NVIDIA GTC Berlin: Live Updates on What’s Next in AI

The NVIDIA GTC Berlin keynote pregame is now live.

Join Deirdre Bosa and Gavin Baker live from Berlin’s Tempodrom ahead of NVIDIA founder and CEO Jensen Huang’s keynote. Hear from Europe’s leading founders, investors and industry leaders on how Europe is building its intelligent future.

NVIDIA GTC Berlin kicked off Tuesday with full-day workshops for developers on topics including AI agents, inference, accelerated networking and robotics.

Berlin has turned bright green for the conference, which will draw thousands of attendees to the expo hall, over 100 sessions, and NVIDIA founder and CEO Jensen Huang’s keynote address, which begins Wednesday, October 21 at 11 a.m. CEST. The keynote pregame, featuring over a dozen NVIDIA partners, starts at 9 a.m. CEST.

Hear from Germany’s AI Ecosystem

Researchers across Germany use NVIDIA AI and accelerated computing to advance scientific discovery. 

Hear Andreas Dengel, executive director of the German Research Center for Artificial Intelligence, share how the center harnesses NVIDIA accelerated computing and drives research projects focused on water forecasts, maritime maintenance and local news.

And learn how Christian Schiffer, team leader for a brain mapping project at the Institute of Neuroscience and Medicine at Forschungszentrum Jülich, one of the largest interdisciplinary research centers in Europe, is tapping AI to accelerate his work.

Stay tuned to this page for live updates from the keynote and the rest of the show. 

At the Siemens Electronics Factory in Erlangen, Germany, Siemens and NVIDIA are showing how AI and simulation can help manufacturers adapt to changing orders and get to production faster.

NVIDIA’s Rev Lebaredian, Kari Briski and Deepu Talla joined Siemens’ Cedrik Neike, Rainer Brehm and Mark Hindsbo for a tour of the factory to explore new industrial AI applications.

For factory planners, that starts with a familiar challenge: an urgent order arrives when the production schedule is already full. Siemens Intelligence Center X agents, built with the NVIDIA Factory Operations Blueprint, use factory data and planning tools to recommend options. They compare available materials, line capacity, delivery commitments and the time and cost of switching between products, helping planners understand how prioritizing one order could affect others.

Teams can test proposed changes in a digital twin while the real production line keeps running. The production controller reviews the trade-offs and approves the plan before the agent applies permitted changes through the connected planning system.

The collaboration also extends to circuit-board inspection. Siemens’ Teamcenter and Process Preparation X software work with NVIDIA Cosmos to create a virtual “golden board” and synthetic images of good and defective boards. These help prepare inspection systems before physical boards are produced, with the project targeting up to 85% less manual engineering effort in generating inspection programs.

The factory has around 1,000 production employees. Siemens is also investing €500 million into its Erlangen Technology Campus, expanding development and manufacturing capabilities and establishing the location as a hub for industrial AI and digital twins.

In NLND Berlin, an innovation campus based in a former cigarette factory, developers, engineers, students and researchers rolled up their sleeves to get their hands on autonomous agents. 

NVIDIA’s Build-a-Claw drew more than 500 people on a rainy afternoon for live demos, technical talks and a head-to-head agent competition. The gathering was the last leg of the European Build-A-Claw tour leading up to GTC Berlin, which officially begins Tuesday.

Anyone Can Build an Agent

All the fun didn’t detract from the importance of what everyone was building, and how, thanks to NVIDIA OpenShell.

The objective of OpenShell is making autonomous systems safe. OpenShell is a containment system for building agents giving them zero trust. It’s how the attendees were all building their agents and how it should be done.

That was just one of the lessons for Berlin’s developer community at the event. There will be much more to learn when GTC Berlin kicks off with NVIDIA founder and CEO Jensen Huang’s keynote Oct. 21 at 11 a.m. CEST. Watch live.

Along Stockholm’s waterway, founders, enterprise leaders and researchers streamed into Münchenbryggeriet’s largest hall to learn about the innovators building on NVIDIA technology across the Nordics. 

NVIDIA CEO Jensen Huang joined the crowd at NVIDIA’s Sweden AI Ecosystem Reception, the company’s capstone event in the country ahead of GTC Berlin. 

Huang pointed to the region's wealth of infrastructure companies and startups hungry for compute capacity and encouraged them to find each other.

“We have the momentum now starting to build, and once the flywheel goes, the sky is the limit,” he said.

He also touched on the importance of moving from general intelligence to specialized intelligence. 

“AI needs domain expertise. General intelligence is not sufficient to come into manufacturing, into drug discovery and become proficient,” Huang said. “You need to take that general intelligence and apply it for yourself and make domain experts.” 

