RAISE Summit
Partager
Where visionaries and AI converge. Coming July 8th & 9th, 2026 at the iconic Louvre, Paris.
Screen. Keyboard. Mouse. That's still how most people "talk" to technology, and it hasn't changed in 40 years.
At RAISE Summit, Gautier Cloix, CEO of H Company, called this out directly: despite how far technology has come, the way humans interact with it is almost prehistoric. He estimates 80 to 90% of work is ex*****on, tasks humans are good at, sure, but not what they're best at. What people are actually great at is giving intention, and that's the layer AI should be built to serve.
Gautier also had a clear challenge for anyone buying AI: demand real ROI, and demand it in weeks, not months. If nothing has changed after three months, his advice was blunt, fire the company.
He backed that up with numbers of his own. H Company's models run at $3 per million tokens, compared to $125 per million tokens on some other frontier models, roughly 40 times cheaper to build and run, with a smaller carbon footprint too.
Full conversation now live on our YouTube channel.
Even Einstein needed a hundred failed attempts before the one that worked. At RAISE Summit Ekin Dogus Cubuk Co-CEO of Periodic Labs, made the case that real scientific progress has never come from a static dataset. It comes from a lab that can iterate with the universe itself, trying, failing, and trying again. He connected this to a deeper loop already underway: better AI helps design better chips, and better chips help build better AI. It's a new kind of Moore's Law, but rooted in physical hardware rather than just transistors. The twist making this iteration loop faster than ever? Cheap robotics. Cubuk pointed to simple robots, the kind that can make you coffee at SF airport, now available for around $10,000. That cost drop means physical experiments can now be run faster and far more often, accelerating the entire cycle of discovery. Which changes the arithmetic entirely. If a hundred failed attempts is the price of one breakthrough, the question stops being whether you have the insight and starts being how cheaply and how often you can be wrong. Einstein had a lifetime to run his hundred attempts. The next Einstein might have an afternoon. Subscribe to our YouTube channel to watch all RAISE Summit 2026 sessions ungated and free.
For two days in July, Paris held the people deciding how the next era of technology gets built.
9,000+ attendees from 90 countries. 350+ speakers. 200+ sponsors. 822 CEOs. 80%+ of participants at C-suite or decision-making level. 168 Fortune 500, CAC 40, FTSE 100, DAX and S&P 500 companies represented. Over $600B in AUM in the building.
Leaders from Google, Meta, OpenAI, Broadcom, Mistral, IREN, Cognition, Fireworks AI, Lovable, and hundreds more set the agenda for the next decade of AI. Mark Cuban, Yann LeCun, Sachin Katti, Arthur Mensch, and Manoj Nair among the voices that defined the programme.
“The size of the conference has tripled since last year, and the calibre of speakers has actually improved. It’s literally the best AI conference in the world.” — Helena Chao, Bank of America
RAISE wasn’t built to be an audience. It was built to be a working floor — thousands of one-to-one meetings running alongside the programme. The stages set the agenda. The tables around them set things in motion.
The 2026 summit ended. What it set in motion didn’t. See you in 2027, Paris.
🎬 Final Chain Films
09/07/2026
The final three are in. RAISE the Stakes, Europe's largest AI startup competition, narrowed the field from ten to three, and they pitch live on the Master Stage today at 6:15pm.
Your 2026 finalists:
Kinga Matuła, CEO and founder of QurieGen. Charles Maddock, CEO of Strawberry Browser. Rodrigo F. Baon, CMO at NeuralTrust.
Three teams, one stage, one title left to win. Come watch the final pitch decide it.
RAISE Summit 2026, Paris.
Stevie Case (Vanta) on the RAISE stage, on a shift happening faster than the job titles can keep up:
Stevie Case: "Just in the last three months, we have seen a 300% increase in functional engineering roles."
Go-to-market engineers, legal engineers, marketing engineers, whole categories of work that didn't exist a year ago. It isn't just upskilling, she argued, it's a wave of new roles being invented in real time, and companies hiring people into jobs the role itself has never had before.
Proof that AI isn't only changing how we work, but what the work even is. RAISE Summit 2026, Paris.
hashtag hashtag hashtag
09/07/2026
Midday on the Master Stage belonged to OpenAI, with two sessions from Sachin Katti (Head of Industrial Compute) bookending a fireside on the economics of speed. The throughline: AI has stopped being something you chat with and become something that does the work.
In his keynote, Katti made the case that the shift is already total inside OpenAI.
Sachin Katti: "Pretty much the entire company now produces its output tokens through Codex, not chat."
And it isn't just engineers, he noted. Customer support, legal, sales and go-to-market now run most of their knowledge work through an agentic interface, with research token use up nearly 50x, and model releases compressing from every six to nine months to roughly one a month, because AI is now speeding up its own research.
He then sat with Cerebras co-founder and CEO Andrew Feldman, hosted by RAISE's Henri Delahaye, for the harder economic argument: that speed itself is the product. Feldman's framing was the sharpest line of the hour.
