AMD Challenges Nvidia's AI Dominance with Helios Rack-Scale System

AMD has fired its latest shot in the battle for AI computing supremacy, unveiling a rack-scale system called Helios that the company says will deliver the highest performance of any AI rack on the market.

Speaking at the sold-out Advancing AI conference in San Francisco on Thursday, AMD Chair and CEO Dr. Lisa Su presented Helios as a system engineered to train and run the world's most demanding frontier AI models at massive scale. The system is scheduled to begin shipping later this year.

Helios Takes Aim at Nvidia's Stronghold

Rack systems bundle large numbers of processors into unified, high-powered units designed for data centers, where they handle the training and execution of AI models and other compute-intensive workloads. Nvidia has long controlled this space with its Vera Rubin and Grace Blackwell rack-scale systems, but AMD is now making a serious push to claim a share of the market.

According to The Register, Helios outperforms Nvidia's Vera Rubin across several metrics, suggesting AMD's offering could be genuinely competitive rather than merely aspirational. Su described Helios as the tech industry's "highest-performance AI rack," built specifically for leading AI companies deploying at gigawatt scale.

A Growing Roster of High-Profile Customers

Helios, which was first revealed in 2025 and demonstrated onstage at CES 2026 in January, has already secured commitments from several major players in the AI space. OpenAI, Meta, Oracle, Anthropic, and Microsoft have all announced plans to deploy the system.

Microsoft CEO Satya Nadella confirmed on Monday that the company would expand its Azure infrastructure using Helios. Separately, Anthropic and AMD announced a strategic partnership on Wednesday to deploy up to two gigawatts of GPUs through the new rack system.

New CPU and a Trillion-Dollar Vision

Alongside Helios, AMD introduced its Venice-X CPU on Thursday, a data center processor built for high-computing workloads. The Venice-X is expected to launch in 2027.

During her remarks, Su offered a broad perspective on where the chip industry is headed. She predicted that by 2030, AI chips would constitute an enormous portion of the overall computing market, driven by what she described as a "step change in compute demand" stemming largely from the rise of agentic AI.

Su explained that agentic AI workloads are far more resource-intensive than traditional AI tasks because they involve multi-step reasoning, tool usage, data access, and iterative problem-solving — all of which require substantial GPU capacity.

"When you ask the agent to do something, it actually has dozens of steps, and it has to reason, and it has to call tools, and it has to access data, and it has to keep doing it over and over until it solves the problem, and so you need lots of GPUs to do all that," Su said.

She projected that the AI accelerator market would reach approximately $1.4 trillion by 2030, a figure that would bring it close to the size of the entire semiconductor market as it stands today. GPUs, she added, are expected to account for the vast majority of that market, given that current AI algorithms remain in their early stages and workloads continue to evolve — a trend that favors programmability in the silicon ecosystem.

"We do expect that GPUs are going to make up the vast majority of that market because the algorithms are still very much in their infancy, and we're still continuing to see the workloads change, and that favors programmability in the overall silicon ecosystem," she said.

As AMD prepares to ship Helios and expand its footprint in the AI hardware race, the competition with Nvidia is set to intensify. With major customers already lined up and a clear vision for the decade ahead, AMD is signaling that it intends to be a formidable player in the AI infrastructure space. What's your take on the AMD-Nvidia rivalry? Share this article and join the conversation.

Source: TechCrunch