RAISE Summit 2026: Paris Emerges as the Global AI Epicenter

RAISE Summit 2026: Paris Establishes Itself as the Epicenter of Global AI

The RAISE Summit 2026, held at the Carrousel du Louvre on July 8 and 9, attracted 9,000 participants from around the world. The event highlighted a central message: artificial intelligence (AI) is now a functional reality, but the real challenge lies in the human transformation it entails. With such a gathering of leaders, Paris is asserting itself as the European capital of AI, a place where technology and humanity must co-evolve to make the most of technological advancements.

An Event on the Rise

The RAISE Summit has seen impressive growth since its inaugural edition in 2024, which brought together 2,000 participants. This year, that number quadrupled to 9,000, and the event is already scheduled to move to the Palais des Congrès in 2027 to accommodate even more attendees. Eric Schmidt, former CEO of Google, described RAISE as the fastest-growing AI conference in Europe, if not in history. On July 8 and 9, the Carrousel du Louvre hosted meetings among 350 speakers, 168 Fortune 500 companies, and prestigious investment funds such as Sequoia, Founders Fund, and Accel. The founders of RAISE, Hadrien de Cournon and Henri Delahaye, even had the honor of ringing the closing bell at the NYSE on April 30, underscoring the international significance of the event.

Macron, Cuban, LeCun: A Stage of Ambition

The summit opened with a video address from President Emmanuel Macron, who reiterated France's commitment to becoming a European hub for AI. He highlighted national champions like Mistral AI, the 23 validated infrastructure sites, and emphasized the importance of digital sovereignty in the face of American and Chinese giants. Mark Cuban, the iconic investor, then spoke alongside Anton Osika, co-founder of Lovable. Cuban pointed out that the main obstacle to innovation is time and imagination, while noting a significant increase in the number of startups and an increasingly entrepreneurial France. Other notable speakers included Yann LeCun (AMI Labs, former Meta), Mati Staniszewski (ElevenLabs), Amin Vahdat (Google), Marco Argenti (Goldman Sachs), Arthur Mensch (Mistral AI), and Clara Chappaz, Minister Delegate for AI.

Sponsors and Exhibitors: A Strong International Presence, Dominated by Americans

The RAISE Summit also showcased a strong international presence, with a predominance of American players in the AI infrastructure sector. Sponsors included giants like NVIDIA, Oracle, Broadcom, Google Cloud, Vultr, and Crusoe (specializing in high-performance GPU cloud), as well as Backblaze, which recently signed a $335 million five-year contract with CoreWeave. Cursor, a native AI code editor, was also present. On the media and finance side, NYSE Wired, Bloomberg, CNBC, and Nasdaq were official partners, while funds like Sequoia, Founders Fund, and Accel rounded out this Anglo-Saxon dominated landscape. A new generation of players, often referred to as neoclouds, is emerging, offering comprehensive solutions that go beyond merely providing GPUs, with data orchestration services and sovereignty concerns.

Two exhibitors particularly caught attention:

• Alibaba Cloud surprised with its presence, showcasing its Qwen 3.7 models and deploying its entire Agentic Cloud stack live, from foundational models to agent orchestration tools. This underscores Alibaba's global ambitions in the race for LLMs and offers a counterpoint to American dominance.

• ElevenLabs, a European gem in voice AI valued at several billion, impressed with live demonstrations of its voice agents, capable of strikingly realistic multilingual conversations in real-time. ElevenLabs' booth was one of the most visited, allowing for 1:1 sessions with technical teams. Mati Staniszewski, co-founder, also spoke during a keynote, embodying this new wave of AI that extends beyond text to conquer voice and multimodal interfaces.

Data Sovereignty: The Real Underlying Debate

Beyond the buzz of the booths, a crucial debate permeated all sessions: how to manage data in the era of large models? Companies are looking to avoid model lock-in, turning to open-source solutions and considering on-premise deployments to protect their proprietary data. Barak Kaufman, CSO of Wonderful, summarized the mindset of large companies facing generative AI: "In closed-door discussions with CxOs of major companies, it is clear they are looking to design architectures independent of models." This approach aims to avoid technological lock-in and maintain control over costs and data. Tokenmaxxing—sending all data to a single large general model—is now seen as naive and risky. The most advanced strategies focus on model-agnostic architectures capable of orchestrating multiple specialized models.

The CxO Summit: Unfiltered Peer-to-Peer Experience Sharing

Alongside the public event, RAISE hosted a CxO Summit—a private invitation-only format reserved for executives. With 822 CEOs, 1,345 CxOs from various sectors, and 168 Fortune 500 companies, the event facilitated strategic exchanges without any sugarcoating. Participants included prominent figures such as the CIO of Mercedes-Benz, the CDIO of Michelin, the Chief AI Officer of Schneider Electric, the CTO of BNP Paribas, and the Global Head of Data & AI of Sanofi. The knowledge partner for this edition was QuantumBlack, AI by McKinsey, ensuring high-level debates on themes such as value creation, scaling data, agentic AI, and global regulation.

Five Takeaways for French Companies

The roundtables I attended brought forth strong convictions, shared by Michelin, Sanofi, Schneider Electric, and Kantar. These large companies provided concrete, sometimes counterintuitive, but always useful insights.

1. Agentic AI works, but efficiency is not yet there. AI agents are operational, but their accuracy and reliability in complex contexts still need improvement. The real mistake would be to remain in a perpetual POC mindset. Impact requires an industrial approach, solid foundations—data, governance, architecture—and above all, a long-term plan.

2. The three-circle model redefines Make or Buy. Michelin proposes a framework of three concentric circles for the "build or buy" dilemma:

   • Circle 1 — Core Data & Secret Sauce: sensitive and differentiating data. Rule: Build, or Buy in a secure framework.
   • Circle 2 — Operational Efficiency: mixed approach possible.
   • Circle 3 — Support Functions: turnkey solutions without strategic risk.

3. Recruiting AI talent or upskilling existing teams? Both—but not just any way. A mixed approach is necessary, but it requires a real strategy. Recruiting sharp AI profiles is essential, but upskilling existing employees is equally strategic. True value emerges where AI expertise meets business knowledge. Sanofi added that it is crucial to calibrate capabilities according to use cases and profiles.