Like King Arthur and Excalibur, here's yet another AI startup trying its luck at dislodging Nvidia from the AI stone

French startup VSORA raises $46 million to take on Nvidia with a custom-built AI inference chip.

May 8, 2025 - 18:40
 0
Like King Arthur and Excalibur, here's yet another AI startup trying its luck at dislodging Nvidia from the AI stone

  • VSORA raises $46 million to build Europe's alternative to AI giants
  • Jotunn8 chip targets inference with lower power and higher throughput
  • French startup challenges Nvidia’s lead in a training-focused chip market

French chip designer VSORA has raised $46 million in new funding as it aims to offer Europe’s main alternative to AI processors from the likes of Nvidia, AMD, Intel, and Google.

The Paris-based startup is betting big on a custom-built inference processor called Jotunn8, which it plans to bring to silicon in the second half of 2025 using 5nm production at TSMC.

While Nvidia’s GPUs dominate training workloads, VSORA is targeting inference with its chip which it says is faster and more efficient where it matters.

Ensuring Europe’s technological AI sovereignty

The funding round was led by Otium and a French family office, with additional backing from Omnes Capital, Adélie Capital, and the European Innovation Council Fund.

The investors are backing VSORA’s goal to carve out a space in a crowded and mostly non-European market.

“In a market dominated by global giants like Nvidia, VSORA is a unique opportunity for France and Europe, home to world-class engineering talent,” said Gaspard de Veyrac, Principal at Otium.

"With this funding, VSORA has the necessary tools to reshape the future of AI computation and secure a significant position in the global AI chip market.”

VSORA’s chip isn’t trying to be everything for everyone. Unlike GPUs, which are built mainly for training models, Jotunn8 is designed just for inference so focuses on performance per watt, latency, and cost per query.

The company claims Jotunn8 delivers more than three times the performance of current solutions while using less than half the power.

It’s aiming for 3,200 teraflops of compute power and will target workloads like ChatGPT-style generative AI, autonomous driving, and edge devices.

“This funding marks a pivotal moment for VSORA as we accelerate our mission to revolutionize AI chips and ensure Europe’s technological sovereignty in AI computing,” said VSORA founder and CEO Khaled Maalej.

“It will drive the finalization of our technology and the launch of our production, enabling VSORA to play a crucial role as the sole alternative to non-European chip designers. We are grateful for our investors’ trust and look forward to continuing our collaboration with industry leaders to bring our chip to market.”

You might also like