Nvidia’s European rival plans to raise R$588 million with a focus on energy efficiency for AI

Dutch startup Euclyd is negotiating with investors to raise funds of at least €100 million, equivalent to around US$118 million or R$588.8 million. The company, founded in 2024 and headquartered at Eindhoven, develops chips aimed at artificial intelligence inference. The founder Bernardo Kastrup confirmed the conversations in an interview. The round seeks resources to scale the technology and begin commercial sales.
Outras European startups are also seeking significant contributions in this segment. The British Optalysys plans to raise more than US$100 million in 2026. The companies Fractile, from Reino Unido, and Arago, from França, would be in discussions for rounds in the nine-figure range.
Avanço from European startups in the semiconductor sector for AI
Startups from the continent accelerates efforts to compete for space in the artificial intelligence infrastructure market. The main focus is on inference, a stage in which already trained models perform tasks with high efficiency. Empresas and Euclyd claim to offer solutions that overcome limitations of traditional GPUs in power consumption and operational cost.
Investidores allocated more than US$200 million in 2026 alone to the Dutch Axelera and the British Olix. The move comes as the global AI chip market records strong expansion. The architecture of conventional GPUs, according to experts, was not optimized for large-scale inference. Export Controles and the quest for computational sovereignty in Europa direct capital to local solutions.
- Euclyd negotiates minimum investment of €100 million
- Optalysys plans to raise over US$100 million
- Fractile and Arago seek nine-figure rounds
- Axelera and Olix have already received more than US$200 million in 2026
Arquitetura from Euclyd promises gains in energy efficiency
Euclyd develops chips that, according to the company, deliver up to 100 times more energy efficiency in inference tasks compared to the Vera Rubin generation of Nvidia. The startup replaces the traditional approach with an architecture that processes data at multiple points simultaneously. Isso reduces power consumption and the physical size required in data centers.
The company has already completed an inference chip and is now working on a multi-chiplet system. Commercial launch is scheduled for 2028. Euclyd claims that its systems reduce both energy expenditure and the space occupied by AI installations. Até At this time, the technology has not been validated on a broad commercial scale.
Bernardo Kastrup, founder and CEO, highlighted that the company is already negotiating with four potential clients. The expectation is to meet two of them from next year and the rest the following year. The startup has support from former ASML executives, including former CEO Peter Wennink as an investor and mentor.
Contexto in the race for alternatives to Nvidia solutions
Nvidia consolidated its dominant position with GPUs initially designed for games and later adapted for training AI models. The company invested more than US$18 billion in research and development in the last fiscal year ending in January 2026. In December, it acquired assets from startup Groq for US$20 billion. In March, it announced an investment of US$4 billion in two companies focused on photonic technology.
The market gradually migrates to inference, where operational efficiency gains relevance. European Startups argue that their solutions best meet this specific demand. Patrick Schneider-Sikorsky, director of Nato Innovation Fund, noted that risks of concentration in suppliers and the need for computational sovereignty drive investments on the continent.
British company Olix works with photonic processors for AI. Investidores like Taavet Hinrikus, from Plural, point out that current electronic architecture reaches limits in miniaturization and heat management. Plataformas based on light would represent, according to him, the next technological advance. Olix is also targeting its first customers for next year.

Barreiras structures faced by European companies
Startups in the sector still encounters obstacles in the European ecosystem. The chip development cycle is long and the path from prototype to scaled production requires robust infrastructure. The continent lacks an equivalent to the American DARPA for strategic innovation funding.
Fabrizio Del Maffeo, CEO of Axelera, mentioned that European governments adopt a conservative stance towards new companies. Fragmented labor legislation between countries makes hiring difficult. Além Furthermore, there is a lack of strong incentives for consuming local products.
According to data from Dealroom, European AI chip startups raised US$800 million in 2026. In the same period, Estados Unidos’s competitors raised US$4.7 billion. American Exemplos include Cerebras Systems, which raised $1 billion in February, and others like MatX, Ayar Labs and Etched, each with $500 million rounds.
Mesmo with lower values, investor interest in European solutions for AI inference grows. Carlos Espinal, partner at Seedcamp, stated that the topic is no longer a niche bet and has become part of the center of discussions on artificial intelligence infrastructure.
Próximos steps and perspectives for the segment
Euclyd plans to use the resources from the new round to scale production and advance commercial partnerships. The company already collaborates with ADTechnology, from Coreia and Sul, to develop silicon solutions. The goal is to reduce costs and energy footprint of data centers dedicated to AI.
Outras European companies follow similar trajectories, focusing on photonics or specialized architectures. The move reflects broader efforts to strengthen the continent’s industrial capacity in advanced semiconductors.
The race for alternatives continues at a rapid pace. Tanto to Nvidia and new entrants invest heavily to meet the growing demand for efficient artificial intelligence processing.






