GPU model
NVIDIA GB200 NVL72: specs, VRAM and rental prices
NVIDIA’s NVIDIA GB200 NVL72: rented from $10.50 per GPU-hour on demand (CoreWeave). Specs from NVIDIA, prices as of Oct 25, 2026.
The GB200 NVL72 is a liquid-cooled rack that links 72 Blackwell GPUs and 36 Grace CPUs into one NVLink domain, announced by NVIDIA in 2024. NVIDIA publishes its figures for the whole rack and for one GB200 Grace Blackwell Superchip, which pairs one Grace CPU with two Blackwell GPUs; it does not publish a separate per-GPU column on that page.
Prices were last checked Oct 25, 2026. A listing not re-checked for 7 days is greyed and left out of every lowest-price figure; after 14 days it is removed from the page.
NVIDIA GB200 NVL72 specifications
| Specification | GB200 NVL72 (rack) | GB200 Superchip (1 CPU, 2 GPUs) | Source |
|---|---|---|---|
| Configuration | 36 Grace CPUs, 72 Blackwell GPUs | 1 Grace CPU, 2 Blackwell GPUs | nvidia.com |
| FP4 tensor | 1,440 PFLOPS sparse | 720 PFLOPS dense | 40 PFLOPS sparse | 20 PFLOPS dense | nvidia.com |
| FP8 / FP6 tensor (with sparsity) | 720 PFLOPS | 20 PFLOPS | nvidia.com |
| FP16 / BF16 tensor (with sparsity) | 360 PFLOPS | 10 PFLOPS | nvidia.com |
| TF32 tensor (with sparsity) | 180 PFLOPS | 5 PFLOPS | nvidia.com |
| FP32 | 5,760 TFLOPS | 160 TFLOPS | nvidia.com |
| FP64 / FP64 tensor | 2,880 TFLOPS | 80 TFLOPS | nvidia.com |
| GPU memory and bandwidth | 13.4 TB HBM3E, 576 TB/s | 372 GB HBM3E, 16 TB/s | nvidia.com |
| NVLink bandwidth | 130 TB/s | 3.6 TB/s | nvidia.com |
| CPU memory and bandwidth | 17 TB LPDDR5X, 14 TB/s | Up to 480 GB LPDDR5X, up to 512 GB/s | nvidia.com |
| Per-GPU figures | Not published | Not published | nvidia.com |
Tensor throughput is stated with sparsity unless marked dense; NVIDIA says dense is half of the sparse figure. The Superchip column covers two GPUs and one CPU, so it is not a per-GPU figure. Not published on the pages cited here: Per-GPU memory, bandwidth, throughput and power; CUDA core count; Tensor core count.
Sources: NVIDIA GB200 NVL72 page · Blackwell announcement (2024). Every figure above is copied from these NVIDIA pages; nothing is estimated.
GB200 NVL72 rental prices by provider
Lowest listed price by pricing type
| Pricing | $ per GPU-hour | Provider | Checked |
|---|---|---|---|
| On-demand | $10.50 | CoreWeave | Oct 25, 2026 |
On-demand price by provider
Each provider’s lowest fresh on-demand listing, $ per GPU-hour.
Every listed GB200 NVL72 price
| Provider | Configuration | Pricing | $ per GPU-hour | Region | Checked | Source |
|---|---|---|---|---|---|---|
| CoreWeave | 4 GPUs per instance · 186 GB each NVIDIA GB200 NVL72 |
On-demand | $10.50 Listed $42.00/hr per 4-GPU instance |
North America and Europe | coreweave.com | |
| Oracle Cloud | 4 GPUs per instance · 4x400 Gb/s InfiniBand BM.GPU.GB200.4 |
On-demand | $16.00 Listed $16 per GPU per hour (OCI - Compute - GPU - GB200) |
— | apexapps.oracle.com | |
| Microsoft Azure | 4 GPUs per instance Standard_ND128isr_NDR_GB200_v6 |
On-demand | $27.04 Listed $108.16/hr per 4-GPU instance |
eastus2 | prices.azure.com |
Download: current prices (CSV) · current prices (JSON) · daily history (CSV) · price-change feed (RSS). Free to reuse with attribution under CC BY 4.0.
Price history
The history chart appears once there are two daily snapshots of an on-demand price. Tracking began Oct 15, 2026; the first snapshot has the lowest listed on-demand price at $10.50 per GPU-hour.
Embed this chart
<iframe src="https://datacentercensus.com/embed/gpu-prices/gb200-nvl72" title="NVIDIA GB200 NVL72 rental prices" width="100%" height="500" style="border:0;max-width:640px" loading="lazy"></iframe>
<p style="font:13px/1.4 system-ui,sans-serif;margin:6px 0 0">Source: <a href="https://datacentercensus.com/gpu-prices/gb200-nvl72">Data Center Census</a></p>
<script>addEventListener("message",function(e){if(e.origin==="https://datacentercensus.com"&&e.data&&e.data.dcc){document.querySelectorAll('iframe[src^="https://datacentercensus.com/embed/"]').forEach(function(f){if(f.contentWindow===e.source)f.style.height=e.data.h+"px"})}})</script>Free with the credit line. Preview the embed.
