NVIDIA · 80 GB · updated 2026-11-02
A100 price: rent per GPU-hour, history and outlook
The NVIDIA A100 (80GB) rents for $0.53 to $5.07 per GPU-hour on demand across hyperscalers and neoclouds, median $1.75; a one-year commitment lists at $1.59. Coloprice tracks published provider price pages only — no purchase-price estimates. Below: the range, 18 historical snapshots, the analyst note with sources and the colocation alternative for owned hardware.
How much does an A100 cost to rent in 2026?
On 2026-11-02, NVIDIA A100 (80GB) on-demand rental ranges from $0.53 to $5.07 per GPU-hour across tracked providers, with a median of $1.75. A one-year contract lists at $1.59 per GPU-hour. At the median, one GPU running around the clock costs about $1278 per month; an eight-GPU node about $10220. Prices are read from provider price pages and change weekly.
A100 price history
Median on-demand price per GPU-hour by snapshot. Peak $3.20 (2024-01-01), low $1.69 (2026-09-09); change versus a year ago: -24%. Snapshots before 2026-08 are seeded from published market indices; later points are collected by Coloprice from provider price pages.
| Date | Median | Min | Max | 1-year |
|---|---|---|---|---|
| 2026-11-02 | $1.75 | $0.53 | $5.07 | $1.59 |
| 2026-10-26 | $1.75 | $0.53 | $5.07 | $1.59 |
| 2026-10-16 | $1.75 | $0.53 | $5.07 | $1.59 |
| 2026-10-09 | $1.75 | $0.53 | $5.07 | $1.59 |
| 2026-10-02 | $1.75 | $0.53 | $5.07 | $1.59 |
| 2026-09-25 | $1.75 | $0.53 | $5.07 | $1.59 |
| 2026-09-16 | $1.69 | $0.53 | $5.07 | $1.59 |
| 2026-09-09 | $1.69 | $0.53 | $5.07 | $1.59 |
| 2026-09-04 | $1.80 | $0.53 | $5.07 | $1.59 |
| 2026-08-28 | $1.80 | $0.53 | $5.07 | $1.59 |
| 2026-08-24 | $1.80 | $1.09 | $5.07 | $1.59 |
| 2026-08-21 | $1.80 | $1.09 | $5.07 | — |
| 2026-08-19 | $1.85 | $1.09 | $5.07 | — |
| 2026-08-12 | $1.85 | $1.09 | $5.07 | — |
| 2026-08-04 | $1.85 | $1.09 | $5.07 | — |
| 2026-08-03 | $1.85 | $1.09 | $5.07 | — |
| 2025-01-01 | $2.30 | $1.35 | $5.07 | — |
| 2024-01-01 | $3.20 | $1.89 | $5.07 | — |
Seed sources: Silicon Data H100 Rental Price Index · SemiAnalysis — GPU rental capacity and pricing · Introl — GPU cloud price collapse, Dec 2025
Analyst note: what moved the A100 price
Legacy workhorse holding roughly flat: floor stays at Vast.ai's marketplace ~$0.53/GPU-hr (range $0.39-0.67 depending on host reliability, unchanged), well below Thunder Compute's own $1.09/GPU-hr anchor and CoreWeave's newly-tracked $2.21/GPU-hr PCIe rate (confirmed today via CoreWeave's own pricing page, a new data point in the mid-cluster). GetDeploying's tracked median ticked up slightly to ~$1.75/GPU-hr (from last cycle's $1.69, +3.6%, within normal noise) with GCP a2-ultragpu still listing ~$5.07/GPU-hr at the top, ceiling unchanged. Civo's tiered 80GB rate card is unaffected — $1.79 on-demand, $1.59 at 12 months, down to $1.39 at 36 months; most other contract demand has moved to H100/H200.
Sources: GetDeploying — A100 price comparison · CoreWeave pricing · Vast.ai live GPU pricing · Civo — NVIDIA A100 GPU cloud pricing
A100 against the other 10 tracked GPUs
Sorted by median on-demand price. A100 ranks #4 of 10 from cheapest. Same tracker, same date.
See also
Weekly digest: GPU and colocation price moves in one email.
FAQ
Why does the A100 price range from $0.53 to $5.07?
Because the tracker covers very different sellers: marketplaces and boutique neoclouds at the bottom, hyperscalers with bundled networking, storage and support at the top. The median $1.75 is the better planning number; the minimum usually comes with capacity limits, older hosts or spot-like terms.
Is the A100 getting cheaper?
The tracked median moved from $3.20 at the first snapshot (2024-01-01) to $1.75 now, with a peak of $3.20 and a low of $1.69; the change against a year ago is -24%. Newer generations, provider capacity and inference demand drive the direction — see the analyst note above for the latest move.
What does it cost to buy an A100 instead?
Coloprice does not publish purchase prices for the A100: vendors and OEMs do not list them publicly, and quoted figures vary by volume and configuration. For a rent-versus-buy comparison use the calculator on the tools page with your own quote and the rental median $1.75.
Where can I host my own A100 servers?
In colocation facilities built for high density: current NVIDIA accelerators need liquid cooling or 40–130 kW racks, which only AI-ready halls provide. The colocation pricing pages show wholesale $/kW benchmarks by market, and the catalog marks AI-ready facilities per city.
Get A100 capacity quotes
Rental from tracked providers or colocation for your own A100 servers — one brief, several offers. Independent; no provider pays for position.
Need A100 capacity? One brief, several quotes.
Request quotes →