H100 rental rates climb 24% while the AI trade wobbles and lenders back off
Renting an Nvidia H100 now runs about $3.70 an hour, up 24% in a year, per Silicon Data — even as lenders slash what they'll advance against chip collateral and AI output prices slide. CoreWeave says its own older fleets repriced 30-50% in a single quarter.
- Original source: cryptobriefing.com.
- The source published it on Nov 20, 2026 at 7:55 PM.
- No second outlet has reported it yet. Treat it as a single-source claim.
Median rental prices for Nvidia's H100 hit roughly $3.70 per GPU-hour in the week of October 19, 2026, according to Silicon Data. A year earlier the same hour cost $2.98, making for a 24% climb. That is a strange look for a sector that spent the autumn absorbing bad headlines.
Silicon Data is a New York firm that tracks GPU rental costs, and its indices cover 95%+ of providers. Shorter windows show the same push: the firm's H100 Neo index rose 11.3% between September 30 and November 10, 2026, a move it tied to tight supply and demand. Longer commitments moved harder still, with H100 contract rates up roughly 38-40% from late 2025 into early 2026 and carrying into the fourth quarter.
CoreWeave, the GPU cloud provider that acts as a bellwether here, told investors on its November 10 Q3 2026 earnings call that older Ampere and Hopper fleets repriced 30-50% quarter over quarter. Silicon Data's indices agree on direction but not size, showing average Q3 gains of approximately 2.4% to 13.1%, with some indices up as much as roughly 30% year over year. Independent tracker GetDeploying shows similar increases across Hopper and Nvidia's newer Blackwell line. The drivers, per the data: demand is up significantly, memory costs are rising, and fewer GPUs are rolling off existing contracts as the market approaches 2027-2028.
The credit side tells a gloomier story. Lenders have cut advance rates on GPU-backed loans from 80% to 50%, meaning borrowers who once got 80 cents for every dollar of chips pledged now get 50 — a squeeze that could slow how fast new capacity gets financed. Silicon Data has also flagged its “crunch spread,” which measures how much work a rented GPU must do to break even when rental costs rise while token prices fall. For Nvidia's B300 chips, the break-even throughput requirement has climbed 75% since May 31, 2026, and the firm expects the pressure to persist into Q4 2026.
ZeroHedge called rental pricing the one chart AI bulls kept pointing to all autumn while the broader supercycle narrative unraveled — by its account the AI trade peaked in June and CoreWeave imploded in October. Why the benchmark gap matters: Silicon Data's indices back financial products including planned CME GPU futures, so a single provider reporting 30-50% increases against broad indices showing far more modest averages is exactly the divergence a trusted benchmark exists to catch.
TopicsNvidia · Silicon Data · CoreWeave · GetDeploying · ZeroHedge · H100 · Blackwell · CME
Source: cryptobriefing.com
Written by AI from verified reporting.
More stories
- Tech & AI blockchain.news 0 viewsFBI Flags ARTEX: Hacking AI Turned Loose on Real Targets Since August
- Tech & AI dailymail.com 0 viewsAnthropic staffer claims Claude just cracked the Riemann hypothesis
- Tech & AI cryptopolitan.com 1 viewAnthropic co-founder joins Musk and the Dells as a Trump Accounts donor
- Tech & AI arstechnica.com 0 views2M digital canaries expose how AI middlemen leak sensitive data

