ticker: AMD
step: 01
generated: 2026-05-12
source: quick-research
Advanced Micro Devices, Inc. (AMD) — Business Overview
Business Description
AMD is a leading designer of high-performance computing, graphics, and AI silicon. The company has transformed over the past decade under CEO Lisa Su from a beleaguered PC competitor into the #2 data center compute provider (behind Intel in x86 server CPU, and the only credible alternative to Nvidia in AI accelerators). The 2022 Xilinx acquisition ($49B) added FPGA + adaptive SoCs (now the Embedded segment). With the Helios rack-scale platform launch (mid-2026) and MI400 series, AMD is transitioning from component supplier to full-stack AI infrastructure provider.
Revenue Model
- Data Center (~50%+ of revenue): EPYC server CPUs (Zen 5 "Turin", upcoming Zen 6) + Instinct AI accelerators (MI300/MI325/MI350/MI400) + Pensando DPUs
- Client (~25%): Ryzen consumer CPUs for PCs + AI PC NPU integration
- Gaming (~10%): Radeon GPUs + semi-custom (PlayStation 5, Xbox Series X/S)
- Embedded (~15%): Xilinx FPGAs + Versal adaptive SoCs across automotive, aerospace, industrial, test & measurement, broadcast
Products & Services
Data Center
- EPYC server CPUs: 5th-gen "Turin" (Zen 5); ~35%+ x86 server market share
- Instinct MI300X / MI325X: Current GPU generation in production
- Instinct MI350 series: Launched mid-2025 (CDNA 4 arch); MI355 matching/exceeding NVDA B200 in key workloads
- Instinct MI400 series: Launching mid-2026; HBM4; targeting NVDA GB200 successor parity
- Helios rack-scale system: 31TB HBM4, 1.4PB/s bandwidth, 2.9 ExaFLOPS FP4 inference, 1.4 ExaFLOPS FP8 training
- Pensando DPU: Data center networking + SmartNIC
- ROCm software: Open AI/HPC software stack (alternative to CUDA)
Client
- Ryzen processors: Zen 5 desktop + laptop; AI 300 series with NPU
- Ryzen AI: On-device AI PC platform
Gaming
- Radeon RX 9000 GPUs: Consumer graphics
- Semi-custom: PlayStation 5 / Xbox Series X/S APUs (declining in 7th year of console cycle)
- FSR4 Redstone: AI-driven upscaling
Embedded
- Xilinx FPGAs / Versal SoCs: ~$17B FY25 design wins; $50B+ since Xilinx acquisition
- Aerospace/Defense, Test & Measurement, Industrial, Automotive, Wired/Wireless
Customer Base & Go-to-Market
- Hyperscalers (data center growth driver): Microsoft (Azure), Meta (6GW MI450 deal Feb 2026), OpenAI (6GW deployment beginning 2H 2026), Oracle, Google, AWS
- Enterprise: Dell, HPE, Lenovo, Supermicro server OEMs
- Consumers: PC OEMs (Dell, HP, Lenovo, Asus, etc.); DIY enthusiasts
- Game console: Sony, Microsoft (semi-custom)
- Industrial / embedded: Tier-1 industrial OEMs, automotive Tier 1, defense contractors
Competitive Position
AMD is the only credible challenger to Nvidia in AI training/inference accelerators and the leader in x86 server CPU share gains. Moats: (1) Zen architecture leadership in CPU + Infinity Fabric chiplet design, (2) Xilinx FPGA franchise with $50B+ embedded design wins, (3) deep OEM relationships, (4) ROCm stack maturity catching up to CUDA. Persistent challenges: (1) Nvidia's CUDA software moat with 15+ years and millions of trained developers, (2) ASIC competition for AI workloads (Broadcom, Marvell custom silicon), (3) console cycle decay, (4) export controls limiting China AI exposure.
Key Facts
- Founded: 1969 (Sunnyvale, CA)
- Headquarters: Santa Clara, CA
- Employees: ~28,000
- Exchange: NASDAQ
- Sector / Industry: Technology / Semiconductors
- Market Cap: ~$360B (May 2026)
- CEO: Dr. Lisa Su (since 2014)
- Major deals 2025-2026: OpenAI 6GW MI400/450 (Oct 2025); Meta 6GW MI450 (Feb 2026)