macrostack
Browse

The AI stack

The whole five-layer mapL3InfrastructureL4Models & toolingL5Applications

Categories

All 79 categories, by layer →

About

How we rank & score

NVIDIA H100 SXM 80GB · prices read 12 Oct 2026 – 13 Oct 2026

What does it cost to rent an H100 SXM?

23 companies publish an on-demand price for the H100 SXM. Read on their own price pages, the cheapest is $1.99 per GPU-hour at Voltage Park and the dearest $11.06 at Google Cloud (A3, A4) — 5.6× apart for the same chip. The middle price is $3.99. Every price below links to the page it was read on.

Cheapest on-demand · per GPU-hour

$1.99

Voltage Park

Middle price · per GPU-hour

$3.99

of 23 published prices

Dearest on-demand · per GPU-hour

$11.06

Google Cloud (A3, A4) · 5.6× the cheapest

What a month costs

The rental alone, at the cheapest, middle and dearest published on-demand price. Storage, network and data transfer are usually billed on top and are not included.

How muchCheapestMiddleDearest
1 GPU for 100 hours$199$399$1,106
1 GPU for a working month (160 h)$318$638$1,770
1 GPU always on (730 h)$1,453$2,913$8,075
8 GPUs always on (730 h)$11,622$23,302$64,598

Work out your own H100 SXM bill →Any number of hours, 1 or 8 GPUs, every company side by side.

Every published on-demand H100 SXM price, cheapest first

One row per company: its own fixed on-demand price per GPU per hour, rentable on the day it was read. Ordered by price and nothing else. Each row says how its number was made and how fresh it is — current for a week, aging to 30 days, stale after that; the site will not build with an on-demand price older than 45 days.

#CompanyTierPer GPU-hourFreshnessSource
1Voltage ParkThe price cards say 'Contact for pricing'. The figure is the page's own FAQ: one H100 for an hour, no contract.
How this number was made

Already per GPU

'starting at $1.99' for one H100 GPU for 1 hour without a contract (FAQ); cards say 'Contact for pricing'. GPUs are HGX H100

Region: US (company is US-only; region not stated on page)

On-demand$1.99Aging · 12 Oct 2026their page ↗
2Vultr
How this number was made

Already per GPU (8-GPU HGX H100, 640 GB GPU RAM)

'Price starting at' $2.300/GPU/hr, Cloud GPU (VM), billed hourly

Region: not stated (FAQ: some locations priced differently; per-location prices are in the Control Panel)

On-demand$2.30Aging · 12 Oct 2026their page ↗
3GMI CloudThe price card does not name its tier. Recorded as the on-demand starting rate — flagged, not certain.
How this number was made

'from $2.60 /GPU-hour'; tier not labelled on the card (see company note); listed as 'NVIDIA H100 GPU', form factor not stated

Region: not stated on page

On-demand$2.60Aging · 12 Oct 2026their page ↗
4Civo
How this number was made

1-GPU instance; 8-GPU instance $23.92/hr ÷ 8 = $2.99 (identical)

Per hour, 1 x H100 80GB (H100 SXM table)

Region: not stated on page

On-demand$2.99Aging · 12 Oct 2026their page ↗
5OVHcloudH100 80GB (h100-380: 1 GPU; form factor not stated)
How this number was made

h100-380 $2.99/hr, 1 GPU, no division.

USD, excl. VAT, hourly, from the international page. h100-760 (2x) $5.98 and h100-1520 (4x) $11.97 are the same per GPU (within $0.01).

Region: GRA9 / GRA11 (Gravelines, France); not offered in OVHcloud US regions

On-demand$2.99Aging · 12 Oct 2026their page ↗
6Hyperstack (NexGen Cloud)
How this number was made

Already per GPU

Per hour per GPU (H100 SXM 80 GB); billed per minute

Region: not stated on page

On-demand$3.20Aging · 12 Oct 2026their page ↗
7JarvisLabsH100 SXM 80GB
How this number was made

none — page prices per GPU

On-demand list price, per GPU-hour, per-minute billing. Read 20:44 UTC (16:44 EDT).