Outside the venue, a Volvo car gleamed — a display of the iconic Swedish automaker’s autonomous driving technology, built on NVIDIA AI and accelerated computing. Volvo Cars today announced an expanded partnership with NVIDIA to unlock new levels of safety in automated driving technology in future cars, building on the companies’ decade-long collaboration.

That’s a wrap in Sweden —  next stop: Berlin. 

Industry leaders and government representatives discussed the strategic integration of AI across various sectors to drive innovation, economic growth and national security in Sweden.

Stockholm’s Eric Ericsonhallen concert hall, a converted church, was filled not just with Sweden’s technology, government and business leaders, but with ideas. Ideas about how the nation can best harness AI and the just launched supercomputer, Sferical One.

Sferical One is a shared supercomputer built on NVIDIA systems that will allow researchers in Sweden and the Nordics to customize and deploy NVIDIA Nemotron open models in a secure environment. A consortium anchored by AstraZeneca, Ericsson, Saab, SEB and Wallenberg Investments, with SKF as strategic partner, is behind the initiative. 

Representatives from those companies, Sweden’s startup and research ecosystem, as well as Sweden’s government shared their ambitions for AI across industries and the Nordic nation at the event.

Each industry leader anticipated SfericalOne would be a catalyst for innovation across every part of Sweden’s economy, from breakthroughs in healthcare and drug discovery to enhanced telecom networks, more efficient R&D in aerospace and standout service in banking. The presenters said that the advances enabled by the new supercomputer should aim to be transformational, rather than an incremental upscaling of existing approaches and services.

Deputy Prime Minister of Sweden Ebba Busch was blunt with the crowd, warning that moving too slowly on AI risked Sweden being left behind. 

“If we overregulate AI, it's not going to happen here,” Busch said. “That's the choice Europe needs to make, and I would like for Sweden to lead the way in terms of finding an innovative regulatory framework that makes sure that AI works for the purpose of overarching aims.”

After a stage reset, NVIDIA founder and CEO Jensen Huang joined Marcus Wallenberg, chair of Wallenberg Investments; Jenny Nordlöw CEO of Sferical AI; and Spotify founder executive chairman Daniel Ek to discuss what Sferical One will enable for Sweden and the Nordics.

Huang described the speed of change and improvement in AI across industries and applications. What it unlocks, Huang said, is ambition.

“Because you now have an instrument that’s so incredible, your ambition goes up. As companies, as societies, we’re going to be much more ambitious,” he said. “The next generation that comes out of school, they’re going to be insanely ambitious. Their imagination of what they think they can do is going to be incredible.”

Spotify’s Ek agreed, and added that the pace of that discovery and strength of that ambition is only increasing.

“There isn’t a week that passes that we’re not learning something new that we thought was impossible. I never thought I’d experience it at this pace,” Ek said. “My urge to everyone in the room is this: This will happen. Even if it feels like it’s not happening right now in your industry, it will happen.”

The claws were out in one of Europe’s leading tech hubs, Stockholm.

Developers, from the most experienced researchers to AI-curious newcomers, gathered at the Norrsken Foundation, a nonprofit hub for entrepreneurs and startups building businesses to solve the world’s biggest problems. The crowd from the Nordics and beyond came to build specialized AI agents, no coding required. It was NVIDIA’s last stop outside of Berlin before GTC.

That was true as well for NVIDIA CEO Jensen Huang, who explained the importance of what everyone was doing and NVIDIA OpenShell makes it possible.

“When you're building autonomous systems, you have to make sure that it operates in a way that's secure and safe,” Huang said. “So, we created a containment system, another operating system, that these agents can sit within. This containment system gives the agent zero trust, which is the way you should build it.”

That containment system is OpenShell.

“OpenShell is one of the most important things that we've done,” Huang said. “And I hope you all get involved.”

Huang described the moment developers in the Nordics — and across the globe — are experiencing as every industry is on the verge of being reimagined. Foundation models are being combined with specialized models and applied to specific problems, creating a flywheel.

“The number of possibilities for startups is really incredible,” Huang said. “You can create all kinds of interesting businesses with that. I think the days of superintelligence have really come.”

Two lucky attendees won the raffle for the latest DGX Spark and RTX Spark computers.

There was only one wrinkle. Demand for both AI machines has been massive, and Huang didn’t have an RTX Spark to hand over. Instead, he gave the excited winner a signed IOU.

“I'm good for it, I'm good for it,” Huang joked. “You know where I work.”

Next stop: Berlin and GTC. Check back for all the latest news.

