At GTC Paris — held alongside VivaTech, Europe’s largest tech occasion — NVIDIA founder and CEO Jensen Huang delivered a transparent message: Europe isn’t simply adopting AI — it’s constructing it.
“We now have a brand new trade, an AI trade, and it’s now a part of the brand new infrastructure, known as intelligence infrastructure, that will probably be utilized by each nation, each society,” Huang mentioned, addressing an viewers gathered on-line and on the iconic Dôme de Paris.
From exponential inference progress to quantum breakthroughs, and from infrastructure to trade, agentic AI to robotics, Huang outlined how the area is laying the groundwork for an AI-powered future.
A New Industrial Revolution
On the coronary heart of this transformation, Huang defined, are techniques like GB200 NVL72 — “one large GPU” and NVIDIA’s strongest AI platform but — now in full manufacturing and powering every little thing from sovereign fashions to quantum computing.
“This machine was designed to be a considering machine, a considering machine, within the sense that it causes, it plans, it spends lots of time speaking to itself,” Huang mentioned, strolling the viewers by way of the scale and scale of those machines and their efficiency.
There’s extra coming, with Huang saying NVIDIA’s companions at the moment are producing 1,000 GB200 techniques per week, “and that is only the start.” He walked the viewers by way of a spread of obtainable techniques starting from the tiny NVIDIA DGX Spark to rack-mounted RTX PRO Servers.
Huang defined that NVIDIA is working to assist international locations use applied sciences like these to construct each AI infrastructure — companies constructed for third events to make use of and innovate on — and AI factories, which corporations construct for their very own use, to generate income.
NVIDIA is partnering with European governments, telcos and cloud suppliers to deploy NVIDIA applied sciences throughout the area. NVIDIA can also be increasing its community of expertise facilities throughout Europe — together with new hubs in Finland, Germany, Spain, Italy and the U.Okay. — to speed up abilities improvement and quantum progress.
Quantum Meets Classical
Europe’s quantum ambitions simply obtained a lift.
The NVIDIA CUDA-Q platform is stay on Denmark’s Gefion supercomputer, opening new prospects for hybrid AI and quantum engineering. As well as, Huang introduced that CUDA-Q is now out there on NVIDIA Grace Blackwell techniques.
Throughout the continent, NVIDIA is partnering with supercomputing facilities and quantum {hardware} builders to advance hybrid quantum-AI analysis and speed up quantum error correction.
“Quantum computing is reaching an inflection level,” Huang mentioned. “We’re inside attain of having the ability to apply quantum computing, quantum classical computing, in areas that may remedy some attention-grabbing issues within the coming years.”
Sovereign Fashions, Smarter Brokers
European builders need extra management over their fashions. Enter NVIDIA Nemotron, designed to assist construct massive language fashions tuned to native wants.
“And so now you understand that you’ve got entry to an enhanced open mannequin that’s nonetheless open, that’s high of the chief chart,” Huang mentioned.
These fashions will probably be coming to Perplexity, a reasoning search engine, enabling safe, multilingual AI deployment throughout Europe.
“Now you can ask and get questions answered within the language, within the tradition, within the sensibility of your nation,” Huang mentioned.
Each firm will construct its personal brokers, Huang mentioned. To assist create these brokers, Huang launched a set of agentic AI blueprints, together with an Agentic AI Security blueprint for enterprises and governments.
The brand new NVIDIA NeMo Agent toolkit and NVIDIA AI Blueprint for constructing information flywheels additional speed up the event of protected, high-performing AI brokers.
To assist deploy these brokers, NVIDIA is partnering with European governments, telcos and cloud suppliers to deploy the DGX Cloud Lepton platform throughout the area, offering immediate entry to accelerated computing capability.
“One mannequin structure, one deployment, and you’ll run it anyplace,” Huang mentioned, including that Lepton is now built-in with Hugging Face, giving builders direct entry to world compute.
The Industrial Cloud Goes Reside
AI isn’t simply digital. It’s powering bodily techniques, too, sparking a brand new industrial revolution.
“We’re engaged on industrial AI with one firm after one other,” Huang mentioned, describing work to construct digital twins based mostly on the NVIDIA Omniverse platform with corporations throughout the continent.
To additional this work, Huang introduced NVIDIA is launching the world’s first industrial AI cloud — to be in-built Germany — to assist Europe’s producers simulate, automate and optimize at scale.
“Quickly, every little thing that strikes will probably be robotic,” Huang mentioned. “And the automobile is the subsequent one.”
NVIDIA DRIVE, NVIDIA’s full-stack AV platform, is now in manufacturing to speed up the large-scale deployment of protected, clever transportation.
And to point out what’s coming subsequent, Huang was joined on stage by Grek, a pint-sized robotic, as Huang talked about how NVIDIA partnered with DeepMind and Disney to construct Newton, the world’s most superior physics coaching engine for robotics.
The Subsequent Wave
The subsequent wave of AI has begun — and it’s exponential, Huang defined.
“We’ve got bodily robots, and now we have info robots. We name them brokers,” Huang mentioned. “The expertise needed to show a robotic to govern, to simulate — and naturally, the manifestation of an unbelievable robotic — is now proper in entrance of us.”
This new period of AI is being pushed by a surge in inference workloads. “The variety of folks utilizing inference has gone from 8 million to 800 million — 100x in simply a few years,” Huang mentioned.
To fulfill this demand, Huang emphasised the necessity for a brand new type of pc: “We want a particular pc designed for considering, designed for reasoning. And that’s what Blackwell is — a considering machine.”
These Blackwell-powered techniques will stay in a brand new class of information facilities — AI factories — constructed to generate tokens, the uncooked materials of contemporary intelligence.
“These AI factories are going to generate tokens,” Huang mentioned, turning to Grek with a smile. “And these tokens are going to turn out to be your meals, little Grek.”
With that, the keynote closed on a daring imaginative and prescient: a future powered by sovereign infrastructure, agentic AI, robotics — and exponential inference — all in-built partnership with Europe.
Watch the NVIDIA GTC Paris keynote from Huang at VivaTech and discover GTC Paris classes.