Region: not stated (India regions mentioned)

On-demand$3.49Aging · 12 Oct 2026their page ↗
8Crusoe
How this number was made

Per GPU-hour (H100 80GB HGX)

Region: not stated on page

On-demand$3.90Aging · 12 Oct 2026their page ↗
9ModalNvidia H100 SXM5Base price: GPU time only, preemptible by default. Non-preemptible runs cost 3x; CPU and memory are extra.
How this number was made

Per GPU per second x 3600, no division.

Per-second billing: $0.001097/sec x 3600. GPU only; CPU/memory extra. Base (preemptible) rate; non-preemptible = 3x.

Region: Not pinned: base price lets Modal place the job; region selection costs 1.15-1.75x

On-demand$3.95Aging · 12 Oct 2026their page ↗
10RunPodH100 SXM 80GB
How this number was made

none — page prices per GPU

Secure Cloud Pod — RunPod's own on-demand list price per GPU-hour, billed per second; ranked from 13 Oct on Abdul's call (one rule for every company: a published hourly price is ranked). Re-read 13 Oct 2026, ~22:35 EDT (14 Oct ~02:35 UTC), on runpod.io/pricing with the Secure Cloud toggle (the default); unchanged since 9 Oct.

Region: not stated (30+ regions)

On-demand$3.99Aging · 13 Oct 2026their page ↗
11Together AINVIDIA HGX H100
How this number was made

Printed per GPU per hour, no division.

GPU Clusters, on-demand, pay as you go; hourly.

Region: Not stated on the page

On-demand$3.99Aging · 12 Oct 2026their page ↗
12Lambda
How this number was made

Already per GPU. By instance size: 8x $3.99, 4x $4.09, 2x $4.19, 1x $4.29

Per GPU per hour, plus applicable sales tax/VAT/GST (H100 SXM 80GB)

Region: not stated on page

On-demand$3.99–$4.29Aging · 12 Oct 2026their page ↗
13DataCrunch (now Verda)
How this number was made

Already per GPU (1x instance)

Per hour, 1x H100 SXM5 80GB (two configs listed, same price)

Region: not stated on page

On-demand$4.05Aging · 12 Oct 2026their page ↗
14Paperspace (DigitalOcean)
How this number was made

Already per GPU

Per GPU per hour, GPU Droplet NVIDIA HGX H100 (1 or 8 GPUs); billed per second, 5-minute minimum

Region: not stated on page

On-demand$4.41Aging · 12 Oct 2026their page ↗
15Nebius
How this number was made

Per GPU-hour, column 'Effective October 1, 2026'; the page's other column 'On-demand, GPU-hour' still shows $3.85

Region: not stated on page

On-demand$4.50Aging · 12 Oct 2026their page ↗
16ReplicateNvidia H100 80GB (gpu-h100; form factor not stated)
How this number was made

1-GPU hardware class, no division.

Per-second billing: $0.001525/sec; Replicate prints $5.49/hr.

Region: Not stated on the page

On-demand$5.49Aging · 12 Oct 2026their page ↗
17CoreWeave
How this number was made

$49.24/hr for an 8-GPU HGX H100 (80GB) instance ÷ 8

Instance price per hour

Region: North America (Europe on-demand identical)

On-demand$6.16Aging · 12 Oct 2026their page ↗
18BasetenH100 80 GiB VRAM (form factor not stated)
How this number was made

Single-GPU instance; per-minute price x 60.

Per-minute billing: $0.10833/min x 60. Instance price (GPU instance with its vCPU/RAM).

Region: Not stated (Baseten: 'Talk to Sales about compute in other countries and regions')

On-demand$6.50Aging · 12 Oct 2026their page ↗
19AWS (P5, P6, G6)H100 80GB HBM3, 8 per instance on 900 GB/s NVSwitch (AWS does not use the word SXM)
How this number was made

p5.48xlarge on-demand Linux $55.04/hr ÷ 8 GPUs = $6.88 (cross-check: p5.4xlarge, 1x H100, $6.88/hr). Read from the price file AWS's on-demand page loads (b0.p.awsstatic.com/pricing/2.0/meteredUnitMaps/ec2/USD/current/ec2-ondemand-without-sec-sel/US East (N. Virginia)/Linux/index.json, published 2026-10-08T18:48:50Z); the page's table is rendered from it.