Inside a converted church in the heart of Mayfair, London, NVIDIA founder and CEO Jensen Huang moved through the crowd, greeting founders, business leaders and — at one point, raising a toast to the startups in attendance.

NVIDIA’s U.K. AI Ecosystem Reception at Mercato Mayfair lauded the U.K. as a thriving center for technology innovation. 

Huang took to the stage in front of a packed room, the crowd spilling over into the balcony sections. He played to the London crowd by mentioning how several NVIDIA architectures have been named after English scientists.

After thanking the audience for their contributions and continued work with NVIDIA, he emphasized the importance of safety — especially with regard to agents, which he said must be transparent — and how AI will help solve some of the world’s greatest challenges. 

“AI is our best shot at accelerating the transition to sustainable energy,” he said.

Huang also underscored that “AI is not just the model,” and that “in order for the U.K. to participate in the AI economy, you have to participate in every single layer of the AI economy, the AI industry.”

To spur those efforts, Huang announced NVIDIA’s planned investments in U.K. startups across medicine, AI research, infrastructure and quantum computing — part of a broader commitment to invest nearly $10 billion into Europe’s technology ecosystem.

Kanishka Narayan, the U.K.’s Minister for Artificial Intelligence, joined Huang and the startup founders on stage for a toast.

“Go U.K.!” Huang said. “You gotta put pedal to the metal, and let’s get the U.K. ecosystem flywheel going in the U.K.”

Developers packed One Triton Square in London today to build their own autonomous agents at Build-a-Claw, an event that gave attendees access to NVIDIA-powered computers to design their own agent, no coding required.

The gathering was part of NVIDIA’s tour through Europe ahead of GTC Berlin next week — and featured a visit by NVIDIA CEO Jensen Huang.

“OpenClaw, as you know, makes it possible for the agent to be able to deal with memory, orchestrate action and use tools. As a result, we have now seen an inflection, and artificial intelligence has become useful,” Huang told the crowd, citing applications including drug discovery, coding, marketing and academic research.

Huang described the need for an isolation layer, “another operating system inside the operating system,” to ensure agents are operating securely. 

“And so we created this thing called OpenShell,” Huang said. “Whatever the task you give to your agent, you then give it minimum rights, minimum privileges, and then you monitor it to make sure it’s compliant with policy.”

Huang also called attendees up to the stage to award raffle prizes including a signed NVIDIA DGX Spark.

AI Safety and Security Research Forum

Earlier in the day, Huang joined U.K. researchers at the AI Safety and Security Research Forum at the London Institute for Healthcare Engineering. His visit followed NVIDIA’s announcement of a compute donation valued at $250 million over five years to support U.K. agentic safety and security research.

The forum brought together some of the U.K.’s most talented engineers working on the biggest problems in AI safety. Two shared challenges emerged from the session: securing multi-agent systems, where individually secure agents can still form insecure systems, and rapidly scaling safety at the same pace as capability. Huang and others made the case for openness as the foundation for effective defense, and highlighted the U.K.’s world-leading strength in AI safety. He argued open models and solutions put defensive power in everyone’s hands.

“We need to democratize self-defense," Huang said. "We need a community of people watching each others’ backs.”

Deepu Talla, vice president of robotics and edge computing at NVIDIA, joined European Union and local leaders at the International AI Summit in Dublin to celebrate the talent, ambition and spirit of innovation in Ireland and across Europe.

Talla spoke on a panel titled The AI Backbone: Power, Cloud and Critical Infrastructure, joined by Robin Colwell, senior vice president of government affairs at Intel; Charlotte Baylac de la Codre, head of public affairs for compute at Mistral AI; and Jason Lynch, CEO of Equal 1, to discuss what it takes to power Europe’s next era of compute.

The panel’s moderator John Thornhill, innovation editor at the Financial Times, began the discussion by invoking NVIDIA founder and CEO Jensen Huang’s description of AI as a five-layer cake, with the layers (from the bottom up) comprising energy, chips, infrastructure, models and applications.

The question for Talla: Which layers can Europe play in? All of them, was the short answer Talla gave — but the most important layer for European nations to focus on is applications.

“When you have an AI factory, you’re providing energy and out comes intelligence,” Talla said. “It’s quite simply the most valuable commodity that humanity can have. The question then is, once you have this intelligence, how much do you need? Well, the answer should be as much as you can get.”

And then, what do you apply that intelligence toward?

Talla’s answer for Europe’s small and large enterprises as well as governments takes into consideration the region’s phenomenal manufacturing expertise. 

