By Lisa Eadicicco, CNN
Las Vegas (CNN) — Nvidia just provided a closer look at its new computing platform for AI data centers, Vera Rubin, a release that could have major ramifications for the future of AI given the industry’s massive reliance on the company’s tech.
Nvidia previously announced some details about Vera Rubin but laid out how the system will work and revealed its launch timing during the CES tech conference in Las Vegas on Monday. Vera Rubin is currently in production and the first products running on it will arrive in the second half of 2026, the company said.
Nvidia has become the poster child for the AI boom, with the pervasiveness of its AI chips and platforms propelling it to briefly become the world’s first $5 trillion company last year. But the company is also combatting fears of an AI bubble amid growing competition and a push by tech companies to make their own AI chips to decrease reliance on Nvidia.
Nvidia CEO Jensen Huang, clad in his signature leather jacket, addressed the question of where AI funding is coming from – a point central to the bubble debate – in his opening remarks on stage at the theater inside the Fontainebleau Las Vegas. He said companies are shifting budgets in research and development in classical computing methods to artificial intelligence.
“People ask, where is the money coming from? That’s where the money is coming from,” he said.
The Vera Rubin platform is an attempt by Nvidia to position itself as the answer to the computing challenges posed by increasingly demanding AI models – such as whether existing infrastructure can handle increasingly complicated AI queries. The company claims in a press release that its upcoming AI server rack, called Vera Rubin NVL72, “provides more bandwidth than the entire internet.”
With Vera Rubin, Nvidia says it’s developed a new type of storage system to help AI models process more complex, context-heavy requests more quickly and capably. Existing types of storage and memory used by traditional computers and even the graphics processing units powering data centers won’t be enough as companies like Google, OpenAI and Anthropic shift from offering simple chatbots to full-fledged AI helpers.
Huang walked through what the transition from chatbots to agents on Monday. In a video demonstration, a person built a personal assistant by connecting a friendly-looking tabletop robot to multiple AI models running on Nvidia’s DGX Spark desktop computer. The robot was able to do things like recount the user’s to-do list and even tell the dog to get off the couch.
Huang said creating such an assistant would have been unimaginable several years ago but is “utterly trivial” now that developers can rely on large language models rather than traditional programming tools to build apps and services.
In other words, the old way simply won’t cut it as AI grows more sophisticated and “reasons” on tasks that take multiple steps like these, Nvidia claims.
“The bottleneck is shifting from compute to context management,” Dion Harris, Nvidia’s senior director of high-performance computing and AI hyperscale solutions, said on a call with reporte