AMD locks Microsoft Helios deal as AI revenue hits $5.8B

AMD's Helios rack-scale platform will power AI inference across Microsoft Azure, expanding the chipmaker's revenue beyond individual components to full systems — a shift reflected in its $5.8 billion data center business, which grew 57 percent from a year earlier.

"These results mark a clear inflection in our growth trajectory and a structural shift in our business," Lisa Su, chief executive officer of AMD, said on the company's Q1 earnings call.

AMD expects to begin shipping Helios systems, including to Microsoft, in the second half of 2026. The deal follows similar commitments from Meta Platforms Inc. and OpenAI, each pledging up to 6 gigawatts of AMD Instinct GPUs. Server CPU revenue is projected to grow more than 70 percent year over year in the second quarter, with AMD doubling its 2030 server CPU total addressable market estimate to more than $120 billion.

AMD shares have surged about 158 percent year to date, giving the company a market capitalization of about $900 billion. The stock trades at 87 times forward earnings, a premium to the sector median of 32 times, reflecting investor expectations that the Helios platform and hyperscale partnerships will sustain growth against Nvidia Corp.'s dominant 85 percent share of the AI accelerator market.

Helios Expands AMD's Revenue Per Rack

The Helios platform combines AMD's Instinct MI455X GPUs, sixth-generation EPYC Venice CPUs, Pensando networking chips and ROCm software into a single rack-scale system. For Microsoft, the deployment will support frontier AI inference, Azure AI services and customer workloads, while Azure also plans to introduce new EPYC-powered virtual machine series and expand Pensando data processing units for networking performance.

The platform expands AMD's total addressable market. Where the company previously sold individual chips, Helios allows it to capture revenue across the entire server rack — GPUs, CPUs, networking and software. Su told investors she sees "a clear path to exceed our long-term financial targets, including delivering more than $20 in EPS over the strategic time frame."

Hyperscale Momentum vs. Nvidia's Moat

AMD's customer roster now includes three of the largest AI spenders. Beyond Microsoft, Meta committed to up to 6 gigawatts of Instinct GPUs, with the first gigawatt built on the custom MI450 design. OpenAI signed on as a core preferred partner for another 6 gigawatts. Anthropic agreed to deploy up to 2 gigawatts of AMD Instinct MI450 Series GPUs in Helios systems, with AMD considering an equity investment of as much as $5 billion. Rackspace Technology signed a phased rollout of 30 megawatts of AMD AI compute across its global data centers starting in late 2026.

The wins come as Nvidia maintains its grip on the AI accelerator market with roughly 85 percent share. Nvidia's data center revenue reached $75.2 billion in the first quarter of fiscal 2027, up 92 percent from a year earlier. Broadcom Inc., another competitor in custom AI chips, reported AI semiconductor revenue of $10.8 billion in its second quarter, up 143 percent year over year.

AMD's opportunity lies in the market's expansion. The total addressable market for AI accelerators is estimated to exceed $200 billion in 2026, leaving room for multiple winners. AMD's data center revenue of $5.8 billion represents about 5 percent to 7 percent of that market, according to analyst estimates.

AMD reports second-quarter earnings on Aug. 4, with analysts expecting earnings per share of $1.35, up from $0.27 a year earlier — 400 percent growth. UBS analyst Timothy Arcuri raised his price target ahead of AMD's Advancing AI event and called the stock his top semiconductor pick. Jefferies analyst Blayne Curtis rates AMD a buy with a $615 target. Of 45 analysts surveyed, the consensus is a strong buy with an average price target of $543.83, implying about 1.5 percent upside from current levels.

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