Advanced Micro Devices reported first quarter 2026 revenue of $10.25 billion on May 5, up 38% year on year and ahead of analyst expectations of $9.89 billion. Non-GAAP earnings per share reached $1.37, up 43% year on year against an expected $1.29. The result sent AMD stock up 16% in after-hours trading, reflecting investor confidence in the company’s data center AI trajectory. The AMD data center revenue Q1 2026 figure of $5.8 billion, up 57% from $3.67 billion a year earlier, is now the primary growth engine of the business and marks the clearest evidence yet that hyperscalers are treating AMD as a serious alternative to Nvidia for both CPU and GPU workloads. Data center segment operating margin expanded from 25% to 28%, with operating income reaching $1.6 billion — a margin profile that reflects the premium economics of AI infrastructure relative to AMD’s consumer-facing businesses.
The quarter’s most strategically significant disclosure was the confirmation of Meta’s commitment to deploy up to 6 gigawatts of AMD Instinct GPUs across multiple product generations, with the first 1 gigawatt deployment using a custom GPU based on the MI450 architecture. Meta will also serve as a launch customer for AMD’s sixth-generation EPYC CPUs under its Venice architecture, representing a dual commitment across both the CPU and GPU product lines. Lisa Su confirmed separately that AMD’s October 2025 deal with OpenAI to deploy 6 gigawatts of GPUs starting in the second half of 2026 is “going really well,” with visibility now established down to which specific data centers the GPUs will be installed in. Looking ahead, AMD guided Q2 2026 revenue to $11.2 billion, plus or minus $300 million, ahead of prior consensus estimates.
What the Numbers Signal for the Competitive Landscape
The $5.8 billion data center quarter represents a run rate that, if sustained and grown at AMD’s stated targets, puts the company on a path to tens of billions of dollars in annual data center AI revenue by 2027. Su confirmed on the earnings call that AMD has “strong and increasing confidence” in reaching that level and in exceeding its long-term growth target of greater than 80% annually in the data center segment. More than 1,600 public EPYC cloud instances are now available globally, up approximately 50% year on year across AWS, Google Cloud, Microsoft Azure, and Tencent Cloud — evidence that AMD’s competitive position extends well beyond GPU workloads into the server CPU market where it continues to take share from Intel.
As covered in our analysis of the custom silicon AI accelerator race entering its most consequential phase, the competitive GPU market emerging in 2026 is creating structural changes in how hyperscalers procure infrastructure, price AI services, and evaluate vendor relationships that will compound for years.
Why the MI450 Deployment Is the Forward Catalyst
The MI450 architecture, which forms the basis of AMD’s custom GPU for Meta’s 6 gigawatt deployment, represents AMD’s most consequential competitive move against Nvidia’s Blackwell architecture. The Helios production ramp begins initial volume in Q3 2026 and scales significantly in Q4, establishing the second half of 2026 as the critical execution window for AMD’s AI infrastructure ambitions. Lisa Su acknowledged supply chain constraints — tightness in data center buildouts and component availability — on the earnings call, but framed them as factors AMD can navigate given its existing visibility into deployment timelines. The combination of the Meta 6 gigawatt MI450 commitment and the OpenAI 6 gigawatt agreement gives AMD a contracted forward pipeline of 12 gigawatts across two of the most strategically significant AI developers in the world — a position that would have been unimaginable twelve months ago.
