AWS, Google, and Microsoft custom-silicon investments compound — Trainium, TPU, and Maia chips collectively encircle Nvidia at the hyperscaler in-house deployment layer
Amazon Trainium, Google TPU, and Microsoft Maia in-house AI chip programs collectively represent a structural competitive challenge to Nvidia at the hyperscaler-deployment layer. The H2 2026 picture: hyperscalers increasingly use in-house silicon for internal workloads while sourcing Nvidia for customer-facing workloads, eroding Nvidia's hyperscaler-procurement leverage.
The substantive piece is the hyperscaler in-house-vs-Nvidia bifurcation. AWS Trainium 2, Google TPU v6, and Microsoft Maia 2 in-house chip programs have reached production-deployment maturity through H1 2026. The hyperscaler procurement pattern that's emerging: in-house silicon for first-party workloads (foundation model training, infrastructure inference) plus Nvidia for customer-facing workloads (Bedrock, Azure OpenAI, Vertex AI services to enterprise customers). Nvidia retains the customer-facing revenue while losing the first-party allocation.
The competitive read for Nvidia's H2 2026 strategic position is that the encirclement is real but not existential. The $4.5T market cap reflects ongoing dominant position in the customer-facing AI-infrastructure layer; the in-house-silicon encirclement reduces growth velocity but doesn't reverse it. The 18-24 month outlook depends on whether hyperscaler in-house silicon expands beyond first-party to customer-facing offerings.
Windows News — AI Chip Wars 2026: Amazon, Google, and Microsoft Surround Nvidia with Custom Silicon → · Al Jazeera — Nvidia unveils new chip to bring AI directly to personal computers →