Chipmaker AMD is taking aim at rival Nvidia with its latest hardware release of a rack-scale system designed to meet the computing needs of the world’s largest AI labs.
At the company’s packed Advancing AI conference in San Francisco on Thursday, AMD Chairman and CEO Dr. Lisa Su touted its new AI rack system, known as Helios, and its growing list of customers, including Microsoft, as the company prepares to ship later this year. Hsu also touted the company’s latest chip, designed to feed the compute-hungry dragon that is the AI industry.
Rack systems combine many processors into one high-power unit. They are built for data centers to train and run AI models and other compute-intensive workloads.
Su called Helios the technology industry’s “highest performing AI rack,” adding that it was “built to train and run the world’s most demanding frontier models at scale.” The company says the system will be deployed at gigawatt scale by major AI companies.
Nvidia has historically dominated this market with rack-scale systems from Vera Rubin and Grace Blackwell. It’s clear that AMD is trying to get in on the action. And Helios’ performance metrics appear to outperform Vera Rubin in a number of metrics, giving her a real chance, the Register reported.
Helios was announced in 2025 and unveiled on stage at CES 2026 in January, but it already has some big-name customers including OpenAI, Meta, Oracle, Anthropic, and Microsoft, all of whom plan to deploy the system. Microsoft CEO Satya Nadella said Monday that the company will expand its Azure infrastructure with Helios. Meanwhile, Anthropic and AMD on Wednesday announced a strategic partnership to deploy up to 2 gigawatts of GPUs through new rack systems.
AMD also announced Thursday its Venice-X CPUs, which are designed for data centers and are designed to handle high computing workloads. Venice-X is scheduled to launch in 2027.
In his talk, Hsu commented on the trajectory of the chip industry, arguing that by 2030, chips that power AI will become a large part of the overall computing market. That’s because the industry is “seeing a significant shift in computing demands,” largely due to the rise of agent-based AI, he said.
“When you ask an agent to do something, there are actually dozens of steps that have to be inferred, tools have to be called, data has to be accessed, and it has to be done over and over again until the problem is solved. So you need a lot of GPUs to do all of that,” the executive said.
“We currently expect the AI accelerator market to reach approximately $1.4 trillion by 2030,” Su said. “What this means is that by the end of the decade, the AI accelerator market will approach the size of the entire semiconductor market today.”
“We expect GPUs to take a large portion of that market as algorithms are still in their infancy and workloads continue to change, which favors programmability across the silicon ecosystem,” she added.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
