According to a report by technology publication TechCrunch, global chipmaker AMD is taking direct aim at its chief rival, Nvidia, with the launch of its latest hardware innovation: a complete rack-scale server system known as Helios. This highly advanced system has been engineered specifically to satisfy the massive computational demands of the world's largest and most prominent artificial intelligence laboratories. During the company's sold-out "Advancing AI" conference, held on Thursday in San Francisco, AMD Chair and CEO Dr. Lisa Su showcased and promoted the new Helios system. Su highlighted the company’s expanding roster of customers for this system, which includes technology giant Microsoft, as AMD prepares to begin shipping the hardware later this year. Alongside this announcement, Su also utilized the conference to promote the company’s newest chips, designed to power an AI industry that consumes vast amounts of processing power—an industry she colorfully described as a compute-hungry dragon.
What is a Rack-Scale System and How Does It Work in Data Centers?
Rack-scale systems integrate and unify numerous diverse processors into a single, cohesive, and exceptionally high-performance unit. These complex hardware systems are designed and built specifically for data centers, where they are utilized to train and run complex artificial intelligence models, as well as to handle other highly intensive, demanding workloads that require high-intensity computing.
During her keynote address at the conference, Dr. Lisa Su defined the Helios system as the "highest performance AI rack" in the entire technology industry. Su went on to explain that the innovative system was purpose-built to train and run the most demanding and advanced "frontier models" currently in existence on a massive and extraordinarily large scale. According to AMD's official statement, the Helios system is set to be deployed and operated by leading AI companies at a gigawatt-scale.
Competition Against Market Leaders and Helios Performance Metrics
Historically, Nvidia has dominated and continues to heavily dominate this market for data center server systems, primarily through its own rack-scale offerings known as Vera Rubin and Grace Blackwell. Now, AMD is making it explicitly clear that it intends to actively participate and directly compete in this market segment.
According to a report published by the professional technology publication The Register, performance metrics and benchmarks for the Helios system indicate that AMD has a genuine and tangible opportunity to challenge the market leader. The report notes that the Helios system actually surpasses Nvidia’s Vera Rubin system across several significant performance metrics.
The Helios system, which was first revealed to the public in 2025 and physically demonstrated on stage this past January at the global technology exhibition CES 2026, already boasts a number of prominent and well-recognized industry customers. This impressive list includes major giants such as OpenAI, Meta, Oracle, Anthropic, and Microsoft, all of which have defined plans to deploy and integrate the new system into their technological infrastructures.
Strategic Partnerships: Microsoft, Anthropic, and Other Giants
The announcements surrounding the Helios system were accompanied by several significant business moves from AMD’s key partners. Microsoft CEO Satya Nadella declared on Monday that the company intends to expand its Azure cloud platform infrastructure by integrating and deploying AMD's new Helios system.
In parallel, Anthropic and AMD announced a broad, large-scale strategic partnership last Wednesday. Under this collaboration, the companies plan to deploy graphics processing units (GPUs) with a massive capacity of up to two gigawatts, powered by the newly developed AMD rack system.
Unveiling the Venice-X Processor for Heavy Data Center Workloads
In addition to presenting the Helios system, AMD leveraged its stage on Thursday to introduce its new Venice-X processor. This CPU is specifically designed and engineered for use in large data centers to manage and process ultra-high-intensity computing workloads. According to the company’s announcement, the Venice-X processor is scheduled to officially launch and hit the market during the year 2027.
The Future of the Chip Market up to 2030 and the Impact of Agentic AI
During her address and commentary at the conference, Dr. Lisa Su addressed the growth trajectory and future of the global chip industry. Su asserted and estimated that by the year 2030, chips powering AI applications will become a massive and highly dominant segment of the overall computing market. According to her, this trend is occurring because the technology industry is currently witnessing a "step change in compute demand," which is heavily driven by the rise and breakthrough of agentic AI.
Su explained the operational nature of these AI agents in detail, highlighting the growing need for processing power: "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."
Following these remarks, Su noted: "We’re now expecting that by 2030, the AI accelerator market is going to reach about $1.4 trillion." The CEO emphasized the deep significance of this figure, pointing out that by the end of the decade, the AI accelerator market will approach the size and total volume of the entire global semiconductor market as it is valued today.
Finally, Dr. Lisa Su explained why graphics processors will continue to lead this trend: "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."