AMD doubles down on AI infrastructure with Helios launch to rival Nvidia

Advanced Micro Devices (AMD) is set to unveil a new lineup of AI hardware on Thursday at an event in downtown San Francisco, as it steps up efforts to challenge Nvidia in the booming AI infrastructure market. The launch comes as demand for data centre chips continues to surge, driven by the rapid growth of inference computing.

AMD is expected to showcase its next-generation data centre hardware, including its first server rack system, Helios, which the company is positioning as a rival to Nvidia's rack-scale AI systems. The launch comes as Nvidia prepares to roll out the second generation of its own rack architecture later this year.

More from Tech

Beyond its hardware announcements, AMD has agreed to invest up to $5 billion in Anthropic under an infrastructure agreement covering tens of billions of dollars' worth of AI systems. Anthropic plans to deploy up to two gigawatts of AI computing capacity using AMD's Instinct MI450-series accelerators, with the first gigawatt expected to come online in the first half of 2027.

AMD's investment will be tied to Anthropic meeting certain deployment milestones, although the companies have not disclosed the specific conditions attached to the agreement.

Financial incentives tied to chip deployments

The Anthropic agreement follows a series of similar infrastructure deals. In October, AMD agreed to supply OpenAI with systems supporting up to six gigawatts of capacity and issued warrants that could allow the ChatGPT maker to acquire about a 10% stake in the chipmaker. In February, AMD finalised another agreement with Meta covering up to six gigawatts of GPU capacity, which also included performance-based warrants linked to shipment and purchase targets.

As part of its deal with Anthropic, the AI startup will deploy AMD's Helios systems featuring Instinct MI455X GPUs from the MI450 series, EPYC Venice processors, Pensando networking, and the ROCm software platform. The deployment builds on Anthropic's existing use of AMD's earlier MI350-series accelerators.

According to a report by The Wall Street Journal, Anthropic plans to install some of the AMD systems in its own data centres, while the remaining infrastructure will be hosted by cloud providers and specialist AI infrastructure companies.

Anthropic's mix of hardware parternships

Anthropic is expanding its AI infrastructure through a mix of hardware partnerships, cloud agreements, and leased data centre capacity. Alongside its new AMD deal, the company will continue using Nvidia GPUs, Amazon Trainium processors, and Google TPUs to train and serve Claude.

Amazon remains Anthropic's primary cloud partner, providing up to five gigawatts of capacity and more than one million Trainium2 processors, while next-generation Trainium3 and Google TPU deployments are planned over the next two years. Anthropic has also secured computing capacity at SpaceX's Colossus facilities and is reportedly in talks to lease infrastructure from Meta.