RAISE Summit

@lqiao @FireworksAI_HQ Watch the full session: youtube.com/watch?v=AhZ0U3…

Specialized intelligence just became a $17.5 billion idea. Last week @lqiao made the case on the RAISE Summit Master Stage. This week @FireworksAI_HQ announced a $1.5B Series D — $1B+ ARR, up 5x YoY, 40 trillion tokens served daily. Her argument: efficiency means specializing the tool to its value. "We don't drive a Ferrari to grocery shopping." Full session 👇

@CharlieKawwas @Broadcom Watch the full conversation with Tony Kim (@BlackRock): youtube.com/watch?v=ADOeIX…

The race to AGI might not be won by the best model. It might be won by the best chip stack. At RAISE Summit, @CharlieKawwas, President of @Broadcom, shared a striking view from inside the frontier labs: the first company to fully integrate GPU, XPU, and CPU into a single agentic AI platform could be the first to reach AGI. Four to five frontier labs are racing toward exactly that. The scale backing this race is staggering. Charlie Kawwas revealed that one major lab alone is putting in 1.5 gigawatts of capacity this year, and expects that number to climb past 5 gigawatts next year. His read on demand: insatiable. Full conversation now live on our YouTube channel. Link in comments 👇

.@RaiseSummit 2026 wrapped up in Paris last week, and Netris had a jam-packed agenda. Over the course of two days, we met with many of the neoclouds, sovereign AI operators, telcos, and enterprise AI factories driving the AI buildout across Europe and beyond — as well as the ecosystem partners supporting them. Between two speaking sessions, two VIP receptions, and a steady stream of conversations at the Netris booth, the throughline was clear: networking is a foundational layer of every AI cloud — and one of the hardest to get right without automation. Thanks to RAISE Summit for bringing the AI industry together, and to everyone who stopped by to learn why Netris has become the leading network automation and multi-tenancy platform for AI cloud operators. #RAISESummit #Paris #AIInfrastructure #NAAM

Watch the full conversation: youtube.com/watch?v=iDpPFA…

Why don't we have real domestic robots yet? Or Level 5 self-driving cars? At RAISE Summit, Yann LeCun, Executive Chairman of AMI Labs, gave a clear answer: today's AI can't do what a 10-year-old, or even a cat, does without thinking twice. Despite years of hype around "full self-driving," we still don't have it. Despite huge investment in robotics, we still don't have machines that can navigate a home the way a child does. LeCun's point: language models are excellent with text and symbols, but when it comes to real-world signals, like video, sensors, and physical space, they're essentially lost. The uncomfortable truth is that scaling language models further won't close this gap. It's not a data problem or a compute problem. It's a fundamentally different kind of intelligence that hasn't been solved yet. Full conversation now live on our YouTube channel. Link in comments 👇

Watch the full conversation: youtube.com/watch?v=jLJ6jZ…

Three countries. 200 megawatts. One phone call that started it all.  At RAISE Summit, Andrew Feldman, Founder & CEO of Cerebras shared what's next: 200 megawatts of new data center capacity across France, Norway, and Finland, built largely to meet OpenAI's demand. Some go live this year, the rest by the end of next. The scale reflects a deal that's already one of the largest in Silicon Valley history: a compute agreement north of $20 billion, closed in December, born out of a conversation Feldman had with Sam Altman just months earlier about the moment being right for fast inference. It's a clear signal of where AI infrastructure spending is headed next, and how quickly a single conversation can turn into billions of dollars of build. Full conversation now live on our YouTube channel. Link in comments 👇

We ended the day with a rooftop reception overlooking the Seine, where we were also fortunate to host Yinon C. (Wiz). It was a privilege to partner with @RaiseSummit, whose team built a phenomenal conference bringing together top minds from around the world to discuss AI across the stack. Thank you to every speaker, enterprise leader, founder, investor, partner, and friend of Vintage who joined us and made the day so meaningful.

Chip design takes years. At @RaiseSummit, Ricursive says AI can dramatically speed up the process—and that better chips will in turn enable better AI models, creating a recursive cycle of improvement.

LIVE from Paris: Jason Interviews Two AI CEOs Disrupting Trillion-Dollar Industries: Voice and Legal Will we see the end of the billable hour? @ElevenLabs @mati @WeAreLegora @MaxJunestrand @Jason (0:00) ElevenLabs' $600M ARR Ramp, 600 Employees & Life Without PMs (15:34) Celebrity Voice Deals, Deepfake Impersonation & Racing OpenAI and Anthropic (31:42) Legora's Hypergrowth, Disrupting Law Firms & the Billable Hour (42:31) LexisNexis Decline, Legal Data Moats & Legora's Narrow AI Models --------------------------- Thanks to our partners for making this possible! @airwallex is a leading global payments and financial platform for modern businesses, offering trusted solutions to manage everything from business accounts, payments, treasury, and spend management to embedded finance. airwallex.com/allin @Oracle powers AI at every scale—from frontier labs to enterprise production. Your data. Leading models. No lock-in. One platform, architected for AI. Built for business. Visit oracle.com/artificial-int… #OracleAIExperienceLive

What will it take for AI to reach its full potential? At @RaiseSummit Paris, #NextSilicon CEO @Elad_raz joined @PGelsinger to discuss why the future of AI depends on adaptable architectures, system-level innovation, and rethinking AI infrastructure from the ground up. The conversation is just getting started. Read Elad's blog here: nextsilicon.com/insights/elads…

Hosting @theallinpod live from @RaiseSummit - a standout moment of the entire week. Two of the most consequential AI companies, @ElevenLabs and @WeAreLegora, led by European founders @mati and @MaxJunestrand, on the Master Stage with @Jason in his element. Worth every minute.