Billed per second. The 1-GPU p5.4xlarge is also $6.88/hr, the same per-GPU rate.

Region: us-east-1 (US East, N. Virginia)

On-demand$6.88Aging · 12 Oct 2026their page ↗
20Fireworks AIH100 80 GB (form factor not stated)
How this number was made

Printed per GPU, no division.

Billed per GPU second; Fireworks prints $0.134/min and $8.00/hr; the hourly figure is used.

Region: Not pinned (region-restricted deployments cost 1.5x, via sales)

On-demand$8.00Aging · 12 Oct 2026their page ↗
21Oracle Cloud (OCI)H100 80GB (BM.GPU.H100.8, bare metal, 8 GPUs; form factor not named by Oracle)
How this number was made

Oracle prices per GPU (metric 'GPU Per Hour'), PAY_AS_YOU_GO $10, no division. BM.GPU.H100.8, part B98415. The page's price cells are empty in its HTML and filled from Oracle's public price-list API (https://apexapps.oracle.com/pls/apex/cetools/api/v1/products/?partNumber=B98415&currencyCode=USD, lastUpdated 2026-10-08), which is where the figure was read.

Pay-as-you-go.

Region: Not region-specific (Oracle's USD price list carries one rate per part number)

On-demand$10.00Aging · 12 Oct 2026their page ↗
22Microsoft Azure (ND, NC)H100 80GB (ND96is H100 v5 family: 8 GPUs; high = ND96isr H100 v5, the same machine with InfiniBand; Microsoft does not use the word SXM)
How this number was made

Low: ND96is_H100_v5 $88.488/hr ÷ 8 — the same $88.488 for ND96is_noIB, ND96isf and ND96is_flex. High: ND96isr_H100_v5 $98.32/hr ÷ 8 (the InfiniBand machine this row showed alone until 13 Oct). Both Linux, East US 2.

Pay-as-you-go, listed per hour. Shown as a range like Google's (a3-highgpu to A3 Mega), on Abdul's call of 13 Oct: Azure sells several 8-GPU H100 machines and the InfiniBand one is the dearest. Azure also sells a 1-GPU H100 NVL 94GB PCIe VM, NC40ads H100 v5, at $6.98/hr — a different variant, kept out of this row. Re-read 13 Oct 2026, ~22:30 EDT (14 Oct ~02:30 UTC), on Microsoft's public Retail Prices API (prices.azure.com), East US 2, Linux pay-as-you-go.

Region: East US 2 (eastus2)

On-demand$11.06–$12.29Aging · 13 Oct 2026their page ↗
23Google Cloud (A3, A4)H100 80GB (a3-highgpu-8g; high = a3-megagpu-8g, A3 Mega)
How this number was made

Low: a3-highgpu-8g on-demand $88.490000119/hr ÷ 8. High: a3-megagpu-8g $93.400712807/hr ÷ 8.

Per-second billing; listed per hour. Google prices the whole machine (GPUs + vCPU + memory + bundled Local SSD); not eligible for sustained-use discounts.

Region: Not stated by Google for this table (the A4/A3/A2 table has no region selector; changing the page's region selector moves only the G-series table, checked in the built-in browser)

On-demand$11.06–$11.68Aging · 12 Oct 2026their page ↗

Listed apart, never ranked

Cheaper, but not a price you can count on (27)

Spot, preemptible, community and marketplace listings. They can be interrupted, or the price is set by an independent host and moves with them.

