inthesingularity.ai
Tracker · GPU compute · collected six times a day

GPU Compute Price Index

Renting an H100 costs $3.82 an hour on demand, a trimmed average of 23 providers’ posted prices, and $2.49 on spot. One index per GPU, for every way of buying it, from interruptible spot capacity to five-year commitments, built from the prices 31 providers publish.

H100
$3.82
on-demand, 23 providers · spot $2.49 (65%) · 7-day avg $3.83
H200
$5.56
on-demand, 18 providers · spot $3.51 (63%) · 7-day avg $5.48
B200
$7.87
on-demand, 15 providers · spot $5.15 (65%) · 7-day avg $7.88
B300
$9.41
on-demand, 10 providers · spot $6.35 (67%) · 7-day avg $9.71
A100
$2.37
on-demand, 19 providers · spot $1.36 (58%) · 7-day avg $2.35

Per GPU-hour, global, all providers, as of Nov 20, 11:14 UTC. Each index is the trimmed mean of providers’ prices, each provider counted once.

On-demand and spot, by GPU

Per GPU-hour, log scale. The line runs from the cheapest provider’s on-demand price to the dearest.

On-demand indexSpot indexProviders’ on-demand range
$0.5$1$2$5$10$20H100H100 on-demand: providers from $1.73 to $14.38H100 spot: $2.49 (8 providers)H100 on-demand: $3.82 (23 providers)$3.82H200H200 on-demand: providers from $3.00 to $13.25H200 spot: $3.51 (7 providers)H200 on-demand: $5.56 (18 providers)$5.56B200B200 on-demand: providers from $3.74 to $17.09B200 spot: $5.15 (5 providers)B200 on-demand: $7.87 (15 providers)$7.87B300B300 on-demand: providers from $6.50 to $23.23B300 spot: $6.35 (4 providers)B300 on-demand: $9.41 (10 providers)$9.41GB200GB200 on-demand: providers from $10.50 to $33.53GB200 spot: $6.22 (1 provider)GB200 on-demand: $16.00 (3 providers)$16.00A100A100 on-demand: providers from $1.30 to $5.33A100 spot: $1.36 (7 providers)A100 on-demand: $2.37 (19 providers)$2.37MI300XMI300X on-demand: providers from $2.59 to $7.50MI300X spot: $1.39 (1 provider)MI300X on-demand: $4.39 (5 providers)$4.39L40SL40S on-demand: providers from $0.48 to $3.77L40S spot: $1.01 (4 providers)L40S on-demand: $1.37 (15 providers)$1.37RTX PRO 6000RTX PRO 6000 on-demand: providers from $0.99 to $5.99RTX PRO 6000 spot: $1.51 (7 providers)RTX PRO 6000 on-demand: $2.25 (19 providers)$2.25
$0.5$1$2$5$10$20H100H100 on-demand: providers from $1.73 to $14.38H100 spot: $2.49 (8 providers)H100 on-demand: $3.82 (23 providers)H200H200 on-demand: providers from $3.00 to $13.25H200 spot: $3.51 (7 providers)H200 on-demand: $5.56 (18 providers)B200B200 on-demand: providers from $3.74 to $17.09B200 spot: $5.15 (5 providers)B200 on-demand: $7.87 (15 providers)B300B300 on-demand: providers from $6.50 to $23.23B300 spot: $6.35 (4 providers)B300 on-demand: $9.41 (10 providers)GB200GB200 on-demand: providers from $10.50 to $33.53GB200 spot: $6.22 (1 provider)GB200 on-demand: $16.00 (3 providers)A100A100 on-demand: providers from $1.30 to $5.33A100 spot: $1.36 (7 providers)A100 on-demand: $2.37 (19 providers)MI300XMI300X on-demand: providers from $2.59 to $7.50MI300X spot: $1.39 (1 provider)MI300X on-demand: $4.39 (5 providers)L40SL40S on-demand: providers from $0.48 to $3.77L40S spot: $1.01 (4 providers)L40S on-demand: $1.37 (15 providers)RTX PRO 6000RTX PRO 6000 on-demand: providers from $0.99 to $5.99RTX PRO 6000 spot: $1.51 (7 providers)RTX PRO 6000 on-demand: $2.25 (19 providers)

What the prices say

Spot capacity, which the provider can take back at short notice, costs 58% of the on-demand price for an A100 and 74% for an L40S, with the other GPUs in between.

The hyperscalers (AWS, Azure, Google Cloud and Oracle) charge the most. On demand, an H100 costs $10.68 an hour from them against $3.68 from neoclouds, 2.9 times as much; the multiple is 2.6 for L40S, 2.1 for A100 and 2.1 for H200.

Committing for a year brings the price down to about 69% of on-demand, comparing each provider with its own on-demand price (H100 69%, H200 69%, A100 75% and RTX PRO 6000 69%). Few providers post term prices, so the longer commitments rest on a handful of quotes; most large term deals are negotiated privately.

Daily fixing, on-demand, indexed

Each GPU’s on-demand index at 16:00 UTC each day, rebased to 100 at its first fixing on October 26, so a move of the same size reads the same whatever the GPU costs. Hover or tap for the dollar prices.

