More than 9,000 attendees packed into the Carrousel du Louvre on July 8 and 9 for the RAISE Summit, and the conversations inside skipped past the demo and hype phase that’s defined so many AI conferences over the past two years. The event opened with a video address from French President Emmanuel Macron, followed by a keynote conversation between investor Mark Cuban and Anton Osika, co founder of the AI app-building platform Lovable. What united the sessions that followed wasn’t a single flashy model launch. It was a shared focus on infrastructure economics, energy costs, and how enterprises actually get return on the AI they’ve already bought a notably sober set of priorities for a summit trying to position an entire city as a global AI capital.

Inside RAISE 2026

RAISE drew attendees from more than 2,000 companies, with roughly 80% holding C-level or founder titles, a concentration that reflects the event’s stated focus on strategic decision making rather than technical demonstration. The programme organized itself around what organizers call a “4Fs” framework Foundation, Frontier, Friction, Future built around sovereign cloud infrastructure, data provenance, the energy compute relationship, and AI governance at scale. An adjacent event called MACHINA, dedicated specifically to physical AI and robotics, including humanoids and industrial automation, debuted alongside the main summit on July 7.

The tone throughout leaned toward deployment mechanics over model-of-the-month excitement. One recurring theme, according to Forbes’ coverage of the event, was how neoclouds cloud infrastructure providers built specifically around AI workloads are evolving from simple GPU rental services into fuller infrastructure platforms offering global services, applications, and stronger data privacy guarantees. That shift matters because it signals enterprises attending RAISE are no longer treating compute access as the bottleneck; they’re treating what happens after they have compute access as the harder problem.

Efficiency Over Scale: The Summit’s Real Theme

If there was a single argument running underneath RAISE’s individual sessions, it was that enterprises are done buying AI capability on faith and have started demanding proof of return. Barak Kaufman, chief strategy officer at enterprise AI platform Wonderful, put the shift plainly at the summit: return on investment increasingly has to come from both revenue growth and cost savings, not one or the other. That calculation is pushing companies toward a genuine debate over whether to keep renting frontier models from OpenAI and Anthropic or build custom infrastructure around open-source models instead, driven by concerns over data sovereignty, avoiding vendor lock-in, and controlling proprietary information.

Dmitry Panenkov, founder of Luxembourg based cloud orchestration company emma, framed the infrastructure shift in similarly practical terms: neoclouds have to evolve from being real estate providers into companies that add genuine value on top of raw compute. That’s a meaningfully different conversation than the one dominating AI conferences even a year ago, when simply securing enough GPU capacity was treated as the main achievement worth discussing.

The Bigger European Ambition

Macron used his platform at the summit to make an argument that extends well beyond any single technology metric: that Europe needs its own AI infrastructure and genuine data sovereignty in a world increasingly split along geopolitical lines. That message isn’t new Macron made a similar case at the 2025 AI Action Summit, where he argued Europe was “back in the race” for AI while acknowledging the continent remained too slow for many investors, and where he specifically pointed to France’s nuclear energy production as a structural advantage for powering AI infrastructure at scale.

The throughline across both summits is that Paris’s bid to be Europe’s AI capital rests as much on energy and infrastructure independence as it does on any single startup’s model quality. A supercomputer, a nuclear-powered grid, and a research tax credit are less exciting than a benchmark score, but they’re the kind of durable advantage that doesn’t disappear the next time a rival lab ships a better model.

What Stands Between Paris and the Claim

None of this settles the question of whether Paris actually becomes Europe’s defining AI capital, rather than one strong node among several. London, Berlin, and Amsterdam all have their own AI ecosystems and their own claims on European tech leadership, and hosting a 9,000 person summit is a different achievement than building the next frontier lab to rival OpenAI or Anthropic at scale. Mistral remains France’s only AI company operating at genuinely global scale, and one decacorn, however impressive, is a thinner foundation than the multiple frontier labs the U.S. can point to. RAISE’s own organizers have already announced the event will need to move to a larger venue, the Palais Royal, next year to accommodate growing attendance a sign of real momentum, but attendance growth and technological leadership are not the same measure of success.

The Bottom Line

What distinguishes this year’s version of “Paris as Europe’s AI capital” from the same claim made a year earlier is the specificity of the argument behind it. The city isn’t betting its case purely on Mistral’s valuation or a single summit’s attendance figures. It’s betting on nuclear-powered energy independence, sovereign compute infrastructure, tax incentives that have already produced measurable startup growth, and a policy environment explicitly designed to keep European AI companies from depending entirely on American or Chinese infrastructure. Whether that combination is enough to make Paris more than a very well organized meeting point for an industry still largely built elsewhere will depend on whether the next Mistral scale company emerges from French infrastructure within the next few years, not on how large next year’s RAISE Summit venue needs to be.