  • Vast.ai (H100 SXM 80GB): $0.65–$1.66 per GPU-hour · Marketplace · Spot — can be interrupted · global marketplace · Aging · 9 Oct 2026 their page ↗
  • Vast.ai (H100 PCIe 80GB): $0.67–$0.75 per GPU-hour · Marketplace · Spot — can be interrupted · global marketplace · Aging · 9 Oct 2026 their page ↗
  • Nebius: $0.79 per GPU-hour · Spot — can be interrupted · Aging · 9 Oct 2026 their page ↗
  • JarvisLabs (H100 SXM 80GB): $1.54 per GPU-hour · Spot — can be interrupted · Aging · 9 Oct 2026 their page ↗
  • Vast.ai (H100 SXM 80GB): $1.73–$2.40 per GPU-hour · Marketplace listing · global marketplace · Aging · 9 Oct 2026 their page ↗
  • Vast.ai (H100 SXM 80GB): $1.80–$2.83 per GPU-hour · Marketplace on-demand (hosts vary) · global marketplace; cheapest host: India (unverified) · Aging · 9 Oct 2026 their page ↗
  • TensorDock (H100 SXM5 80GB): $1.91 per GPU-hour · Marketplace · Spot — can be interrupted · Aging · 9 Oct 2026 their page ↗
  • RunPod (H100 PCIe 80GB): $1.99 per GPU-hour · Community cloud — other people's hardware · Aging · 9 Oct 2026 their page ↗
  • Together AI (NVIDIA HGX H100): $1.99 per GPU-hour · Preemptible — can be interrupted · Aging · 9 Oct 2026 their page ↗
  • Vast.ai (H100 NVL 94GB): $2.00–$2.13 per GPU-hour · Marketplace · Spot — can be interrupted · global marketplace · Aging · 9 Oct 2026 their page ↗
  • DataCrunch (now Verda): $2.02 per GPU-hour · Spot — can be interrupted · Aging · 12 Oct 2026 their page ↗
  • Akash Network (H100 SXM5 80GB): $2.04–$2.56 per GPU-hour · Marketplace listing · decentralised providers; not stated · Aging · 9 Oct 2026 their page ↗
  • TensorDock (H100 SXM5 80GB): $2.25–$2.50 per GPU-hour · Marketplace on-demand (hosts vary) · Aging · 9 Oct 2026 their page ↗
  • Microsoft Azure (ND, NC) (H100 80GB (ND96isr H100 v5, 8x)): $2.27 per GPU-hour · Spot — can be interrupted · East US 2 (eastus2) · Aging · 9 Oct 2026 their page ↗
  • Vast.ai (H100 NVL 94GB): $2.39–$2.47 per GPU-hour · Marketplace on-demand (hosts vary) · global marketplace · Aging · 9 Oct 2026 their page ↗
  • Vast.ai (H100 NVL (page label 80GB)): $2.39–$2.93 per GPU-hour · Marketplace listing · global marketplace · Aging · 9 Oct 2026 their page ↗
  • CoreWeave: $2.46 per GPU-hour · Spot — can be interrupted · North America · Aging · 9 Oct 2026 their page ↗
  • Vast.ai (H100 PCIe 80GB): $2.59–$2.77 per GPU-hour · Marketplace on-demand (hosts vary) · global marketplace · Aging · 9 Oct 2026 their page ↗
  • RunPod (H100 NVL 94GB): $2.59 per GPU-hour · Community cloud — other people's hardware · Aging · 9 Oct 2026 their page ↗
  • Vast.ai (H100 PCIe 80GB): $2.59–$2.77 per GPU-hour · Marketplace listing · global marketplace · Aging · 9 Oct 2026 their page ↗
  • io.net (H100 PCIe 80GB): $2.60–$3.67 per GPU-hour · Marketplace on-demand (hosts vary) · US (low); France / Poland (high) · Aging · 9 Oct 2026 their page ↗
  • RunPod (H100 SXM 80GB): $2.69 per GPU-hour · Community cloud — other people's hardware · Aging · 9 Oct 2026 their page ↗
  • Spheron (H100 80GB (menu and detail page: SXM5)): $2.77 per GPU-hour · Marketplace · Spot — can be interrupted · Aging · 9 Oct 2026 their page ↗
  • Spheron (H100 80GB (menu and detail page: SXM5)): $2.98 per GPU-hour · Marketplace on-demand (hosts vary) · Aging · 9 Oct 2026 their page ↗
  • AWS (P5, P6, G6) (H100 80GB (p5.48xlarge, 8x)): $3.61 per GPU-hour · Spot — can be interrupted · us-east-1 (US East, N. Virginia) · Aging · 9 Oct 2026 their page ↗
  • io.net (H100 SXM5 80GB): $4.22 per GPU-hour · Marketplace on-demand (hosts vary) · France · Aging · 9 Oct 2026 their page ↗
  • Google Cloud (A3, A4) (H100 80GB (a3-highgpu-8g, 8x)): $6.29 per GPU-hour · Spot — can be interrupted · Aging · 9 Oct 2026 their page ↗

Reserved, committed and promo prices (20)

A contract term or a limited offer, not a price for renting by the hour. Published ones are listed here; negotiated rates are lower and are not published by anyone.

