TLDR
- AMD reports Q3 earnings Tuesday with Wall Street expecting $8.76 billion revenue and $1.17 adjusted EPS
- Data Center segment projected at $4.18 billion, up 17.6% year-over-year, driven by EPYC and Instinct demand
- Morgan Stanley sees strong quarter ahead due to server demand and Intel supply issues
- OpenAI deal includes six gigawatts of AMD GPUs starting with MI450 servers in late 2026
- MI350 Series GPUs compete directly with NVIDIA B200 at lower cost
AMD releases third quarter earnings Tuesday after the bell. Analysts are optimistic about what the chip maker will deliver.
Wall Street expects revenue of $8.76 billion and adjusted earnings per share of $1.17 for the September quarter. Current quarter projections sit at $9.21 billion in revenue with $1.32 earnings per share, per FactSet data.
Advanced Micro Devices, Inc., AMD
Morgan Stanley analyst Joseph Moore kept his Equal weight rating and $246 price target Monday. But he’s looking for solid results.
“AMD should have a very strong data center quarter, given strong server demand and Intel’s [supply] constraints,” Moore noted. Intel recently confirmed demand is exceeding supply as companies accelerate computer and server upgrades.
Data Center Revenue Could Hit $4.18 Billion
The Data Center business looks poised to shine. Consensus estimates put Q3 Data Center revenue at $4.18 billion, marking 17.6% growth versus last year.
AMD’s EPYC processors continue gaining enterprise traction. Over 100 new AMD-powered cloud instances launched in Q2 2025, including Turin instances from Google and Oracle.
By Q2’s end, approximately 1,200 EPYC cloud instances were live globally. This expansion fuels enterprise cloud adoption.
HPE, Dell Technologies, Lenovo and Super Micro rolled out 28 new Turin platforms last quarter. The EPYC 4005 series targets small and medium businesses plus hosted IT customers.
OpenAI Partnership Validates AI Strategy
AMD secured a major OpenAI deal last month. The agreement makes AMD a key supplier for OpenAI’s AI infrastructure buildout.
OpenAI will deploy six gigawatts of AMD GPUs. The first gigawatt of rack-scale MI450 GPU servers arrives in late 2026.
AMD’s Instinct MI350 Series GPUs launched in June 2025. The company says MI355 matches or beats NVIDIA’s B200 in key training and inference workloads while offering lower cost and complexity.
Oracle is building an AI cluster with over 27,000 nodes. The project uses MI355X accelerators, fifth-gen EPYC Turin CPUs and Pollara 400 SmartNICs.
Competition Heats Up in AI Chips
AMD faces strong competition from NVIDIA and Broadcom. NVIDIA’s Hopper 200 and Blackwell platforms are seeing fast adoption as customers build out AI infrastructure.
Broadcom benefits from networking product demand and custom AI accelerators. The company expects higher XPU demand in late 2026 as hyperscalers shift focus to inference.
AMD’s partner ecosystem includes Cohere, IBM, Google, HPE, Dell Technologies, Lenovo and Super Micro. Tuesday’s earnings will show whether this network and the EPYC momentum translated into Q3 results that meet or exceed Wall Street’s expectations for the data center business.