French semiconductor company VSORA has announced the production launch of Jotunn8, a new artificial intelligence (AI) inference chip that the company claims is Europe’s most powerful to date. Designed to power large-scale AI applications in modern data centers, the chip promises to deliver unprecedented efficiency and scalability.
It would position VSORA as one of the few European challengers to U.S. tech giants dominating the AI hardware market.
Now entering production, Jotunn8 marks a milestone for both VSORA and the European semiconductor ecosystem. The chip is engineered to match the raw computational performance of leading AI accelerators from the United States while consuming less than half the power. According to the company, Jotunn8 achieves more than three times higher real-world inference performance than competing solutions, offering data centers a decisive edge in energy efficiency and sustainability.
“Jotunn8 allows us to compete on the global stage with a homegrown, energy-efficient AI chip,” said Khaled Maalej, founder and CEO of VSORA. “It’s about performance and a pivotal moment not just for VSORA, but for European technology. Europe now has a seat at the table to guide the future of AI.”
The timing of the launch coincides with mounting pressure on data centers to handle increasingly complex AI workloads while managing escalating power costs and sustainability mandates. As artificial intelligence systems grow in scale, inference – the process by which trained models generate responses or decisions – has become a dominant computational task. VSORA’s Jotunn8 chip is purpose-built for that role, designed to run inference workloads more efficiently than general-purpose GPUs that were originally optimized for training AI models.
AI Hardware Innovation
The company reports that Jotunn8 delivers up to 3,200 teraflops of compute power with over 50 percent utilization efficiency, while maintaining 50 percent lower power consumption than conventional GPUs. This efficiency, coupled with its scalability, enables cloud providers and enterprises to deploy AI inference at lower operational costs and with reduced environmental impact.
Unlike traditional architectures, the Jotunn8 chip was built from the ground up for inference acceleration rather than repurposed from training-focused designs. Its architecture minimizes data transfer bottlenecks and maximizes throughput, making it well-suited for emerging generative AI, language model, and computer vision workloads.
VSORA has partnered with leading global semiconductor manufacturers to produce the chip at scale, ensuring access to the latest fabrication technologies. The company is also engaging with data center operators and AI infrastructure providers to prepare for industrial deployment and mass adoption.
The introduction of Jotunn8 signals Europe’s growing ambitions in AI hardware innovation. With governments and industry leaders pushing to reduce dependency on foreign technology ecosystems, VSORA’s production-ready chip could represent a major step forward in achieving greater technological sovereignty in the AI era.
