Jensen Huang CES 2025: Agentic AI Is the New Platform

Why Jensen Huang's CES 2025 Keynote Signals a Computing Reset

Jensen Huang's CES 2025 keynote wasn't just a product launch—it was a declaration that computing has fundamentally changed. With 6,000+ attendees and millions streaming worldwide, the NVIDIA founder laid out his vision: we're witnessing two simultaneous platform shifts (to AI and to physical AI), and agentic systems are the new application architecture. The scale is staggering: $10 trillion of computing being modernized, hundreds of billions in annual VC investment, and a hundred trillion dollars of industry R&D shifting to AI methods.

On the platform shift: "Every 10 to 15 years, the computer industry resets. A new platform shift happens—from mainframe to PC, PC to internet, internet to cloud, cloud to mobile. Except this time there are two simultaneous platform shifts happening at the same time." The first shift is obvious—applications built on AI. The second is physical AI: systems that understand and interact with the real world.

On how software is made now: "You no longer program the software, you train the software. You don't run it on CPUs, you run it on GPUs. And whereas applications were pre-recorded, pre-compiled and run on your device, now applications understand the context and generate every single pixel, every single token completely from scratch every single time." This is Jensen's clearest articulation of why this shift is fundamental, not incremental.

On agentic architecture: "Agentic models that have the ability to reason, look up information, do research, use tools, plan futures, simulate outcomes—all of a sudden started to solve very very important problems. One of my favorite agentic models is called Cursor, which revolutionized the way we do software programming at NVIDIA." When NVIDIA's own engineers are transformed by agentic tools, the signal is unmistakable.

On multi-model systems: "Of course, an AI would also call upon all of the world's great AIs to solve the problem it wants to solve at any part of the reasoning chain. This is why AIs are multi-modal, multi-model, multi-cloud, and hybrid cloud by definition." The architecture of future applications is clear: smart routers directing tasks to specialized models, combining frontier APIs with custom fine-tuned models.

On enterprise transformation: "Not only is this the way that you develop applications now, this is going to be the user interface of your platform. The agentic system is the interface. It's no longer Excel with a bunch of squares that you enter information into." Palantir, ServiceNow, Snowflake, NetApp, CrowdStrike—all integrating NVIDIA's agentic blueprints as their primary interface layer.

On physical AI and Cosmos: "The physical world is diverse and unpredictable. Collecting real world training data is slow and costly and it's never enough. The answer is synthetic data. Cosmos turns compute into data." This is the breakthrough: a world foundation model that generates physically plausible training scenarios, solving the long-tail problem for robotics and autonomous vehicles.

On the first thinking car: "Today, we're announcing Alpamo, the world's first thinking reasoning autonomous vehicle AI. It tells you what action it's going to take, the reasons by which it came about that action, and then the trajectory." End-to-end from camera to actuation, with reasoning that explains its decisions—trained on human demonstration plus Cosmos-generated scenarios.

7 Insights From Jensen Huang on Agentic AI and Physical AI

• $10 trillion modernization underway - The entire computing industry is being rebuilt for AI; R&D budgets across every industry are shifting to AI methods
• Agentic systems are the new app layer - Reasoning models that use tools, do research, and plan are replacing traditional software architectures
• Open models hit the frontier - DeepSeek R1 "caught the world by surprise"; open innovation is now 6 months behind frontier (and closing)
• Multi-model architecture wins - Future apps combine frontier APIs, custom models, and smart routers that match tasks to capabilities
• Cosmos solves the data problem - World foundation models generate synthetic training data grounded in physics, turning compute into data
• Physical AI is next - Three-computer stack (training, inference, simulation) enables robots and vehicles that understand the real world
• Enterprise UI becomes conversational - ServiceNow, Snowflake, Palantir all moving to agentic interfaces as their primary user experience

What This Means for Organizations Deploying AI

Jensen's thesis is bold: we're not adding AI features to existing software—we're rebuilding the entire computing stack from the ground up. The five-layer cake (chip, infrastructure, models, applications, users) is being reinvented at every level. For organizations, the message is clear: the platform shift is here, agentic systems are the architecture, and the companies that master this transition will define the next era. NVIDIA is betting that physical AI—machines that reason about the real world—will be as transformative as the language model revolution that preceded it.