#2 AI accelerator (MI-series) + EPYC server CPUs — the credible second source to NVIDIA
AI exposure 43% — derived from reported segment revenue, not asserted.
Silicon Layer → Semiconductor Devices · figures rebased to Q2 2026, reported 2026-08-04
EPYC server CPUs + Instinct AI accelerators (MI300/MI400/MI450 Helios) leads 4 businesses in this sub-tier that together are 100% of revenue, so Semiconductor Devices is effectively the whole company rather than one leg of it.
AMD (~$25.8B FY2024, $11.54B Q2 2026 alone, +50% YoY) — CPUs/GPUs/FPGAs. Data Center 58% (EPYC CPUs + Instinct AI accelerators = primary B1, the AI core; Helios rack-scale now ramping with four anchor hyperscalers — Meta, OpenAI, Anthropic, Microsoft Azure) + Client 27% (Ryzen) + Embedded 8% (Xilinx FPGAs) + Gaming 7%. All B1 Semiconductor Devices. Primary B1. NVDA is the main rival. ZT Systems acquisition adds rack-scale AI systems.
Moat. Strong x86 server-CPU franchise (EPYC) + the only credible GPU alternative to NVIDIA (MI300/MI350). Chiplet design lead.
Bottleneck / pricing power. Pricing power limited by NVIDIA dominance in GPUs, but the "second source" premium is real — hyperscalers want AMD to succeed to escape NVIDIA.
Role in the AI stack. The challenger compute layer — CPU strength + GPU optionality.
All 174 companies · What moved this quarter · How the map is built