Andrew Feldman: "Why is there no market for slow search? Why is there no market for dial-up internet?"
Give people fast tools, he argued, and they use them more, on harder problems, so the metric that matters isn't tokens burned but productivity delivered. He pointed to a compute deal north of $20 billion and 200 new megawatts across Lyon, Norway and Finland, much of it to meet OpenAI's demand. Katti's version of the same point: after quality comes interactivity, and their next frontier model [see flag: your notes say GPT-5.6] is built to run about an order of magnitude faster than anything out there.
Just before them, Synopsys CEO Sassine Ghazi, with CNBC's Arjun Kharpal, took the same automation story down to the silicon, arguing there's no chip designed today without Synopsys, and that chip design has reached "L4" autonomy, a cognitive layer that reasons and calls agents on its own. Two years ago he'd have put full physical-implementation automation five or six years out, he admitted. He's been wrong, in the same direction, two years running.
Chat commoditizes. Agents do the work. And speed, not size, is becoming the product.
May Habib, co-founder and CEO of WRITER, on the RAISE stage, on the real test for enterprise AI: can you actually trust an agent to run on its own?
May Habib: "Do you have autonomy that you can scale with trust? And that really comes down to accuracy and precision across a very long time horizon."
Her point: models drift, guardrails fall short, evals are often missing, and too many agents still need constant babysitting. And if you're babysitting the agent, where did the productivity actually go?
The question underneath the ROI debate, named on the Master Stage. RAISE Summit 2026, Paris.
09/07/2026
Day 2's late morning followed the money: what AI really costs to run, and who pays to build it.
Marco Argenti (Goldman Sachs), with Tom Mackenzie (Bloomberg), opened on the mindset shift. He's seeing a steady 20% productivity lift, with engineers no longer asking for more people but for more to do, and development turning prototype-driven rather than spec-driven: business teams vibe-code something that looks right, engineers then harden it into real material. But he drew one firm line.
Marco Argenti: "What I don't outsource is decisions. I like to make the final call, that's human judgment, and I haven't seen evidence that AI judgment is superior yet."
Then the economics. In Cheap Tokens, Expensive Mistakes, Dylan Patel (SemiAnalysis) and Liran Zvibel (WEKA), with Karen Kwok (Reuters Breakingviews), laid out how sharply inference has turned. Patel's read: demand is insatiable, nobody has enough compute, and the labs now hold real pricing power, with OpenAI gross margins around 55%. Zvibel's angle: the newest GPU alone doesn't win, layer in storage and memory and a lesser chip can outperform, with cache-hit rates above 95% freeing GPUs to generate instead of recompute.
And underneath all of it, the financing question. The Trillion Dollar Question panel, with Erwan Menard (Crusoe), Kent Draper (IREN), Tim Davis (Modular), Wayne Nelms (Orin), Jeremie Eliahou Ontiveros (SemiAnalysis) and James Krellenstein (Alva Energy), mapped the capital stack: Menard's three ways to sell infrastructure, megawatts, GPU-hours or tokens, Draper's A-rated GPU debt backed by a Microsoft offtake, Davis on CUDA lock-in as the real moat, Nelms on building a futures market for compute. Krellenstein went one layer lower still.
James Krellenstein: "We're not in the data center business, we're in the business of providing the megawatts so that Crusoe can sell megawatts."
Mindset. Margins. Megawatts. The closer you look at AI at scale, the more it comes down to what it costs, and who can finance it.
Day 2 opens with Google.
Amin Vahdat, SVP and Chief Technologist of AI and Infrastructure, took the Master Stage to open the final day with a keynote on the foundations: The Physics of Thought and the Architecture of Intelligence.
His frame for the whole era:
Amin Vahdat: "The industrial revolution was the muscle multiplier. We're now in the era of intelligence, the mind multiplier."
The biggest day, underway. RAISE Summit 2026, Paris.
09/07/2026
Day 2. The final day. And RAISE Summit 2026 opens it at the deep end.
First on the Master Stage, Amin Vahdat, SVP and Chief Technologist of AI and Infrastructure at Google, with a keynote that goes straight to the foundations, The Physics of Thought and the Architecture of Intelligence.
Yesterday mapped where AI value moves: who builds, who powers, who funds.
Today starts one layer deeper, with the question underneath all of it. What is intelligence actually made of, and how do you build the machine to run it?
If Day 1 set the stakes, Day 2 goes looking for the blueprint.
The biggest day starts now. RAISE Summit 2026, Paris.
Cliquez ici pour réclamer votre Listage Commercial.
Type
Contacter l'entreprise
Site Web
Adresse
Notifications
Soyez le premier à savoir et laissez-nous vous envoyer un courriel lorsque RAISE Summit publie des nouvelles et des promotions. Votre adresse e-mail ne sera pas utilisée à d'autres fins, et vous pouvez vous désabonner à tout moment.