Nvidia's European AI Expansion

Summary

Nvidia Founder and CEO Jensen Huang delivered the keynote at Nvidia's first GTC event in Paris, addressing European partners and announcing new collaborations, technological advances, and AI infrastructure initiatives for the region.
Key topics included the evolution of accelerated computing, advances and inflection points in quantum and agentic AI, the unveiling of next-generation Grace Blackwell systems, partnerships with European companies such as Mistral, Schneider, Seimens, and Perplexity, and the role of digital twins, robotics, and AI factories in the new industrial revolution.
Major announcements included new AI cloud and infrastructure projects in Europe, the release of CUDA Q for quantum computing, and the worldwide deployment of DGX Cloud Lepton for multicloud AI workloads.
The session concluded by emphasizing Nvidia’s commitment to advancing AI for industrial, autonomous, and agentic applications, and Europe’s increasing strategic investment in AI infrastructure.

(no explicit due dates or individual ownership mentioned in transcript—action items listed as inferred next steps for referenced stakeholders)

Nvidia: Continue building out AI technology centers in seven European countries.
Nvidia & Mistral: Collaborate to launch a new AI cloud-service in Europe.
Nvidia & Schneider: Further develop digital-first AI factories with digital twin operations.
Nvidia & Perplexity: Integrate regional language models and connect them to the Perplexity reasoning engine.
European Partners: Continue ecosystem partnerships with Nvidia for industrial AI, cloud, and supercomputing infrastructure.
Nvidia & Disney Research/DeepMind: Advance partnership to enhance physics simulation in Omniverse for humanoid robotics.

Nvidia’s evolution from CPU acceleration to a comprehensive stack of over 400 domain-specific libraries, including libraries for quantum computing (CUDA Q), deep learning (QDNN), and AI model orchestration (Dynamo).
Introduction of CUDA Q for both quantum and hybrid quantum-classical computing, now accelerated on Grace Blackwell 200, marking a significant industry inflection point.
Overview of Nvidia’s core libraries and frameworks for various scientific, industrial, and AI applications.
Announcement: Entire quantum algorithm stack now accelerated for Grace Blackwell hardware.

Presentation of the new GB200 and GB300 Grace Blackwell systems, described as “thinking machines” and architected as massive virtual GPUs linked by the MVLink compute fabric.
MVLink enables unprecedented bandwidth (130 TB/s) and interconnectivity, supporting large-scale, high-throughput reasoning models and digital twins.
NVidia’s manufacturing of these large AI systems has scaled to over 1,000 supercomputers produced per week, outpacing previous generations in efficiency and performance.
Introduction of RTX Pro servers for enterprise IT: versatile systems running all major platforms, integrating seamlessly with legacy IT infrastructures.

Nvidia positions AI data centers (“AI factories”) as fundamental new infrastructure, focusing on generating intelligent tokens rather than data storage.
Europe’s AI compute capacity is expected to increase by 10x in two years; 20 new AI factory projects are in development, including several gigawatt-scale facilities.
AI technology centers are being built in seven European countries, fostering research, startups, and ecosystem development.
Partnerships highlighted with leading European companies (Seimens, Schneider, Mistral) and public sector initiatives, including collaboration with President Macron.

Open models (such as Mistral, Llama, Deep Seek) are progressing rapidly and are post-trained and enhanced by Nvidia under the “Neotron” strategy.
Neotron models (including Llama variants) receive NVIDIA’s post-training, architecture search, and reasoning augmentation, producing top benchmark results and supporting extensive enterprise context.
Nvidia emphasizes the right for companies to use and enhance open models with their proprietary data for regional and industry-specific needs.

Agentic AI described as the next major wave, allowing AI agents to reason, plan, and execute complex, multi-step tasks using context, memory, and specialized tools.
Demonstration of agentic AI with Perplexity, showcasing how a single prompt can trigger coordinated work across multiple agents for business planning tasks.
Launch of DGX Cloud Lepton and integration with Hugging Face enables seamless multicloud AI deployment and management of NIM endpoints and microservices.
The Nemo AI platform provides a full toolkit for onboarding, fine-tuning, monitoring, and deploying AI agents, now widely adopted by enterprise partners (Cisco, SAP, DeepL, Kodo, etc.).

Industrial AI revolution enabled through partnerships (notably with Seimens), leveraging Nvidia’s Omniverse platform to create digital twins for design, simulation, manufacturing, and logistics.
Photorealistic, physics-accurate digital twins are critical for training robot perception and interaction in virtual environments before deployment in the real world.
Announcement of the first industrial AI cloud in Europe, dedicated to high-fidelity simulation, digital twin operations, and industrial process automation.
Demonstrations of Omniverse-based digital twins for leading automakers (BMW, Mercedes-Benz, Toyota) and infrastructure projects (train stations, warehouses).

Nvidia Drive platform highlighted for safe, robust autonomous vehicle development, including end-to-end AV system safety and simulation in Omniverse.
Next step: Humanoid robotics powered by Nvidia Thor processors, running transformer-based reasoning models to generate motion plans; trained intensively in Omniverse digital twin environments.
Partnership with Disney Research and DeepMind established for advanced physics simulation in Omniverse, enabling more capable humanoid robots.

Launch of AI cloud partnership with Mistral in Europe — to support local AI model delivery and cloud applications for startups, leveraging both proprietary and open AI advances.
Introduction of CUDA Q quantum stack on Grace Blackwell — to address quantum-classical algorithm acceleration needs at industry inflection point.
Establishment of regional AI technology centers and AI factories in Europe — to drive local innovation, ecosystem growth, and infrastructure sovereignty.
Announcement of industrial AI cloud in Europe — to enable digital twins and advanced simulation for next-generation industrial automation.

Details regarding the industrial AI cloud’s full capabilities and launch timing in Europe to be provided on Friday (per Jensen Huang comment).
Further specifics to be announced regarding collaboration scope and model delivery timeline for Nvidia & Mistral’s AI cloud partnership.
Ongoing integration and regional adaptation plans for open Neotron models and Perplexity language support to be monitored.
Continued expansion and specifics of new European AI technology centers (countries, partners, and focus areas) pending further updates.