Nvidia presented its strategic direction at CES, emphasizing artificial intelligence (AI) and its foundational platforms rather than consumer graphics processing units (GPUs). The company, which recently surpassed a $5 trillion valuation, detailed its expansion into areas such as factories, autonomous vehicles, and robotics, aiming to power simulation and training across various domains.
Focus on Physical AI
A key concept introduced by Nvidia was "physical AI," defined as AI systems that perform actions in the physical world after being trained in virtual environments using synthetic data. These models are then deployed into physical machinery.
Nvidia CEO Jensen Huang highlighted several developments:
- Cosmos: A world foundation model designed for simulating environments and predicting movement.
- Alpamayo: A reasoning model specifically designed for autonomous driving.
These technologies are intended to support applications in robotics, industrial automation, and self-driving vehicles. As a demonstration, a Mercedes-Benz CLA was shown utilizing AI-defined driving on stage. Nvidia also announced plans to test its own Level 4 autonomous robotaxi service with a partner by 2027. This service would operate without human intervention in limited regions. Robotics, including autonomous vehicles, has been identified by Huang as Nvidia's second most significant growth category, following AI itself.
AI Platforms Take Center Stage Over New GPUs
The CES presentation did not include announcements of new consumer GeForce GPUs. Instead, the keynote primarily addressed Rubin, Nvidia's next-generation AI platform. Rubin is described as an integrated system encompassing GPUs, CPUs, networking, and storage, designed to manage the compute demands of modern AI models at data center scales. Nvidia positioned this platform as essential for addressing the increasing demands, energy consumption, and potential bottlenecks associated with AI training.
The absence of new gaming hardware indicates a strategic emphasis on enterprise and data center solutions. Nvidia's growth objectives are currently driven by hyperscalers, governments, and industries focused on automation.
Open AI Models and Platform Strategy
Nvidia reiterated its commitment to offering open AI models for developers. These models, trained on Nvidia supercomputers, are presented as foundational components for use, fine-tuning, and deployment. The company offers open models across various sectors including healthcare, climate science, robotics, embodied intelligence, reasoning AI, and autonomous driving.
Demonstrations also featured personal AI agents operating locally on Nvidia’s DGX Spark hardware. Nvidia's stated objective is to serve as the underlying platform for AI systems ranging from large data centers to individual desktop applications.
The keynote communicated Nvidia's positioning as a foundational provider for an AI-powered global infrastructure, with significant announcements centered on platform and ecosystem development rather than consumer-facing products.