A few hours before the board of directors of OpenAI ousted its CEO and initiated a five-day saga that has captivated the global artificial intelligence community, Xavier Niel disclosed plans to establish a French version of Silicon Valley in the 13th district of Paris.
Kyutai, a non-profit research laboratory, aims to develop and experiment with advanced speech models, algorithms that predict sentence endings, and generate essays and scripts. It mirrors the original ethos of OpenAI rather than its current contentious state. Niel, accompanied by fellow European billionaire Rodolphe Saadé, the head of shipping giant CMA CGM, and former Google CEO Eric Schmidt, announced a €300 million initiative at his startup hub Station F.
Niel’s initiative represents a final push in Europe to compete in the realm of relational AI. Together with the emerging Mistral AI project, supported by Nel and Saadé, it has the potential to position Paris as the epicenter of the continent. Niel emphasizes the importance of Europe’s ability to influence, benefit from, and regulate these powerful algorithms, which are often shaped by the biases inherent in their training data, in shaping our future.
The media-shy telecommunications entrepreneur expressed to French television station FranceInter, “We are taking swift action, dedicating resources to this cause, and progressing rapidly.” He added, “I am adamant that our youth should not be reliant on foreign-made algorithms.”
In contrast to Kyutai’s open approach, OpenAI’s collaboration with Microsoft to commercialize ChatGPT and transition into a profit-driven entity has deviated from the original vision. Kyutai remains committed to openness, allowing anyone to utilize its analysis for commercial purposes. The lab is led by former employees of Deepmind, Meta, and Microsoft, overseen by prominent figures such as Yann LeCun, Bernhard Schölkopf, and Yejin Choi. Transparency, including access to the models’ source code, is a key focus.
The recent power struggles at OpenAI in the United States, stemming from concerns about the rapid advancement of conceptual AI tools, underscore the necessity for alternative governance structures in managing potentially disruptive technologies. The fallout from the controversy may provide European competitors with opportunities to attract investors. It also emphasizes the need for Europe to enhance its computational capabilities for training these advanced models.
Martin Tisné, leader of the AI Collaborative supported by Pierre Omidyar and an AI consultant to the French government, highlighted the dependence of AI companies on major tech firms for computing resources. He emphasized the market power held by companies like Google Cloud, Amazon Web Services, and Microsoft, which often support AI start-ups.
Niel, cognizant of this landscape, aims to leverage Nvidia’s technology through his cloud company Scaleway to train Kyutai’s models. He mentioned, “We have established the fifth most powerful computer globally, with the top four located here.” He also pointed out the potential of OVH, a French company, to support Western start-ups. Despite skepticism about Europe’s competitiveness in AI, Niel remains optimistic, stating, “I have built my career on defying skeptics’ doubts.”