“Physical AI is the next big opportunity,” Talla told the audience, describing the use of robotics to move atoms, create physical objects and perform physical tasks. “Around 70% of GDP is in the physical world, and that makes physical AI one of the largest opportunities ahead of us. Europe can combine AI with its human capital, engineering expertise and industrial capabilities to create more value.”

That value will also extend to Europe’s small and medium businesses.

“Physical AI can bring capabilities that were previously available mainly to large enterprises to small and medium-sized companies,” Talla said. “That creates a major opportunity for Europe’s manufacturing base and its small and medium-sized manufacturers.”

And it will result in applications of AI — at every layer of the stack — that are uniquely European.

“At the model and application layers, local knowledge and local market understanding really matter,” Talla said. “We want to enable the entire ecosystem across startups and universities — everywhere — because ultimately those can become the large companies of the future.”

Salle des Tirages, a historic hall on Place Vendôme, has hosted events of consequence for more than a century. On Tuesday evening, it held France's latest ecosystem moment as founders, policymakers, researchers and press gathered for NVIDIA's France AI Ecosystem Reception.

NVIDIA founder and CEO Jensen Huang took the stage to discuss NVIDIA’s investments in Europe and the need for safe and secure technology development.

“Europe is one of the great regions of industrial manufacturing,” Huang said. “This is the beginning of the AI journey for the EU.”

Roland Lescure, France's Minister of the Economy, joined Huang onstage to talk about France’s strategy for innovation and economic growth.

“We want Europe to enjoy this revolution that’s going to create prosperity, jobs and revenue,” Lescure said. “For this we need pioneers.”

As attendees mingled at the reception, crowds gathered around a couple tiny guests: Microducks, 10-inch biped robots by Hugging Face that run on open source software.

Huang also stepped outside to address the press gathered at the venue about secure technology deployment.

“Containment and monitoring are the fundamental pillars of AI,” he said.

Later that evening, Huang met with French President Emmanuel Macron at the Élysée Palace to discuss NVIDIA’s contribution to France’s thriving AI ecosystem through investments spanning startups, AI-native companies and computing infrastructure.

They also spoke about the importance of open source for France and Europe, AI safety and NVIDIA’s recent launch of its Open Agent Safety Platform.

On the afternoon before NVIDIA's France Ecosystem Reception, hundreds of developers gathered at Pavillon Vendôme to build physical AI — and get a surprise visit from NVIDIA founder and CEO Jensen Huang.

Pavillon Vendôme sits just off Place Vendôme in the 1st arrondissement — one of the most storied addresses in Paris. On Tuesday afternoon, it was full of claws.

NVIDIA’s Build-a-Claw came to France with developers from the nation’s startup community and big companies alike packing into the pavilion. The excited crowd was lined up around the block to try their hands (and brains) at building AI-powered robotic claw machines using open models and NVIDIA’s platform.

NVIDIA founder and CEO Jensen Huang dropped in on the robotic fun chatting with the crowds inside and outside, and showing off NVIDIA DGX Spark and RTX Spark AI computing powerhouse machines.

“The world’s first computers that unite all the software the world has ever created,” Huang said.

Huang then raffled off one of each of the machines, signing the DGX Spark and handing out an autographed “IOU” for the wildly popular RTX machine, since it’s sold out at the moment.

The crowd of developers did not miss a beat, claws were waiting, and went back to building.

The Paris Build-a-Claw was NVIDIA's second event of the day, after a morning Build-a-Claw in Barcelona drew another big crowd at the Palacio de Pedralbes. He next headed to the nearby Salle des Tirages for an evening ecosystem reception with France's AI leaders.

A single day, and already there is a sense of the huge excitement and ambition across the tech ecosystem in both countries. More news, and more countries to come.

Before NVIDIA's keynote in Berlin, the road to GTC began in Barcelona — with a royal palace, hundreds of developers and a very particular kind of robot.

The Palacio de Pedralbes rises up on land that once belonged to the Spanish royal family at the edge of Barcelona. On Tuesday morning, it belonged to builders.

NVIDIA's Build-a-Claw brought together hundreds of developers, researchers and AI enthusiasts from across Spain and Europe for a morning of hands-on AI at the historic site. Using open models and NVIDIA's platform, participants built robotic claw machines capable of picking up objects — physical AI, made from scratch, in a few hours. The crowd was rapt as it watched robotic arms perform grasping tasks, learn from their own failures and improve over  the span of time it takes to eat breakfast.

For the crowd on hand it was an example of how to build and train robots using open models not just quickly, but safely and securely, something Spain's developer community is embracing. Check back for more robotic breakthroughs across Europe on the road to GTC Berlin.