How these prices are collected
- What counts
- Only prices a provider publishes on a public pricing page or in its public price list, opened by the Census on the day shown. A marketplace is listed under the same rule as every other provider: it must publish per-GPU-hour prices publicly, with no login, and its rows are shown with the type Marketplace. Quotes and “contact sales” rows are not listed, and neither are purchase prices.
- $ per GPU-hour
- The listed price per GPU-hour. Where a provider lists a price per instance, the Census divides it by the GPUs in that instance and shows the original figure under it. Nothing else is converted or estimated.
- Pricing types
- On-demand, spot (including tiers a provider calls preemptible or low-priority), reserved with the term the provider states, and marketplace. Reserved rows without a stated term are not listed. A marketplace row is the lowest per-GPU-hour price its public page shows for that GPU, which comes from third-party hosts and changes often. Marketplace rows are a separate type: they appear in the table, filters, downloads and feeds, but never in the “lowest on-demand” or “lowest spot” figures.
- Variants
- SXM, PCIe, NVL and similar versions are kept separate when the provider names them. A provider that does not name the variant is shown without one.
- Freshness
- Every listing carries the date it was last confirmed, and the daily run re-opens each one. If a listing goes 7 days without a re-check it stays on the page greyed, labelled with the date it was last confirmed, and drops out of the “lowest” figures and the calculator’s default results. At 14 days without a re-check it is taken off the site; the daily history still holds it.
- History
- Each daily run saves a snapshot, so the model pages can chart the lowest listed on-demand price over time and the change log can show what moved. History began Oct 15, 2026.
- Differences between providers
- Listed prices are not like for like: regions, minimum instance sizes, networking, storage, commitments and availability differ and are not priced here. Check the provider before relying on a figure.
Providers checked but not listed
- Google Cloud: The compute pricing pages (cloud.google.com/compute/gpus-pricing and vm-instance-pricing) showed no GPU prices when opened.
- Vast.ai: Its pricing page loads live "from" prices from a public JSON file whose terms allow display and citation with attribution but require a data license for bulk collection, redistribution or use in any index, benchmark or derivative product. The Census republishes prices as open data, so it does not list Vast.ai without that license.
- Voltage Park: Its pricing pages say "Contact for pricing" for every GPU; one homepage line says "Starting at $1.99/hr" without a GPU model or unit.
GB200 NVL72: common questions
How much GPU memory does the GB200 NVL72 have?
NVIDIA lists GPU memory and bandwidth for the NVIDIA GB200 NVL72 as 13.4 TB HBM3E, 576 TB/s (GB200 NVL72 (rack)), 372 GB HBM3E, 16 TB/s (GB200 Superchip (1 CPU, 2 GPUs)). The specification table on this page shows each figure with the page it came from.
What is in the GB200 NVL72?
NVIDIA lists the configuration as 36 Grace CPUs, 72 Blackwell GPUs (GB200 NVL72 (rack)), 1 Grace CPU, 2 Blackwell GPUs (GB200 Superchip (1 CPU, 2 GPUs)).
How much NVLink bandwidth does the GB200 NVL72 have?
NVIDIA lists NVLink bandwidth of 130 TB/s (GB200 NVL72 (rack)), 3.6 TB/s (GB200 Superchip (1 CPU, 2 GPUs)).
How much does it cost to rent a GB200 NVL72 per hour?
The lowest listed on-demand price among listings checked in the last 7 days is $10.50 per GPU-hour at CoreWeave, checked Oct 25, 2026. 3 fresh listings from 3 providers are shown in the table above, including spot, reserved and marketplace prices.
Which providers list GB200 NVL72 prices?
CoreWeave, Oracle Cloud, Microsoft Azure.
Which architecture is the GB200 NVL72 built on, and when was it announced?
The NVIDIA GB200 NVL72 uses the Blackwell architecture and was announced or launched in 2024, according to the NVIDIA page cited in the specifications.
Related: All GPU prices · GPU comparisons · Rental cost calculator · H100 · H200 · B200 · B300 · GB300 NVL72 · A100 · L40S · RTX PRO 6000 · L4 · T4 · MI300X · MI325X · MI355X
Cite this page
Last updated Oct 25, 2026. Figures may change as records are updated — include the date you retrieved them.
| APA | Data Center Census. (2026). NVIDIA GB200 NVL72 specs and rental prices. Retrieved from https://datacentercensus.com/gpu-prices/gb200-nvl72 |
|---|---|
| MLA | “NVIDIA GB200 NVL72 specs and rental prices.” Data Center Census, Oct 25, 2026, https://datacentercensus.com/gpu-prices/gb200-nvl72. |
| Link | https://datacentercensus.com/gpu-prices/gb200-nvl72 |
Republishing a figure or table? Attribute it to Data Center Census with a link to this page. The underlying data is downloadable as JSON and CSV; the press kit has standing figures and contact details.