Las Vegas - Nvidia founder and CEO Jensen Huang on Monday highlighted the rapid rise of open-source AI models like the Chinese DeepSeek R1 and revealed that Nvidia's Rubin AI computing platform is in full production.

In his keynote speech to open the 2026 Consumer Electronics Show (CES), attracting more than 2,000 attendees from around the world, Huang also underscored the role of Chinese and Chinese-linked partners in Nvidia's supply chain and system-building ecosystem.

He said the computer industry typically undergoes a reset every 10 to 15 years through major platform shifts, arguing that AI is now driving a fundamental transformation in how software is built and run.

In his two-hour keynote, Huang said developers are increasingly "training the software" and running new workloads on graphics processing units rather than relying solely on central processing units.

The Nvidia CEO pointed to the growing momentum of open models, particularly those released for broad use and adaptation. He said that open models "really took off" and described an "open model" reasoning system as a breakthrough that surprised the industry, noting that DeepSeek R1 "caught the world by surprise."

Open models have "reached the frontier" but remain about six months behind the most advanced frontier systems, Huang said, noting that new models are emerging on roughly six-month cycles and are "getting smarter," linking rising download numbers to strong global interest from startups, large companies, researchers, students, and countries.

On hardware, Huang revealed that Nvidia's major AI computing platform, Rubin, is in full production, describing it as the company's next-generation AI platform.

The Rubin platform is built as a system of six new chips designed to operate as one, Nvidia said Monday in a release.

Huang also emphasized the manufacturing and integration work required to build rack-scale AI systems, like the Rubin platform, referencing multiple major partners involved in their construction. He said "every single major computer company" knows how to build these systems, citing Foxconn and Lenovo among companies in the ecosystem, alongside other global suppliers.

To illustrate the physical scale of modern AI infrastructure, Huang said a rack-level system can include "2 miles" of copper cabling, or about 3.2 kilometers.

Huang's message to CES was that the AI shift is being shaped by both new computing platforms and broader participation enabled by open models. With Rubin moving into production and open-source reasoning models gaining attention, he framed the next stage as a period of faster deployment across industries and economies.