GPU spot prices just doubled in seven months. Gavin Baker, chief investment officer at Atreides Management, flagged the jump on August 4, 2026, tracking how rental rates for the same cluster climbed from under $2 per hour to nearly $4 over that span. The disconnect is sharp: AI stocks tanked in July, yet infrastructure spending never slowed.
The real squeeze shows in the gap between spot and contract rates. Spot rentals now run at least twice as high as locked-in pricing, Baker observed in late July. H100 one-year contracts themselves jumped 40 percent, from $1.70 per hour in October 2025 to $2.35 by March 2026. That suggests hyperscalers still have room to push prices higher.
OpenAI, Anthropic, Google, and other AI companies kept infrastructure spending aggressive through the summer selloff. Memory supply will stay constrained through 2027, keeping pressure on both new and older GPU hardware. Baker expects hyperscaler operating cash flow to accelerate from 31 percent growth in Q1 2026 to 50 percent in Q2, driven mostly by repricing existing contracts at higher rates.
For decentralized compute networks like Render, Akash, and io.net, the math shifts when centralized spot rates double. These platforms pitch themselves as cheaper alternatives to cloud giants through tokenized GPU marketplaces. Rising centralized costs strengthen their value case. GPU-dependent mining also feels the pinch, since owned hardware now beats rented compute on margin when hourly rates hit $4.
This article is informational only and should not be taken as financial or investment advice.
