AMD launches Helios, Venice chips with $5B Anthropic deal to challenge Nvidia

AMD is taking direct aim at Nvidia's AI infrastructure dominance with a new generation of server racks, central processors, and a $5 billion investment in Anthropic.

Advanced Micro Devices on Thursday launched its first-generation Helios server racks and Venice central processing units in San Francisco, directly challenging Nvidia's dominance in the AI data center chip market with a $5 billion investment in Anthropic.

AMD is marketing Helios as a rival to Nvidia's second-generation rack product rolling out this year, according to Reuters. The company also formally launched its Venice CPU, built on the Zen 6 architecture, for data center inference workloads at the Moscone West event, where hundreds of executives and engineers gathered for technical presentations.

The company announced plans to sell up to two gigawatts of its Instinct MI450 chips to AI lab Anthropic beginning in the first half of 2027, a deal that also includes an investment of as much as $5 billion in the Claude maker. In October, AMD announced a multiyear deal with OpenAI that would bring in tens of billions of dollars in annual revenue while giving the ChatGPT creator the option to buy up to roughly 10% of the chipmaker.

The launches come as Nvidia released technical details about its Vera CPU this week, aiming to show that the chip, when combined with Nvidia's Rubin graphics processing unit, maximizes AI workload efficiency per watt of electricity. AMD displayed the Helios rack at booths from cloud providers Vultr and TensorWave, both of which operate data centers with AMD hardware.

Helios and Venice: Closing the Gap

AMD's Helios represents the company's first attempt at a full rack-scale system, a design approach Nvidia pioneered with its DGX and HGX platforms. The Venice CPU, built on the Zen 6 architecture, targets data center inference — the computation that occurs when a user queries a chatbot such as OpenAI's ChatGPT. AMD is offering both products as open alternatives to Nvidia's proprietary CUDA platform, which has long been the industry standard for AI training and inference.

The $5 Billion Bet on Anthropic

The Anthropic deal, valued at up to $5 billion in investment alongside the 2 GW chip supply agreement, reflects deepening ties between chipmakers and AI labs. AMD's separate agreement with OpenAI, which includes an option for the ChatGPT creator to acquire roughly 10% of the chipmaker, could generate tens of billions in annual revenue. These partnerships give AMD a foothold in the two most prominent AI labs, both of which have historically relied on Nvidia hardware.

Investor Implications

AMD shares are likely to face heightened scrutiny as investors assess whether the company can convert these product launches and partnership deals into meaningful market share gains against Nvidia. Nvidia's data center business generated more than $100 billion in annual revenue in its most recent fiscal year, giving it a commanding lead. AMD's ability to secure supply agreements with Anthropic and OpenAI suggests growing confidence among AI labs in AMD's roadmap, but the company must now deliver on production timelines and performance benchmarks to justify the deals.

This article is for informational purposes only and does not constitute investment advice.