Nvidia Corporation · NASDAQ:NVDA
Nvidia — GTC Berlin 2026 Keynote
55 slides from Nvidia's conference presentation, Oct 21, 2026
Slide 1: NVIDIA
GA NVIDIA
Slide 2: Slide 2
Slide 3: Speakers
Frédéric Pascal David Jeannette Godemel Daloz ¢ Silver zu Firstenberg Arthur Arkady Deirdre Jensen Gavin n Ola Zhaopeng Kaindl Volozh ? Bosa Huang AK Baker DA __ Killenius ee Chen,
Slide 4: What Do L'Oréal, Revolut, Spotify, Merck, Vodafone, Zara, Porsche, Nestlé, and HSBC Have in Common?
What Do L’Oréal, Revolut, Spotify, Merck, Vodafone, Zara, Porsche, Nestlé, and HSBC Have in Common?
Slide 5: $10B Invested, 14 New Start-Up Investments, 12K+ Companies in Ecosystem
$10B 14 New Investments 12K+ Over the Last Year In Start-Ups This Week Companies in Our Ecosystem
Slide 6: Slide 6
Slide 7: AI Powers a New Industrial Revolution
AI Powers a New Industrial Revolution
Slide 8: Building Europe's AI Economy
Building Europe’s AI Economy Applications Models AI Factories Compute Energy
Slide 9: The NVIDIA DSX Full-Stack AI Factory Platform
The NVIDIA DSX Full-Stack AI Factory Platform CCCL NIXL cuDNN cuSPARSE NVTX NVSHMEM CUDA Python NCCL cuBLAS NVRTC Megatron TensorRT-LLM cuTILE Python Core CuTe DSL DOCA Transformer Dynamo Engine CUTLASS CUDA-X Libraries Platform Libraries Open Models Vera Rubin NVLink 6 Vera BlueField-4 Spectrum-6 G
Slide 10: NVIDIA Vera Rubin Powers the Whole Agent
NVIDIA Vera Rubin Powers the Whole Agent Vera Rubin NVL72 Vera MGX Vera CMX Vera RTX Compute Node Compute Node Storage Node Compute Node Model Processing Secure Agent Processing Knowledge Serving CUDA-X Tools Processing
Slide 11: Vera Rubin Will Generate Significantly Cheaper Tokens
“Vera Rubin will generate significantly cheaper tokens than the next-leading accelerator and thus make you significantly more money.” 30X 30X Tokens/$ Tokens/$ NVIDIA Vera Rubin NVL72
Slide 12: Slide 12
Slide 13: Announcing NVIDIA Vera Rubin DSX POD
Announcing NVIDIA Vera Rubin DSX POD Full-Stack Modular AI Factory Reference Architecture Factory-Built, Integrated, Pre-Tested Clusters Delivery-to-Deployment Cut From Months to Days The Container is the AI Factory
Slide 14: One CUDA Architecture. Many Markets.
One CUDA Architecture. Many Markets. Productive, Fungible, and Durable. Auto Gaming Public Sector AWS Microsoft Azure Defense Healthcare Retail & Logistics Google Cloud Oracle Cloud Energy M&E Supercomputing Meta SpaceXSI FSI Manufacturing Telco ENTERPRISE & INDUSTRIAL CSPs 10K+ Startups AMI Labs In
Slide 15: e¢
e¢
Slide 16: Useful AI Creates Demand for More Intelligence
Useful AI Creates Demand for More Intelligence
Slide 17: AI Factories Produce Intelligence. Agents Put Intelligence to Work.
