Custom Silicon — Competitive Displacement Risk

NVIDIA's 85%+ share of AI training GPUs commands a valuation premium based on assumed persistence of hardware dominance.

Google's TPUs, Amazon's Trainium/Inferentia, Meta's MTIA, and Microsoft's Maia chips represent a structural threat:

If hyperscalers can handle 30-40% of their AI workloads with custom silicon, NVIDIA's total addressable market shrinks meaningfully.

The signal activates when hyperscalers provide quantitative metrics on custom silicon workload share in earnings calls — any 5%+ shift to custom silicon reduces NVIDIA's effective TAM by billions.

The competitive displacement risk is particularly relevant for inference workloads (more price-sensitive than training), which will eventually represent the majority of AI compute demand.