AI infrastructure spending stays strong as usage outruns price cuts

Conviction: 78% · Horizon: 2Y · 2026-07-06
Cheaper inference accelerates demand more than it erodes total spend

Token volumes are scaling rapidly even as average model prices drift lower, so aggregate infrastructure investment continues to rise. Premium U.S. model providers still capture the bulk of paid spend despite growing share of low-cost alternatives in raw usage.

GPU compute remains supply-constrained with rising rental economics

Rental rates for datacenter-class Nvidia accelerators continued to climb through mid-2026, including legacy A100 and H100 fleets, implying depreciation assumptions far below economic useful life and sustained demand from workload-specific chip selection.

Instrument Side Target Reason
NVDA Long We believe rising lease rates on A100, H100, and B200 capacity signal that AI workloads keep absorbing available compute faster than new supply depreciates legacy hardware, supporting durable revenue and pricing power for the leading accelerator vendor.

Themes

The content on this page is for informational purposes only and does not constitute financial advice. Stoquate is not a licensed financial advisor. Always conduct your own research and consult a qualified professional before making any investment decisions.