AI Factories Produce Intelligence. Agents Put Intelligence to Work. Sandbox Knowledge Model Tools Tokens Harness
Slide 18: Announcing NVIDIA OpenShell Reference Architecture for Safe AI Agents
Announcing NVIDIA OpenShell Reference Architecture for Safe AI Agents NVIDIA Knowledge Model OpenShell Gateway Tools Supervisor Tokens Policy Prover Harness BlueField-4 DPU NVIDIA Sentry
Slide 19: NVIDIA OpenShell
NVIDIA OpenShell Gateway a Supervisor a Policy Prover Permissions Ceiling vy Within the Ceiling
Slide 20: Over 100 Partners Advance Agent Safety with NVIDIA OpenShell
Over 100 Partners Advance Agent Safety with NVIDIA OpenShell
Slide 21: A New Approach to Agent Security with Claude Managed Agents and NVIDIA
Anew approach to agent security with Claude Managed Agents and NVIDIA On September 28, NVIDIA announced the Open Agent Safety Platform, an open software platform and referen ‘em design for strengthening Al security. Anthropic and NVIDIA collaborated to bring itio -curity and control to the agent sta
Slide 22: Open Models Turn Expertise Into Advantage
Open Models Turn Expertise Into Advantage NVIDIA Leads Open Model Contribution NVIDIA Alibaba Cloud Hugging Face Meta Microsoft IBM Google Baidu AMD ByteDance Z.ai Alpamayo Ising Autonomous GR00T Vehicles Quantum Earth-2 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug
Slide 23: Turn Proprietary Data Into Sovereign Intelligence
Turn Proprietary Data Into Sovereign Intelligence NVIDIA Nemotron is “Incredibly Customizable” 100 80 +41 60 Accuracy 40 20 0 GPT 6.1 Sol Nemotron 3.5 Super VL
Slide 24: Turn Proprietary Data Into Sovereign Intelligence
Turn Proprietary Data Into Sovereign Intelligence NVIDIA Nemotron is “Incredibly Customizable” 100 +40 80 60 Accuracy 40 20 0 GPT 5.4 Nano Nemotron 3.5 Lightning
Slide 25: Turn Proprietary Data Into Sovereign Intelligence
Turn Proprietary Data Into Sovereign Intelligence NVIDIA Nemotron is “Incredibly Customizable” 100 +30 80 60 Accuracy 40 20 0 GPT 5.5 Nemotron 3 Ultra
Slide 26: Turn Proprietary Data Into Sovereign Intelligence
Turn Proprietary Data Into Sovereign Intelligence NVIDIA Nemotron is “Incredibly Customizable” 100 +15 80 60 Accuracy 40 20 0 Claude Opus 5 Nemotron 3.5 Super VL
Slide 27: Turn Proprietary Data Into Sovereign Intelligence
Turn Proprietary Data Into Sovereign Intelligence NVIDIA Nemotron is “Incredibly Customizable” 100 +20 80 60 Accuracy 40 20 0 Claude Opus 5.5 Nemotron 3.5 Lightning
Slide 28: An Ecosystem to Help Build Your Super Agents
An Ecosystem to Help Build Your Super Agents
Slide 29: NVIDIA Agent Knowledge Foundation
NVIDIA Agent Knowledge Foundation. Turn Company Data Into Company Knowledge. Unstructured Data Nemotron Nemotron cuVS Nemotron Parse Embed Rerank Analytical cuDF Structured Data Auto- Predictive Ontology Kumo
Slide 30: NVIDIA Agent Knowledge Foundation
NVIDIA Agent Knowledge Foundation ParseBench Unstructured nvidia/Nemotron-3.5-Parse Data Nemotron Nemotron cuVS Nemotron Parse Embed Rerank lightonai/LightOnOCR-3-0.8B Analytical PaddlePaddle/PaddleOCR-VL-1.6 cuDF Structured Data 4 PaddlePaddle/PaddleOCR-VL-1.5 Auto- Predictive Ontology Kumo 5 Datal
Slide 31: NVIDIA Agent Knowledge Foundation
NVIDIA Agent Knowledge Foundation LMEB Unstructured #1 nvidia/Nemotron-3.5-Embed-8B-BF16 PARETO Data Nemotron Nemotron cuVS Nemotron Parse Embed Rerank #2 BAAI/bge-multilingual-gemma2 Analytical #3 nvidia/Nemotron-3-Embed-1B-BF16PARETO cuDF Structured Data #4 DataScience-UIBK/uibk-embedding-v1 Auto-
Slide 32: NVIDIA Agent Knowledge Foundation
NVIDIA Agent Knowledge Foundation ViDoRe v3 Unstructured #1 nvidia/Nemotron-3.5-Rerank-8B-BF16 PARETO Data Nemotron Nemotron cuVS Nemotron Parse Embed Rerank #2 Qwen/Qwen3-VL-Reranker-8B Analytical #3 Qwen/Qwen3-VL-Reranker-2B PARETO cuDF Structured Data #4 nvidia/llama-Nemotron-rerank-vl-1b-v2 Auto
Slide 33: NVIDIA Agent Knowledge Foundation
NVIDIA Agent Knowledge Foundation BeyondArena Unstructured 0 Kumo-Tabular 1426 Data Nemotron Nemotron cuVS Nemotron Parse Embed Rerank 1 TabPFN-3.5 1376 Analytical 2 Kumo-Tabular-Medium 1376 cuDF Structured Data 3 Causilo 1327 Auto- Predictive Ontology Kumo 4 Kumo-Tabular-Small 1320
