The stakes, according to Huang, are no longer measured in mere millions of transistors, but in trillions of dollars. Huang told the capacity crowd that computing demand has increased by 1 million times over the last few years. He projected the company’s revenue could reach $1 trillion between 2025 and 2027, driven by a world that is moving toward autonomous systems that reason, plan, and act.
The Next Generation: Vera Rubin and Feynman
The centerpiece of the presentation was the introduction of Vera Rubin, NVIDIA’s next-generation full-stack computing platform. Named after the astronomer who provided evidence for dark matter, the platform is designed to handle agentic AI—systems capable of completing complex tasks without constant human prompting. Huang also teased a subsequent architecture, Feynman, featuring a CPU dubbed Rosa (named for Rosalind Franklin), which underscores a shift in how NVIDIA views its work: as a foundational science for the 21st century.
AI Beyond the Screen: The Automotive Expansion
While much of the AI conversation has centered on digital text and images, Huang leaned heavily into Physical AI. This is the integration of AI into machines that move through the physical world, with the automotive sector serving as the primary commercial laboratory for high-stakes deployment.
Hyundai and Kia: Data-Driven Scalability
Hyundai Motor Group and Kia Corporation announced a broadened strategic collaboration to accelerate the deployment of software-defined vehicle (SDV) architectures. The partnership focuses on integrating NVIDIA DRIVE Hyperion into select vehicle models for Level 2 and above autonomy, while simultaneously advancing Level 4 robotaxi capabilities through the Motional joint venture. By leveraging NVIDIA’s accelerated computing, the Group aims to internalize proprietary AI models and establish a unified data learning pipeline.
This approach is designed to enhance functional safety and scalability across the Group’s global fleet, transitioning from traditional hardware-centric designs to high-performance, AI-driven systems. The collaboration positions Hyundai Motor Group to compete in the evolving autonomous ecosystem by utilizing NVIDIA’s full-stack hardware and software solutions to maintain a flexible, data-driven response to global market dynamics.
The Uber Robotaxi Rollout: 28 Cities by 2028
In a landmark move for the ride-hailing industry, NVIDIA and Uber announced a global expansion of their autonomous partnership. The deal will see a massive fleet of robotaxis—entirely powered by the NVIDIA DRIVE AV software stack—deployed on the Uber network. The rollout strategy is phased and systematic:
- Phase 1 (Early 2027): Launching in Los Angeles and the San Francisco Bay Area with data-collection and operator-supervised vehicles.
- Phase 2 (2028): Expanding to a total of 28 cities across four continents, including major hubs in North America, Europe, Australia, and Asia.
These vehicles will utilize “Alpamayo,” a next-generation reasoning-based AI model designed to handle complex “long-tail” scenarios. Uber CEO Dara Khosrowshahi emphasized that this partnership establishes a “multi-player” autonomous ecosystem, allowing various automakers using NVIDIA technology to seamlessly plug into Uber’s global demand network.
Visteon: Edge-to-Cloud AI Arbitration
Visteon Corporation announced the development of an edge-to-cloud AI arbitration architecture designed for software-defined vehicles. The platform utilizes a specialized automotive-grade AI operating system (AIOS) to dynamically distribute AI workloads between vehicle hardware and cloud infrastructure. Latency-sensitive functions are processed at the edge for enhanced privacy and speed, while complex reasoning models for fleet learning and predictive maintenance are managed via the cloud. The architecture integrates NVIDIA DRIVE AGX Orin and NVIDIA DriveOS, utilizing the TensorRT-Edge-LLM inference SDK alongside NVIDIA Nemotron open models and NIM microservices to support generative AI capabilities and vision-based in-cabin assistants.
Hesai: Standardizing Lidar Safety
Hesai Technology has joined the NVIDIA Halos AI Systems Inspection Lab, the primary ANAB-accredited facility for AI-driven physical systems validation. The collaboration focuses on evaluating Hesai lidar platforms—recently selected for the NVIDIA DRIVE AGX Hyperion 10 reference architecture—within a unified framework covering functional safety, cybersecurity, and AI compliance. Utilizing a “Safety Triad” engineering approach that combines functional safety (ISO 26262 ASIL B), SOTIF, and cybersecurity, this integration aims to standardize safety architectures as AI-defined fleets transition from development to large-scale commercial deployment.
Industrial Autonomy
The ecosystem is further bolstered by the general availability of the NVIDIA IGX Thor platform. This industrial-grade system brings real-time AI to the edge for companies like Caterpillar and Hitachi Rail, managing autonomous heavy machinery and predictive rail maintenance in complex, safety-critical environments.
The Orbital Frontier
In the most ambitious claim of the night, Huang announced the Space-1 Vera Rubin project. This initiative seeks to bring AI data centers into orbit, processing massive streams of satellite data in real-time. By moving the “AI Factory” into space, NVIDIA intends to extend accelerated computing from Earth to the stars.