H100 rentals have halved to around 3.38 dollars an hour while the compute tightness index jumped…
h100 rentals have halved to around 3.38 dollars an hour while the compute tightness index jumped 13.7 points into tight territory in thirty days. prices fell because the fleet grew, and the fleet grew because everyone is buying the same chips.
cheap compute and scarce compute are the same market right now.
Context
Slicast of 25 September 2026, citing AIMultiple's Cloud GPU Rental Price Index updated 25 August 2026, reports a median H100 rental of around 3.38 dollars an hour, down from more than 7 dollars in early 2024, with spreads from 1.49 to nearly 13 dollars. AIMultiple's own page, updated 22 September 2026, shows a September on-demand H100 median of 3.25 dollars per GPU-hour. GPU Economy's index shows 3.37 dollars, a median of 30 listed on-demand rates from 28 providers. Bargo's Compute Tightness Index read 57.7 of 100, Balanced, on 30 September 2026, with thresholds of 45 for balanced and 60 for tight.
The halving is measured against early 2024, about two and a half years, so it is not a recent move and does not sit with a 30-day index change. The 3.38, 3.25 and 3.37 figures are different monthly or snapshot medians from different indexes and are not merged. The jump of 13.7 points into tight territory was not found on the Bargo index, and which index the note means is not stated. The secondary source itself says H100 pricing rose through most of 2026 year over year. The line that prices fell because the fleet grew is the author's opinion.
Watch next
- The tightness index the note names, and AIMultiple's dated monthly series.
Sources
- Cloud GPU rental price index (Slicast, 25 Sep 2026)slicast.com
- GPU index (AIMultiple)aimultiple.com
- GPU price index (GPU Economy)gpueconomy.com
- Compute tightness (Bargo)bargo.ai
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 19 September 2026 at 01:17 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →