NVIDIA GTC Berlin 2026

NVIDIA GTC Berlin 2026 is the European edition of NVIDIA's GPU Technology Conference, running 20-22 October 2026 at STATION-Berlin, with the headline keynote by founder and CEO Jensen Huang held at the nearby Tempodrom on 21 October. It is a three-day developer conference on accelerated computing and artificial intelligence aimed at engineers, researchers, enterprise architects and the European AI ecosystem, and it has quickly become one of the most significant AI infrastructure events on the continent because NVIDIA's GPUs underpin the overwhelming majority of AI training and inference worldwide.

What's on the 2026 programme

GTC Berlin compresses NVIDIA's vast technical agenda into a focused European format: expert-led sessions, hands-on workshops through the Deep Learning Institute, and an exhibition connecting hardware, software and the developer community. The published topic list spans the full accelerated-computing stack and the AI workloads it powers. Core themes for 2026 include:

• AI infrastructure and accelerated computing — GPUs, networking, CUDA updates, and the systems that determine how large models can grow and how cheaply they run.
• Agentic AI — building, deploying and scaling autonomous agents that plan, reason and call tools, the dominant enterprise-AI theme of 2026.
• Physical and embodied AI — robotics, autonomous systems, simulation and digital twins, where vision-language-action models and sim-to-real training are reshaping the field.
• Generative AI and inference at scale — large language and multimodal models, and the relentless drive to make inference faster and more efficient.
• Sovereign AI — national and regional compute, a theme NVIDIA has pushed hard across its European events as governments build their own AI capacity.

The keynote and who should attend

The centrepiece is Jensen Huang's keynote at the Tempodrom on 21 October, scheduled from late morning and livestreamed free with no registration required to watch. NVIDIA keynotes routinely set the direction for the wider industry, unveiling new silicon, platform updates and partnerships, so it functions as a read on where the AI hardware-software stack is heading. Beyond the keynote, the conference is built for practitioners: machine-learning engineers and data scientists working on model training and deployment, cloud and enterprise architects designing AI infrastructure, roboticists and autonomous-systems developers, and research teams across Europe's labs and universities. The hands-on labs and technical sessions make it as much a working event as a showcase.

Why Berlin, and why it matters in 2026

Bringing GTC to Berlin reflects a deliberate European strategy. The continent is investing heavily in sovereign AI — building its own compute infrastructure, data centres and models partly in response to dependence on US providers and the demands of the EU AI Act — and Germany sits at the centre of that push. GTC Berlin is where NVIDIA courts the enterprises, public institutions and startups driving that buildout, and where European developers get direct access to the accelerated-computing roadmap rather than waiting for the flagship US event in California.

GTC is, more than any single-vendor conference, a window onto the supply side of the AI boom. The questions it answers — how much compute is available, how fast inference can scale, what custom silicon and networking can deliver — set the cost and tempo of AI development for everyone building on top of the stack. For 2026, the convergence of agentic AI, physical AI in robotics, and ever-larger inference workloads runs straight through NVIDIA's platforms, because the company both enables and anticipates those shifts.

The training and ecosystem programme

Much of GTC’s practical value sits outside the keynote. The Deep Learning Institute runs hands-on workshops where engineers earn certifications in topics such as building large language model applications, deploying inference at scale, and developing robotics and simulation workflows on NVIDIA’s platforms. The exhibition and surrounding sessions pull in the wider European ecosystem — cloud providers, server makers, data-centre operators, and the startups in NVIDIA’s Inception programme — so attendees can trace a single thread from silicon and systems through frameworks to deployed applications. For technical leaders, that end-to-end visibility is the reason to attend in person rather than only watch the livestream.

Registering and attending

Registration is handled through the official NVIDIA GTC site, with paid conference passes for the full in-person programme; the Jensen Huang keynote livestream is free and open to all without registration. STATION-Berlin and the Tempodrom sit close together in central Berlin, making the venue well suited to a compact, high-density technical event. Anyone planning to attend the in-person sessions, workshops and exhibition should register early, as NVIDIA's European editions have drawn strong demand. Whether you join in person or follow the keynote remotely, GTC Berlin 2026 offers the clearest single view of the AI infrastructure powering generative AI, agents and robotics across Europe and beyond.