VSORA launched Europe's most powerful AI inference chip, expanding the continent's semiconductor design capabilities.

VSORA, a French semiconductor innovator founded in Paris in 2015, announced the production launch of Jotunn8, an artificial intelligence inference chip engineered for next-generation data center applications. With this launch, VSORA becomes the only European company delivering AI inference chips that compete directly with those of leading U.S. technology giants. The company is headquartered in France with offices in Japan, Korea, Singapore, and Taiwan.

Jotunn8 matches the raw compute power of today's most advanced solutions while achieving more than 3X higher real-world performance and operating at less than half the power, providing significant advantages in efficiency, energy savings and sustainability for modern data centers. The chip delivers 3,200 teraflops of compute power at more than 50% utilization and 50% less power consumption, shattering the performance barriers of conventional GPUs. "Jotunn8 allows us to compete on the global stage with a homegrown, energy-efficient AI chip," remarks Khaled Maalej, VSORA's founder and CEO. "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."

Unlike most chips optimized for AI training, Jotunn8 is purpose-built for AI inference—the process of running trained models to generate responses, predictions or decisions—enabling fast decision-making with minimal power draw. This purpose-built architecture overcomes inherent GPU limitations and is designed to be scalable, integrating easily into modern data centers and AI infrastructure to set a new standard for efficient, high-performance inference computing.

VSORA is partnering with leading global semiconductor manufacturers to bring the chip to production, ensuring quality, scalability and access to cutting-edge fabrication technologies, while working with strategic stakeholders to prepare for large-scale deployment and industrial adoption. Full volume production is expected in Q1 2026, with pricing available upon request.