AI is reshaping every industry—and now we’re redefining its power. Our mission is to democratize AI, generative and agentic AI as well as reasoning models, delivering maximum profitability for customers while minimizing environmental and societal impact.
Unlike conventional accelerators built for training, VSORA’s Jotunn 8 is engineered exclusively for inference. By optimizing latency-sensitive workloads, it achieves lightning-fast response times, dramatically higher throughput, and a significantly lower cost-per-query.
Discover how Jotunn8 makes AI more accessible, efficient, and sustainable than ever.
VSORA is a French fabless semiconductor company delivering ultra-high-performance AI inference solutions for both data centers and edge deployments. Our proprietary architecture achieves exceptional implementation efficiency, ultra-low latency, and minimal power draw—dramatically cutting inference costs across any workload.
Fully programmable and agnostic to both algorithms and host processors, our chips serve as versatile companion platforms. A rich instruction set lets them seamlessly handle pure AI, pure DSP, or any hybrid of the two, all without burdening developers with extra complexity.
To streamline development and shorten time-to-market, VSORA embraces industry standards: our toolchain is built on LLVM and supports common frameworks like ONNX and PyTorch, minimizing integration effort and customer cost.
Our rapid growth is powered by a talented, multicultural team united by pride in our work and a commitment to helping everyone shine. We’ve traded rigid hierarchies for a relaxed, friendly atmosphere where dedication, ownership, collaboration, agility, and engagement guide everything we do. For us, work isn’t a checklist—it’s a shared mission. You’re empowered to shape a leading company and celebrate in its success.
HQ
13 rue Jeanne Braconnier
Immeuble Le Pasteur
92360 Meudon-La-Forêt
France
Asia
Taipei
Taiwan
Japan
Tokyo
Japan
Korea
Seoul
Korea
USA
San Diego, CA
USA
Unmatched Performance at the Edge with Edge AI.
Fully programmable
Algorithm agnostic
Host processor agnostic
RISC-V core to offload & run AI completely on-chip
Tyr 4
fp8: 1600 Tflops
fp16: 400 Tflops
Tyr 2
fp8: 800 Tflops
fp16: 200 Tflops
Tyr 4
fp8/int8: 50 Tflops
fp16/int16: 25 Tflops
fp32/int32: 12 Tflops
Tyr 2
fp8/int8: 25 Tflops
fp16/int16: 12 Tflops
fp32/int32: 6 Tflops
Close to theory efficiency
Fully programmable
Algorithm agnostic
Host processor agnostic
RISC-V cores to offload host
& run AI completely on-chip.
fp8: 3200 Tflops
fp16: 800 Tflops
fp8/int8: 100 Tflops
fp16/int16: 50 Tflops
fp32/int32: 25 Tflops
Close to theory efficiency