Slide 34: Slide 34
Slide 35: NVIDIA CUDA-X Tools Give Agents the Tools of the Trade
NVIDIA CUDA-X Tools Give Agents the Tools of the Trade Simulators | Solvers | Databases | Digital Twin Engines Optimization Engines | Search Engines | Recommendation Engines Earth-2 Aerial Warp Parabricks NVFlare cuQuantum Sionna PhysicsNeMo NemoClaw Newton CUDA-Q OpenShell cuLitho MONAI Isaac Lab N
Slide 36: Meta
O Meta cuML Newton pvcoMP Retriever cuba Foundation Stereo cuPotential = 7 Nemotron | cuSOLVERMp =
Slide 37: Lila Sciences Lab Assistant Agents Use NVIDIA BioNeMo and CUDA-X Tools
Lila Sciences Lab Assistant Agents Use NVIDIA BioNemo and CUDA-X Tools NVIDIA Knowledge Model OpenShell Gateway Tools Lila Supervisor Iris Agents Policy Prover Harness Target Identification Molecular Modeling Molecular Design Lab Digital Twin Lab Operations BioNeMo MONAI Parabricks CodonFM Boltz-2 P
Slide 38: Siemens Factory Agents Use NVIDIA Cosmos and CUDA-X Tools
Siemens Factory Agents Use NVIDIA Cosmos and CUDA-X Tools NVIDIA Knowledge Model OpenShell Gateway Tools Intelligence Supervisor Center X Agents Policy Prover Harness Process Preparation Technomatix Teamcenter X Digital Twin Composer Opcenter NeMo Helix Nemotron Omniverse Isaac cuOpt Kumo Cosmos Met
Slide 39: Dassault Systèmes Virtual Companions Use NVIDIA Nemotron and CUDA-X Tools
Dassault Systèmes Virtual Companions Use NVIDIA Nemotron and CUDA-X Tools NVIDIA Knowledge Model OpenShell Gateway Tools LEO Supervisor 3DX Agents Policy Prover Harness AURA Project Planning Sub-Agent LEO Design Sub-Agent LEO Simulation Sub-Agent LEO Validation Sub-Agent Nemotron NeMo Eval NeMo NeMo
Slide 40: Slide 40
Slide 41: Enterprise Platforms Build Agents on NVIDIA Agent Toolkit
Enterprise Platforms Build Agents on NVIDIA Agent Toolkit
Slide 42: Slide 42
Slide 43: Telecom Networks Become Distributed AI Infrastructure With NVIDIA Aerial AI-RAN
Announcing Telecom Networks Become Distributed AI Infrastructure With NVIDIA Aerial AI-RAN NVIDIA AI for Telco Nokia Aerial RAN Aerial Nemotron Cosmos TensorRT CUDA-RAN Computer
Slide 44: Siemens Healthineers Next-Generation Adaptive Radiotherapy
Announcing Siemens Healthineers Next-Generation Adaptive Radiotherapy NVIDIA AI for Healthcare Nemotron Omniverse Holoscan Cosmos-H Isaac-H IGX Thor
Slide 45: NVIDIA Hyperion: Building a Global Robotaxi Fleet and Ride-Hailing Ecosystem
NVIDIA Hyperion: Building a Global Robotaxi Fleet and Ride-Hailing Ecosystem Mobility Platforms AV Software DRIVE AGX Thor Auto Manufacturers NVIDIA AI for Robotaxis Alpamayo Halos SAFE Halos OS Cosmos Hyperion AlpaSim
Slide 46: Europe's Automakers Build on NVIDIA Hyperion
Announcing Europe’s Automakers Build on NVIDIA Hyperion NVIDIA AI for Robotaxis Alpamayo Halos SAFE Halos OS Cosmos Hyperion AlpaSim
Slide 47: NVIDIA Isaac: Building the Platform for the Next Great Industry
NVIDIA Isaac: Building the Platform for the Next Great Industry Model Isaac OS Builders NVIDIA Jetson Thor Onboard Compute Halos Safety OS Robot Platforms NVIDIA AI for Robotics GR00T Omniverse Halos Cosmos Isaac Lab Isaac OS
Slide 48: Announcing Agile ONE SN Humanoid Robot Built on Isaac GR00T
Isaac OS Announcing Agile ONE SN Humanoid Robot NVIDIA Jetson Thor Onboard Compute Built on Isaac GR00T Halos Safety OS Agile ONE SN 31-DOF Body Available Q1 2027 WUJI Hand 2 20-DOF 5-Finger Hands NVIDIA AI for Robotics GR00T Omniverse Halos Cosmos Isaac Lab Isaac OS
Slide 49: NVIDIA RTX Spark Changes What a Windows Laptop Can Be
“NVIDIA RTX Spark changes what a Windows laptop can be” ― CGMagazine “NVIDIA’s RTX Spark is an earthquake for the PC industry” ― PCMag UK
Slide 50: Slide 50
Slide 51: Slide 51
Slide 52: Slide 52
Slide 53: Slide 53
Slide 54: Partner logos
| 5 AppRobotics —accerture Sadvania ERacierosors —MNaiMotwe’ Ave Grams — BAKiia~—Qaterian «= ama angw = ASBIS 7s GD a A AVELIOS Gs meus: = [ave ae] © Santander . Fbaseten GIBEARROBOTICS Qbechtle gh FBenchling Ly Be TE = @Ycneck por — CINECA “Cisco cura Gea — cLou Q@orouian = IR cohere caynin ind Cor
Slide 55: NVIDIA
GA NVIDIA