Companies/AMD/infrastructure
scalinginfrastructure

Helios rack-scale AI systems

In production per the Q2 2026 call; initial shipments guided to later in Q3 2026 and a larger ramp in Q4 2026 into 2027.

AMD market data

The numbers

Anthropic agreementup to 2 GW of MI450 Series GPUs in Helios racks; first GW starts H1 2027Announced Jul 22, 2026 · source ↗
Company revenue guidance (Helios ramp drives H2)Q3 2026 revenue about $13.0B, plus or minus $300MQ3 2026 guidance · +41% y/y at midpoint · source ↗

What it is

AMD's rack-scale AI system built around Instinct MI455X GPUs, 6th Gen EPYC Venice CPUs, Pensando networking and ROCm software. It is the path for hyperscalers and AI labs to buy complete systems rather than parts. The Q2 2026 release names OpenAI, Meta and Anthropic among the AI labs and cloud providers deploying Helios.

Why it matters to the stock. Helios decides how much of AI rack revenue AMD captures and how fast the Instinct ramp shows up in reported Data Center revenue. Supply constraints on HBM and advanced packaging are the main risk to timing.

News

AMD introduces Helion agentic AI rack blueprint, racks planned for November

AMD unveiled Helion, an open blueprint for agentic AI infrastructure built on the Helios rack approach and contributed to the Open Compute Project.

Product

AMD Q2 2026 revenue rises 50% to $11.5B; Data Center more than doubles

AMD reported Q2 2026 revenue of $11.54 billion, up 50% year over year, with Data Center revenue of $6.72 billion, up 107%.

Earnings

AMD adds MI350P PCIe GPU and 256-core EPYC 9996 at Advancing AI

Alongside the Helios launch, AMD introduced the Instinct MI350P, a PCIe card meant to add AI acceleration to existing servers, and the EPYC 9996 with 256 cores and 512 threads.

Product

AMD and Cerebras pair Helios racks with Cerebras WSE for inference

AMD and Cerebras announced a disaggregated inference setup in which Helios handles prompt processing and the Cerebras wafer-scale engine handles token generation.

Deal

OpenAI says Helios comes online from Q4 2026 and joint kernel work deepens

At the Advancing AI keynote, OpenAI said it expects to bring AMD Helios online through several deployment partners in the second half of 2026, with deployments accelerating through 2027.

Deal

AMD and Anthropic agree to deploy up to 2 GW of MI450-series GPUs

Anthropic will deploy up to 2 gigawatts of Instinct MI450 Series GPUs in AMD Helios racks, with the first gigawatt starting in the first half of 2027.

Deal

AMD and Microsoft expand Azure deal to deploy Helios racks

AMD and Microsoft announced an expanded collaboration to deploy AMD Helios rack-scale systems on Azure, mainly for frontier model inference.

Deal

AMD grants Meta a performance warrant for up to 160 million AMD shares

As part of the Meta agreement, AMD issued Meta a performance-based warrant for up to 160 million shares of common stock.

Capital

AMD and TCS plan 200 MW Helios rack-scale AI data center blueprint for India

AMD and Tata Consultancy Services said they will co-develop a rack-scale AI infrastructure design based on the Helios platform for India, announced during the New Delhi AI Impact Summit.

Deal

AMD previews Helios rack, MI455X GPUs and MI500 roadmap at CES 2026

At CES 2026 Lisa Su previewed Helios, AMD's rack-scale AI platform built on Instinct MI455X GPUs and Venice EPYC CPUs.

Product