Kyutai is a French AI research lab with a $330 million budget that will make everything open source

This morning at Scaleway’s ai-PULSE conferenceFrench billionaire and Iliad CEO Xavier Niel gave more details about his plans for an AI research lab based in Paris.

This new lab called Kyutai will be a privately funded nonprofit working on artificial general intelligence. It will work with PhD students, postdocs and researchers on research papers and open-source projects. When Iliad was originally unveiled in this research lab, the company said that Niel committed €100 million to this project ($109 million at today’s exchange rate).

“Thanks to some amazing friends who are here today, we are now close to €300 million for financing this initiative,” said Xavier Niel at the conference. Among the “friends” is another French billionaire, Rodolphe Saadé, the CEO of the French shipping and logistics giant CMA CGM, who also put in €100 million. There are other smaller contributors, such as Eric Schmidt’s foundation and some unnamed donors.

This is another starting point because Kyutai is open to many donations. “What is interesting to many journalists in the room is that the project may interest other investors,” Saadé said at a press conference after the announcement.

As Kyutai will work on foundational models, they also need some computing power. The good news is that Scaleway, the cloud division of Iliad, recently acquired a thousand Nvidia H100 GPUs. These top-of-the-line GPUs are essential for inference and model training and are cost effective for Kyutai.

Kyutai has begun hiring for its core scientific team. Six men took the stage this morning to talk about their previous work and what they think for the research lab – Patrick Perez, Edouard Grave, Hervé Jegou, Laurent Mazaré, Neil Zeghidour and Alexandre Defossez. They previously worked with Meta’s AI research team FAIR, Google’s DeepMind division, Inria, and others.

Patrick Perez, who previously worked at Valeo, will be the director of the research lab. Kyutai also put together a team of scientific advisors who are famous AI researchers – Yein Choi, Yann LeCun and Bernhard Schölkopf. They only check everyone’s work once or twice a year and give feedback.

One of the reasons why Kyutai thinks it will convince some researchers to join its lab is that researchers will be able to publish research papers.

“Unfortunately, the big tech companies that allow scientific publications are very few. More than an ego boost for researchers, it helps to improve research and contribute to the common good,” said Niel during the press conference.

Of course, this is not the first open AI research lab. OpenAI, as the name still indicates, started as a nonprofit. But things changed a lot after Sam Altman started working full time at OpenAI in 2019. OpenAI moved to a more traditional corporate structure and raised funding from Microsoft.

Other companies are also working on open source foundational models, such as Meta with this one Llama model and Mistral AI. Kyutai’s models may also be open source, but the researchers describe their work as open science. They plan to release the open-source models, but also the training source code and data that explains how they released these models.

“When it comes to the timeline, I don’t think our goal should go as fast as Mistral, because our ambition is to provide a scientific objective, an understanding and a code base to explain the results ,” said Alexandre Defossez in the press conference. But they expect to have something to share within a year.

Laurent Mazaré, another researcher from Kyutai’s group, still describes the first open-source model of Mistral AI as a success because many members of the community are improving it and exploring use cases based on Mistral 7B model.

It will also be interesting to see if a research laboratory is more efficient in releasing foundational models compared to private companies, and how private companies use Kyutai’s work for commercial applications. .

Image Credits: Romain Dillet / TechCrunch

“I am also a firm believer in open source, and we must make it a French asset,” French President Emmanuel Macron said in a pre-recorded video message at the conference.

France’s position: regulating use cases, not models

Macron also used this opportunity to outline and defend France’s position on Europe’s AI Act, saying that use cases should be regulated, not model makers. France is pushing to bring down the AI ​​Act in trilogues (a trilogue is a negotiation between the three main European bodies, the Parliament, the Commission and the Council).

“Regulation is not the enemy of innovation, on the contrary. It is not a question of determining good models, but we must ensure that the services available to our citizens are safe for them, for other players in economy and for our democracy,” Macron said.

“With the work on the European regulation for artificial intelligence that is currently in the ‘trilogues’, the regulation should be controlled and not punishing, to preserve innovation and control the use rather than the technology,” he added.

Xavier Niel basically sided with France’s position on this topic during the press conference. According to him, Europe is lagging behind when it comes to AI innovation, and regulation will slow down newcomers to Europe and reduce the chances of them catching on.

“At the moment we are more on the side of innovation than on the side of regulation. Making regulation means it creates barriers to competitors,” said Niel.

Perhaps if French AI companies become more successful, things may change. “I’d love if one day we could talk about French imperialism in AI,” Niel added later in the conversation.

Leave a comment