RAISE Summit 2026. That's a wrap. Queues at 6am for the Hackathon. Queues at 6pm on the final day of the Summit. The last hour of the last day. That's not obligation, that's an industry realizing it needs to be in the same room, right now. @ylecun: "You can't have agentic systems that are reliable unless they have world models. It's going to be a new AI revolution." @kmariappan: "For every human identity, we're now seeing 109 AI identities. Two months ago that figure was 82 to 1." @Mark_Papermaster: "The CPU-to-GPU ratio went from one-to-eight, to one-to-four, and now it's becoming one-to-one." @mcuban: "There's no better time to be an entrepreneur, because that's where AI is the most impactful." Thank you to every speaker, sponsor, and attendee who queued at 6am and stayed till 6pm. Next year, we go bigger. RAISE Summit 2026, powered by @IREN_Ltd

The results are in. RAISE the Stakes 2026, Europe's largest AI startup competition, has its winners, crowned live on the Master Stage after the final pitch. 🥇 First place: QurieGen, led by CEO and founder @KingaMatula . 🥈 Second place: Strawberry Browser, @charles_maddock , CEO. 🥉 Third place: @NeuralTrustAI , Rodrigo F. Baon, CMO. Three standout teams from a field that started at ten. Congratulations to all of them, and to Kinga and QurieGen for taking the title. This is what raising the stakes looks like. RAISE Summit 2026, Paris. #RAISE2026 #AI #RAISEtheStakes #Paris

Yann LeCun (@ylecun ) took the RAISE stage with a contrarian message for a room full of language-model believers: the current path doesn't lead where everyone thinks. Yann LeCun: "A four-year-old has seen as much data as the biggest LLMs, except it's video." His argument: LLMs are excellent with discrete symbols and useless with real-world signals like video and sensor data, which is why we still have no level-five self-driving, no domestic robots, nothing that matches what a ten-year-old, or even a cat, does effortlessly. A cat runs on 800 million neurons and still intuits gravity, inertia and object permanence, things our machines can't reliably grasp. The missing piece, he argued, is world models, and the stakes go beyond capability. Yann LeCun: "You can't have agentic systems that are reliable unless they can predict the consequences of their actions." He closed on culture: if the world's entire information diet runs through a handful of proprietary engines, that's the end of local culture, and no AI system is ever truly unbiased. The next AI revolution, in his telling, won't come from bigger language models. It'll come from machines that actually understand the physical world. #RAISE2026 #AI #Paris

Late on the Master Stage, two firesides split the same question in two: what does it take to build AI at scale, and what does it take to make it pay? First, the infrastructure. In Scaling AI Infrastructure for the Agentic Era, @CharlieKawwas (@Broadcom ) sat with Tony Kim (@BlackRock ) to explain why the biggest labs are walking away from general-purpose compute. His metaphor did the work. Charlie Kawwas: "This is why people come to the haute couture. You come to us. You tell us what kind of tuxedo or dress you want. We will customize it and build it for you and remove that general compute tax." That tax, he argued, is paid twice, in efficiency and in margin, which is why four of the five hyperscalers now spending most of the world's compute budget are moving custom at the rack level. The demand behind it is staggering: from under $1 billion in this space two years ago to a projected $100 billion-plus, and labs scaling from one or two gigawatts toward more than ten. Whoever first integrates GPU, XPU and CPU into a single platform, he suggested, may reach AGI first. Then, the payoff. In his fireside, Mark Hura (@Oracle ), with Alex Banks (@thealexbanks ) (The Signal @thesignalAI_ ), moved the conversation from stacks to outcomes. Mark Hura: "They're not thinking about buying an AI stack. They're looking at how do we bring an outcome." Customers have crossed from experimenting to executing, he said, and the wins are concrete: AI cutting construction billing errors by over 72% and removing 130,000 support calls a year, clinicians handed back time through assistants that prep every patient summary. The real unlock, in his view, is private enterprise data, where the IP lives, and the rule is simple: start with the business problem, and the technology follows. Custom silicon below. Business outcomes above. The full stack of turning intelligence into results, in twenty-minute windows. #RAISE2026 #AI #Paris

Some events you attend. This one you remember. For two days, the Carrousel du Louvre held the people building the future of AI: the founders shipping it, the investors backing it, the operators putting it to work. Three stages. Thousands of conversations. One current running through all of it. We didn't want to tell you what RAISE was. We wanted them to. So we asked. And this is what they said. RAISE Summit 2026, Paris. Until next time. #RAISE2026 #AI #Paris