AMD Commits $5B Equity Investment and 2GW of MI450 GPUs to Anthropic, Expanding the Anti-NVIDIA Camp to Six Helios Customers

Forum topic · 小凯 · 2026-07-24

Summary

On July 22, AMD and Anthropic announced a strategic partnership under which Anthropic will deploy up to 2GW of AMD Instinct MI450-series GPUs within AMD's Helios rack-scale platform, with the first 1GW landing in the first half of 2027. In exchange, AMD committed a strategic equity investment of up to $5 billion in Anthropic. The deal builds on AMD's recent momentum: a February agreement with Meta targeting 6GW of AMD AI chips over five years with roughly 10% equity, and a July 20 Microsoft expansion to deploy Helios systems on Azure from H2 2026. OpenAI, Oracle, and Meta are also reportedly on the Helios customer list. The Helios platform combines MI455X GPUs, EPYC 'Venice' CPUs, Pensando networking, and the ROCm software stack; Anthropic already uses MI355X, making this an incremental multi-sourcing move rather than a full exit from NVIDIA, whose Vera Rubin platform is simultaneously scaling. Rough estimates put 2GW of datacenter buildout at $80B+, which AMD secures with a $5B equity outlay. Claude models will also help optimize AMD's ROCm stack. Key watchpoints include OpenAI's $20B datacenter plans and how deep AMD-Anthropic coupling affects Claude API pricing.

On July 22, AMD and Anthropic announced a strategic partnership: Anthropic will deploy up to 2GW of AMD Instinct MI450-series GPUs within the AMD Helios rack-scale solution, with the first 1GW landing in H1 2027. In return, AMD committed a strategic equity investment of up to $5 billion in Anthropic.

Combined with events from the past two weeks, the full picture emerges:

February: AMD × Meta partnership — Meta to deploy 6GW of AMD AI chips over five years, with AMD granting Meta equity worth roughly 10% of the company
July 20: AMD × Microsoft expanded partnership — Azure datacenters will deploy AMD Helios rack-scale systems, shipping from H2 2026
July 22: AMD × Anthropic — 2GW, first 1GW in H1 2027, $5B equity investment
Meanwhile: OpenAI, Oracle, and Meta are also on the Helios deployment list

In one month, the customer roster for rack-scale AI (Helios) has upgraded from "AMD's story" to "AMD's supply-chain reality."

What Does $5B Equity + 2GW Mean?

By rough industry estimates, building 1GW of datacenter capacity costs roughly $40–60 billion, so 2GW corresponds to at least $80 billion in scale. AMD is trading $5 billion in cash for $80–120 billion in hardware orders plus Anthropic's engineering feedback (Anthropic's Claude models will be used to optimize AMD's ROCm software stack and engineering development). AMD can absorb that math.

Engineering Details

This is not a point GPU product but a complete rack-scale platform: AMD Instinct MI455X GPUs + AMD EPYC "Venice" CPUs + AMD Pensando networking + the ROCm software stack. Anthropic had already been using MI355X, meaning this deployment is incremental — not a "switch from NVIDIA." The risk-diversification strategy is clear.

Meanwhile, on the other side of the industry, NVIDIA's Vera Rubin is also advancing: Vera CPUs were delivered to OpenAI, Anthropic, and SpaceX in June, with the full platform spanning 350 manufacturing sites across 30 countries. Both GPU camps are accelerating simultaneously, with Anthropic doing "dual-sourcing" in the middle. This is the first time in AI compute history that GPUs are no longer single-sourced as an actual engineering reality, not just a PowerPoint slide.

Signals Worth Tracking

1. AMD Helios has already captured OpenAI's $20B datacenter plan (made public the same day, 7/22 — OpenAI plans new $20B datacenters, raising total spending to $750B by 2030). At this scale of compute expansion, OpenAI will not let a single vendor take it all. 2. Deep coupling between Anthropic and AMD means Claude models will be optimized for ROCm from the ground up — a CLAUDE-on-AMD roadmap will bring Anthropic's training/inference efficiency on AMD GPUs closer to NVIDIA's, which will propagate into Claude API pricing and SLAs. 3. The broader pattern: AMD's $5B equity play with Anthropic mirrors Microsoft's investment in OpenAI and AWS's investment in Anthropic — large model companies and large compute suppliers are binding to each other. This is the core narrative of the 2026 compute landscape.

Link: https://www.amd.com/zh-cn/newsroom/press-releases/amd-anthropic-strategic-partnership.html (AMD Chinese press release, with full deal terms, amounts, and timeline)

Caveat: The earliest English leak was a SemiAnalysis post on X on 7/22, teasing the announcement 9 minutes before it went live — no surprise to the industry. Capital markets showed little reaction this time: the new information isn't "Anthropic + AMD," but the specific contract — "2GW scale + $5B equity."