European AI Chip Startups Target Nvidia’s Dominance with New Funding Rounds

European AI chip startups are aggressively pursuing capital to challenge Nvidia's grip on the semiconductor market, focusing specifically on the critical area of AI inference. While Nvidia's GPUs currently dominate the training of AI models, the industry is shifting toward inference—the process of running those trained models—where efficiency and power consumption become paramount. Dutch startup Euclyd, backed by former ASML leadership, is currently in discussions for a funding round of at least 100 million euros ($118 million). The company claims its architecture can deliver 100x higher power efficiency for inference compared to Nvidia's latest Vera Rubin chips. Similarly, the UK-based Optalysys is planning a fundraise exceeding $100 million later this year, while firms such as Fractile and Arago are also pursuing nine-figure rounds. This surge in activity is underpinned by a broader 'sovereign compute' imperative within Europe. Investors and policymakers are increasingly concerned with concentration risks surrounding TSMC and the impact of U.S. export controls, pushing capital toward homegrown silicon solutions. In 2026 alone, more than $200 million has already been funneled into the Netherlands' Axelera and the UK's Olix. Despite the ambitious funding targets, these technologies remain largely unproven at commercial scale. Euclyd is currently developing a multi-chiplet system it aims to produce by 2028, with initial customer deliveries expected to begin as early as next year for some partners. Olix is also targeting initial customers in 2027, utilizing photonics-based processors to move data using light rather than electricity to bypass the physical limits of traditional electronic architecture.