Chipmaker AMD is taking purpose at rival Nvidia with its newest {hardware} launch of a rack-scale system designed to fulfill the computing wants of the world’s largest AI labs.
On the firm’s packed Advancing AI convention in San Francisco on Thursday, AMD Chairman and CEO Dr. Lisa Su touted its new AI rack system, often called Helios, and its rising record of shoppers, together with Microsoft, as the corporate prepares to ship later this yr. Hsu additionally touted the corporate’s newest chip, designed to feed the compute-hungry dragon that’s the AI business.
Rack methods mix many processors into one high-power unit. They’re constructed for knowledge facilities to coach and run AI fashions and different compute-intensive workloads.
Su known as Helios the expertise business’s “highest performing AI rack,” including that it was “constructed to coach and run the world’s most demanding frontier fashions at scale.” The corporate says the system will likely be deployed at gigawatt scale by main AI corporations.
Nvidia has traditionally dominated this market with rack-scale methods from Vera Rubin and Grace Blackwell. It is clear that AMD is making an attempt to get in on the motion. And Helios’ efficiency metrics seem to outperform Vera Rubin in numerous metrics, giving her an actual probability, the Register reported.
Helios was introduced in 2025 and unveiled on stage at CES 2026 in January, however it already has some big-name prospects together with OpenAI, Meta, Oracle, Anthropic, and Microsoft, all of whom plan to deploy the system. Microsoft CEO Satya Nadella stated Monday that the corporate will develop its Azure infrastructure with Helios. In the meantime, Anthropic and AMD on Wednesday introduced a strategic partnership to deploy as much as 2 gigawatts of GPUs via new rack methods.
AMD additionally introduced Thursday its Venice-X CPUs, that are designed for knowledge facilities and are designed to deal with excessive computing workloads. Venice-X is scheduled to launch in 2027.
In his speak, Hsu commented on the trajectory of the chip business, arguing that by 2030, chips that energy AI will grow to be a big a part of the general computing market. That is as a result of the business is “seeing a step change in computing demand,” largely because of the rise of agent-based AI, he stated.
“Once you ask an agent to do one thing, there are literally dozens of steps, it has to purpose, it has to name instruments, it has to entry knowledge, it has to do it over and over till the issue is solved, so that you want a variety of GPUs to do all of that,” the manager stated.
“We at the moment anticipate the AI accelerator market to succeed in roughly $1.4 trillion by 2030,” Su stated. “What this implies is that by the top of the last decade, the AI accelerator market will method the dimensions of the complete semiconductor market in the present day.”
“We anticipate GPUs to take a big portion of that market as algorithms are nonetheless of their infancy and workloads proceed to vary, which favors programmability throughout the silicon ecosystem,” she added.
In the event you purchase via hyperlinks in our articles, we might earn a small fee. This doesn’t have an effect on editorial independence.
