AMD announced one of the most important launches in its artificial intelligence strategy by unveiling Helios, its first complete rack-scale AI system. The platform enters the market to compete with Nvidia's solutions, which currently dominate data center infrastructure focused on training and inference of AI models.
Microsoft confirmed that it will use Helios in its Azure data centers, becoming one of the main customers of the new platform. The company now joins a list that also includes Meta, OpenAI, Oracle, and Tata Consultancy Services, organizations that are expanding their investments in computing capacity to meet the growing demand for artificial intelligence.
The first Helios deliveries are expected to take place in 2026. AMD did not disclose the value of the contracts or the number of systems acquired by each customer, but the announcement was enough to boost its shares by more than 4% during the trading session. Microsoft's shares also rose.
“We are expanding Azure's infrastructure portfolio with AMD Helios to offer customers the performance, scalability, and choices they need to build and run the next generation of AI applications,” Microsoft CEO Satya Nadella wrote in a press release.
According to Microsoft, Helios will be used primarily to accelerate inference processes for advanced artificial intelligence models and expand the services offered by Azure. At the same time, the company will also add two new computing instances equipped with AMD's Venice processors, aimed at agent-based AI applications, processing large volumes of data, and semiconductor development.
The partnership between AMD and Microsoft has already extended for several years. The chipmaker supplies processors for Surface line devices and Xbox consoles. In 2023, Microsoft was also among the first companies to adopt the MI300X GPU, developed by AMD to compete in the artificial intelligence accelerator segment.
Helios represents an important shift in the company's strategy because it brings together, in a single system, Instinct GPUs, CPUs, high-speed networking, and a software layer optimized for AI workloads. The goal is to deliver an integrated solution capable of reducing operating costs and simplifying deployment in large data centers.
“We are very focused on delivering the best total cost of ownership, the lowest cost per token, all in,” said Forrest Norrod, head of AMD's data center division. “And our customers are telling us that we are achieving that goal.”
Today, Nvidia accounts for more than 95% of the data center GPU market, while AMD has a share close to 4,5%. Even so, analysts believe that the arrival of Helios could significantly expand this share in the coming years.
Daniel Newman, CEO of Futurum Group, believes that AMD has the potential to capture between 20% and 25% of this market, a segment that moves hundreds of billions of dollars and continues to receive record investments from the world's largest technology companies.