wennink serving as an advisor and investor. Having completed a seed funding round of under €10 million, the company is now seeking new capital to expand its technological deployment and begin supplying its first customers.

Kastrup explained that Euclyd is building chip systems to replace GPUs but with a different architecture. While GPUs expend time and energy transferring data between memory stacks, Euclyd's chips process data in multiple locations, which he claims will enhance AI inference efficiency.

He further noted that the company's chip systems for foundational models will reduce the energy consumption, cost, and physical footprint of AI data center infrastructure. However, unlike NVIDIA's chips, Euclyd's systems have not yet been validated through large-scale deployment with commercial partners.

Euclyd is advancing its research and development. The firm has already developed a chip for AI inference and is currently working on a multi-chiplet system that will be faster than existing versions, with mass production targeted for 2028. Kastrup mentioned that the company is in talks with four potential customers, with two expected to begin receiving supplies next year and the other two the following year.

Taavet Hinrikus, partner at investment firm Plural, told CNBC that Olix, which is developing AI processors based on photonic technology, plans to onboard its first customers next year, though it remains in the R&D phase.

Photonics processors use light to transmit data and, in some cases, perform computations.

Hinrikus stated that the startup will serve all clients requiring AI inference, including hyperscale cloud providers and governments. Olix did not respond to requests for comment.

Hinrikus pointed out that the electronic architecture of existing chips, including GPUs, is "approaching its limits" in terms of process scaling. Chip manufacturers are striving to shrink processor sizes to integrate more components on a wafer, thereby improving the economics of system operation.

"Heat generation from current chips is becoming a major challenge. We strongly believe that photonic platforms will define the next technological paradigm," he added.

NVIDIA is also working aggressively to maintain its industry leadership. In its most recent full fiscal year ending January 2026, the chip giant invested over $18 billion in research and development. In December of last year, the company acquired assets from AI inference startup Groq for $20 billion, and in March, it announced a $4 billion investment in two companies developing photonic technology.

European startups still face significant challenges.

Schneider-Sikorsky of the NATO Innovation Fund noted, "Chip development cycles are long, the journey from tape-out to mass production is arduous, and Europe's foundry ecosystem still requires maturation."

Fabrizio Del Maffeo, CEO of Axelera, highlighted that European governments remain cautious about investing in startup products, and Europe lacks an equivalent to the US Defense Advanced Research Projects Agency (DARPA), which provides funding for startups and other technology initiatives.

He added that Europe also lacks mechanisms encouraging the procurement of local products, and fragmented labor laws across countries complicate the recruitment of local talent.

According to market data platform Dealroom, European AI chip startups continue to lag in funding: they have raised $800 million so far in 2026, compared to $4.7 billion raised by their US counterparts.

In the US, Cerebras Systems raised $1 billion in February, while MatX, Ayar Labs, and Etched each completed $500 million funding rounds this year.

Nevertheless, European startups developing chips for AI inference to compete with NVIDIA are attracting increasing investor attention.

Carlos Espinal, Managing Partner at Seedcamp, an investor in chip startup Vaire Computing, remarked, "We observe this trend in our deal flow and conversations with founders in the sector. This is no longer a niche bet; it is becoming a central focus for those building AI infrastructure."