Listed, but not comparable

  • Scaleway (H100 SXM 80GB (H100-SXM-8-80G; high = H100-SXM-2-80G)) publishes 3.1664–3.3099 EUR per GPU-hour only, read 9 Oct 2026. We do not convert a price at an exchange rate the company did not publish. their page ↗
  • CUDO Compute rents it without a public price — 'Quote on request' on CUDO's pricing page.. their page ↗
  • Genesis Cloud rents it without a public price — Company page unreachable on 2026-10-09 (TLS error / Cloudflare 522); no price read.. their page ↗
  • Nscale rents it without a public price — Listed on the GPU nodes page; price only via 'Contact sales' / 'Get a quote' or inside the console after sign-up.. their page ↗
  • Prime Intellect rents it without a public price — Listed as 'Unavailable' in the picker at read time; no price shown.. their page ↗
  • SF Compute rents it without a public price — SF Compute prices sit behind a sign-in (pricing page redirects to login).. their page ↗

Questions

How much does it cost to rent an H100 SXM per hour?
Between $1.99 and $11.06 per GPU-hour on the published on-demand prices of 23 companies, read on their own price pages 12 Oct 2026 – 13 Oct 2026. The middle price is $3.99: about half the companies charge less, about half more. Voltage Park was the cheapest and Google Cloud (A3, A4) the dearest — 5.6× apart for the same chip.
What is the cheapest place to rent an H100 SXM?
On fixed on-demand prices, Voltage Park at $1.99 per GPU-hour, read 12 Oct 2026. Spot, community and marketplace listings can go lower, but they can be interrupted or depend on someone else's machine, so they are listed separately and never ranked. Reserved and committed contracts are lower again — and nobody publishes what large buyers actually pay.
How much does an H100 SXM cost per month?
Running one H100 SXM around the clock (730 hours) costs $1,453 at the cheapest published on-demand price, $2,913 at the middle price and $8,075 at the dearest. A machine of eight costs eight times that. Storage, network and data transfer are usually extra and are not included.
Why is the same GPU so much cheaper at one company than another?
Because you are not only buying the chip. The cheapest end is often brokered or resold capacity, priced per GPU with everything else extra. The dearest end is a hyperscaler list price for a whole 8-GPU machine with its CPUs, memory and storage, which exists mainly to make the committed rate look like a discount. Between them sit differences in support, networking, region, billing and contract terms — and in who the counterparty is.

How these numbers were made, and how to reuse them

Macrostack does not rent clusters and does not benchmark them. Every figure is a published price read on the company's own price page on the date in its row. Committed, reserved and negotiated rates are lower than anything shown as a price here, and the real ones are not published by anyone.

Each price read on the company's own public price page on the date shown. On-demand per GPU per hour in USD; multi-GPU machines divided by GPU count (see normalization); per-second billing converted to per hour. Not measured by Macrostack. Committed, reserved and negotiated rates are lower and are not published by anyone. The full table — every chip, every company, 78 H100 SXM rows among them — is free to reuse under CC BY 4.0 with credit and a link.

Download CSVDownload JSON

Cite as

Macrostack (2026). H100 SXM rental prices, read 13 Oct 2026. CC BY 4.0. https://www.macrostack.net/compute/gpu-prices/h100

Every chip side by side: the GPU cost calculator · Other chips: H200 rental prices · B200 rental prices · A100 80GB rental prices · L40S rental prices · L4 rental prices · The chip itself: NVIDIA H100 SXM 80GB specs · Who these companies are: Layer 2.5 — Compute.

Prices read 12 Oct 2026 – 13 Oct 2026.

The Macrostack brief

New swaps, worth your inbox.

A short, occasional email when we add a high-intent alternative or ship a new head-to-head. No spam, no selling your address — unsubscribe in one click.