AMI Labs, l'Europe est-elle enfin dans la course mondiale ?

In this edition of *Silicon Carnet*, host Carlos Diaz and his guests—tech entrepreneur Fred Lanier, Pierre Gbille, and analyst Fabrice Epelboin—dive into the massive shifts currently reshaping the global technology landscape. The discussion centers on three major pillars: the record-breaking funding of the French AI startup Hamilabs, the philosophical and geopolitical divide over the future of artificial intelligence, and the influence of digital platforms on the recent French municipal elections. ### Hamilabs: A New Paradigm for European Tech The headline news is the historic funding round for Hamilabs, co-founded by Yann LeCun, the renowned "godfather of AI" and former Chief AI Scientist at Meta. The startup has raised over $1 billion at a $3.5 billion pre-money valuation. The investor pool is notably global, led by Cathay Innovation—a Franco-Chinese-American fund—and includes high-profile names like Bezos Expeditions, Nvidia, Temasek, and various South Korean and German funds. Interestingly, top-tier Silicon Valley firms like Sequoia or Andreessen Horowitz are absent from this specific round. The panel notes that while Hamilabs is headquartered in Paris, its footprint is intentionally international, with offices in New York, Singapore, and Montreal, but notably nothing in Silicon Valley. This suggests a strategic move to position Paris as a global hub for "DeepTech." Pierre Gbille argues that this isn't necessarily an investment in Europe as an ecosystem, but rather an investment in the individual genius of Yann LeCun. However, the scale of the deal—the largest seed round in European history—signals that the continent is finally moving away from "sprinkling" small amounts of capital and is now capable of concentrating massive resources into single, high-stakes projects. ### World Models vs. Large Language Models (LLMs) Beyond the money, the guests discuss the scientific bet Hamilabs is making. Yann LeCun has been vocal about his belief that current LLMs (like those powering ChatGPT) are a "dead end" for achieving General Artificial Intelligence (AGI). His argument is that LLMs rely solely on linguistic relationships and lack a fundamental understanding of the physical world, causality, and spatial reasoning—things a human infant learns through observation and interaction. Hamilabs is focused on "World Models." These are designed to understand the physical laws of reality, which makes them particularly vital for the future of robotics and hardware like smart glasses. While OpenAI and Google are likely exploring similar avenues, Hamilabs is marketing itself as the lone challenger to the linguistic paradigm. The panel suggests that we are entering a phase where AI will be split by use cases: LLMs will remain dominant for linguistics, law, and marketing, while World Models will power the physical and robotic revolutions of the next decade. ### The Geopolitics of Compute and Energy A significant portion of the conversation focuses on the "bottleneck" of AI: energy and infrastructure. Fabrice Epelboin points out that while the US leads in software, it faces massive hurdles in building physical data centers due to energy shortages and local opposition. Plans to outsource this infrastructure to the Middle East have stalled due to regional instability and rising insurance costs. In this context, China holds a distinct advantage. Unlike the West, China lacks the social, legal, or "mental" friction that slows down technological deployment. They can build data centers and energy grids at a pace the West cannot match. The guests suggest that Europe’s nuclear energy capacity might actually give it a rare edge in supporting massive compute needs, provided the regulatory environment doesn't stifle progress. ### The "Treason of Progress" and the Death of Work The panel addresses the growing technophobia in the West compared to the techno-optimism in Asia. Elon Musk’s recent predictions serve as a catalyst for this debate. Musk suggests that within 20 years, work will be "optional"—an analogy he draws to growing vegetables in a backyard: it’s a hobby, not a necessity. This vision requires a Universal Basic Income, a concept that feels increasingly "French" despite Musk’s libertarian roots. The guests discuss why the Western middle class is so fearful. Historically, technological progress improved the lives of the working class. However, since the 1970s, the "Information Age" has primarily shifted value toward capital rather than labor. Now, AI threatens the "creative class"—the highly educated elites (lawyers, doctors, professors, and journalists) who previously felt safe from automation. Alex Karp, CEO of Palantir, is quoted as saying that AI will disrupt the economic power of highly educated Democratic voters while potentially empowering vocational, working-class voters. Fabrice Epelboin warns that when a highly educated, articulate population feels it has no future, the result is often social revolution. However, Fred Lanier offers a demographic counterpoint: with aging populations in Europe and Asia, AI might not be a choice but a necessity to maintain growth as the workforce shrinks. ### Digital Influence and the Municipal Elections The final segment analyzes the digital performance of candidates in the French municipal elections, specifically Sarah Knafo. While some media outlets and politicians (like Benoît Hamon) have accused Elon Musk of "tilting" the X (formerly Twitter) algorithm to favor right-wing candidates, the panel is skeptical of a direct conspiracy. Fabrice Epelboin notes that Knafo’s success—reaching 10% from a standing start—is due to her understanding of modern digital militancy. She, along with candidates from LFI (La France Insoumise), has mastered the "recipe" for the current era of social media: clashing tones, anti-establishment messaging, and content designed for virality. While X’s editorial line under Musk certainly favors "anti-woke" and "free speech" narratives, the guests argue that Knafo simply used the tools more effectively than traditional candidates like Rachida Dati, who still treat social media as a secondary broadcast channel rather than a primary battlefield. Ultimately, the panel concludes that the West is at a crossroads. Between the massive capital flowing into French AI labs and the deep-seated societal fear of job displacement, the next decade will be defined by how well Western democracies can adapt their political visions to a world where intelligence has become a utility, much like water or electricity.