90100110Oct 27Oct 31Nov 4Nov 8Nov 12Nov 16Nov 20H100 on-demand, rebased to 100 on Oct 26 ($3.86)H200 on-demand, rebased to 100 on Oct 26 ($5.14)B200 on-demand, rebased to 100 on Oct 26 ($7.72)B300 on-demand, rebased to 100 on Oct 26 ($9.16)GB200 on-demand, rebased to 100 on Oct 26 ($16.00)A100 on-demand, rebased to 100 on Oct 26 ($2.27)MI300X on-demand, rebased to 100 on Oct 26 ($4.16)L40S on-demand, rebased to 100 on Oct 26 ($1.42)RTX PRO 6000 on-demand, rebased to 100 on Oct 26 ($2.17)H200 108.3MI300X 105.7A100 104.1RTX PRO 6000 103.6B300 102.7B200 102.0GB200 100.0H100 98.9L40S 96.0
90100110Oct 30Nov 6Nov 13Nov 20H100 on-demand, rebased to 100 on Oct 26 ($3.86)H200 on-demand, rebased to 100 on Oct 26 ($5.14)B200 on-demand, rebased to 100 on Oct 26 ($7.72)B300 on-demand, rebased to 100 on Oct 26 ($9.16)GB200 on-demand, rebased to 100 on Oct 26 ($16.00)A100 on-demand, rebased to 100 on Oct 26 ($2.27)MI300X on-demand, rebased to 100 on Oct 26 ($4.16)L40S on-demand, rebased to 100 on Oct 26 ($1.42)RTX PRO 6000 on-demand, rebased to 100 on Oct 26 ($2.17)H200 108.3MI300X 105.7A100 104.1RTX PRO 6000 103.6B300 102.7B200 102.0GB200 100.0H100 98.9L40S 96.0

Each GPU, on-demand and spot

Daily fixing at 16:00 UTC, per GPU-hour, each on its own scale; the latest point is provisional until today’s fixing. Change is since the first fixing.

On-demand indexSpot index
H100OD $3.82 −1.1% · spot $2.49 +11.0%
$2$3$4Oct 30Nov 6Nov 13Nov 20$3.82$2.49
H200OD $5.56 +8.3% · spot $3.51 +16.6%
$3$4$5Oct 30Nov 6Nov 13Nov 20$5.56$3.51
B200OD $7.87 +2.0% · spot $5.15 +4.7%
$6$8Oct 30Nov 6Nov 13Nov 20$7.87$5.15
B300OD $9.41 +2.7% · spot $6.35 +20.1%
$6$8$10Oct 30Nov 6Nov 13Nov 20$9.41$6.35
GB200OD $16.00 0.0% · spot $6.22 0.0%
$10$15Oct 30Nov 6Nov 13Nov 20$16.00$6.22
A100OD $2.37 +4.1% · spot $1.36 +2.5%
$1.50$2.00Oct 30Nov 6Nov 13Nov 20$2.37$1.36
MI300XOD $4.39 +5.7% · spot $1.39 0.0%
$2$3$4Oct 30Nov 6Nov 13Nov 20$4.39$1.39
L40SOD $1.37 −4.0% · spot $1.01 +6.6%
$1.00$1.20$1.40Oct 30Nov 6Nov 13Nov 20$1.37$1.01
RTX PRO 6000OD $2.25 +3.6% · spot $1.51 +6.9%
$1.50$2.00Oct 30Nov 6Nov 13Nov 20$2.25$1.51

Every GPU and term

Per GPU-hour, global, all providers, with the number of providers in each index. Commitments are grouped by length; a figure from one or two providers is that provider’s price, not a market.

GPUSpotOn-demand1M3M6M1Y2Y3Y5Y
H100H100 SXM 80GB$2.498$3.8223$5.624$5.341$5.291$6.185$4.131$4.973$5.751
H200H200 SXM 141GB$3.517$5.5618$7.423$6.831$6.761$6.675$5.281$4.653—
B200B200 HGX 180GB$5.155$7.8715$10.514$9.641$9.541$11.123$7.461$7.142—
B300B300 HGX 288GB$6.354$9.4110$14.532$12.781$12.651$12.133$9.881$10.511—
GB200GB200 NVL72$6.221$16.003$10.581——$21.461—$14.751—
A100A100 SXM 80GB$1.367$2.3719$2.482$2.401$2.371$3.104$1.851$2.343—
MI300XMI300X 192GB$1.391$4.395———$3.352—$3.291—
L40SL40S 48GB$1.014$1.3715$2.041$2.021$2.001$2.212$1.561$1.731—
RTX PRO 6000RTX PRO 6000 Blackwell (server)$1.517$2.2519$2.921$2.891$2.861$3.103$2.241$2.372—

Terms: Spot, spot / interruptible; On-demand, on-demand; 1M, commit ≤1 month; 3M, commit 1-3 months; 6M, commit 3-6 months; 1Y, commit 1 year; 2Y, commit 2 years; 3Y, commit 3 years; 5Y, commit 5 years. Download: gpu-index-latest.csv.

Price against on-demand

Each provider’s price for a term divided by its own on-demand price, the median across providers quoting both. This compares like with like: providers that post term prices tend to be cheaper on demand too, so the board above can show a term costing more than on-demand.

GPUSpot1M3M6M1Y2Y3Y5Y
H10044%784%497%196%169%575%144%340%1
H20050%685%397%196%169%575%146%3—
B20050%4100%399%198%184%276%142%1—
B30045%284%298%197%185%276%145%1—
GB20019%1———64%1—44%1—
A10050%779%297%196%175%475%145%3—
MI300X18%1———69%2—44%1—
L40S49%498%197%196%179%275%146%1—
RTX PRO 600051%798%197%196%169%375%145%2—

By type of provider, on-demand

Per GPU-hour, with the number of providers. Hyperscalers: AWS, Azure, Google Cloud and Oracle. Neoclouds sell their own capacity at posted rates. Marketplaces: prices set by many independent hosts or by auction.

GPUAll providersHyperscalersNeocloudsMarketplaces
H100$3.8223$10.684$3.6815$2.284
H200$5.5618$10.304$4.8311$4.353
B200$7.8715$15.542$7.7910$6.723
B300$9.4110$19.122$9.006$6.722
GB200$16.003$24.762$10.501—
A100$2.3719$4.644$2.1712$1.393
MI300X$4.395$6.752$3.193—
L40S$1.3715$3.632$1.4210$0.703
RTX PRO 6000$2.2519$4.503$2.1113$1.733

By region, on-demand

Per GPU-hour, from providers that say where their capacity is, with the like-for-like difference from North America below: each provider’s price in the region against its own North American price, the median across providers quoting both. Rate cards without a location count only in the global index.

GPUNorth AmericaEuropeAsia-PacificMiddle East & AfricaLatin America
H100$5.4011 · base$5.6510 · 0%$5.848 · 0%$10.003 · 0%$10.782 · +34%
H200$6.4911 · base$5.9211 · 0%$7.177 · +5%$8.775 · +20%$10.003 · +33%
B200$8.0610 · base$10.785 · 0%$10.994 · 0%$8.503 · 0%$14.001 · 0%
B300$11.165 · base$11.117 · 0%$15.003 · 0%$15.001 · 0%$25.072 · +49%
GB200$16.003 · base$16.003 · 0%$26.252 · +17%$33.152 · +43%$35.042 · +50%
A100$2.8012 · base$3.159 · 0%$3.177 · +8%$3.885 · 0%$6.102 · +33%
MI300X$6.003 · base$6.832 · +3%$6.972 · +5%$7.292 · +10%$9.002 · +33%
L40S$0.827 · base$2.108 · +9%$2.352 · −10%$3.503 · 0%$2.262 · +22%
RTX PRO 6000$2.3514 · base$2.5410 · 0%$3.046 · 0%$4.501 · 0%$4.501 · 0%

Method

What is priced. US dollars per GPU-hour, before tax, for 9 reference GPUs (H100 SXM 80GB, H200 SXM 141GB, B200 HGX 180GB, B300 HGX 288GB, GB200 NVL72, A100 SXM 80GB, MI300X 192GB, L40S 48GB and RTX PRO 6000 Blackwell (server)). A price quoted per instance or node is divided by its GPUs. Each index covers one GPU, one way of buying it (spot or interruptible, on-demand, or a commitment grouped by length, from up to a month to five years), one region (global or one of five) and one type of provider.

Sources. 32 sources: providers’ own price lists and rate cards, floating spot markets and auctions, marketplace asks and catalogs of posted prices, collected six times a day (03:07 to 23:07 UTC, every four hours). 32 of 35 answered in the latest run; failing: Cirrascale, Crusoe and Together AI GPU clusters. A source that fails keeps its last prices for up to 36 hours. These are offered prices, not transactions; large term contracts are negotiated privately and usually below posted rates.

Index. Each provider counts once, at the median of its prices in scope (its regions, instance sizes and configurations). The index is the 20% trimmed mean of the provider prices: with five or more providers the lowest and highest fifth are dropped and the rest averaged; with fewer, the median. Providers are equally weighted, since capacity isn’t public. Listing placeholders (under 15% of the GPU’s on-demand median) and outliers (more than four times away from the median of comparable providers, those of the same type when there are three or more) are set aside, and a spot price at or above the same provider’s on-demand price is dropped.

Fixing. One fixing a day at 16:00 UTC, from each price in force at that time; a final fixing stands as published. Between fixings this page shows the same calculation on the latest prices. Seven- and thirty-day moving averages of the fixings appear once they have five and twenty days. The first fixing is October 26.

Updates. Prices are collected and this page rebuilt six times a day; the time at the top is the latest collection. Downloads: gpu-index-latest.csv, the latest print of every GPU and term; gpu-index-daily.csv, every daily fixing, by GPU, term, region and type of provider.

Source: providers’ published prices. Indexes are inthesingularity.ai calculations. Not investment advice. See also GPU spot prices, Nebius’s spot market every 15 minutes.