B200 (Blackwell) Cloud Pricing: What the Newest GPU Costs to Rent
Pricing guide · Updated November 11, 2026 · Every rate below was read from the provider's official pricing page on November 11, 2026
How to read these prices. Each figure comes from the provider's own public pricing page, checked on November 11, 2026 — not from third-party roundups. B200 capacity is still scarce and prices move fast, so treat this as a dated snapshot and always confirm the live rate before budgeting or committing.
On this page
The short answer
On November 11, 2026, a single B200 hour costs $7.69 on Lambda (self-serve instance), $7.99 on RunPod, $8.50 on Nebius, and $8.60 on CoreWeave — where that last figure is an 8-GPU HGX node at $68.80/hr divided by eight. The on-demand spread is unusually tight for a flagship GPU (about 12% from cheapest to most expensive), because the real price differences hide in how each provider sells Blackwell: single pods, managed clusters on term contracts, or full nodes. The B200 carries 180GB of memory against the H100's 80GB — and, on the one provider that lists both as single GPUs, it costs almost exactly twice as much per hour.
The B200 price table
All rates are per GPU per hour, US dollars, as published on each provider's pricing page on November 11, 2026. CoreWeave sells B200s only as 8-GPU nodes, so its node price and the per-GPU equivalent are both shown.
| Provider | B200 offering | On-demand $/GPU/hr | Spot / interruptible | How it is sold |
|---|---|---|---|---|
| Lambda | HGX B200 180GB (instance) | $7.69* | — (reserved capacity via sales) | Self-serve instances, per-GPU rate |
| RunPod | B200 180GB (Secure Cloud) | $7.99 | — | Self-serve pods, billed per second |
| Nebius | HGX B200 | $8.50 | Preemptible from $0.99 (dynamic spot pricing) | Per-GPU instances, self-serve console |
| CoreWeave | HGX B200 180GB (8-GPU node) | $8.60 (node: $68.80/hr) | Spot node $34.11/hr ($4.26/GPU) | Full nodes and clusters, North America list |
* Lambda's pricing page lists $7.69/GPU/hr with a note that the posted rates take effect November 16, 2026; until then, current rates are shown after login. Sources: lambda.ai/pricing, runpod.io/pricing, nebius.com/prices, coreweave.com/pricing — all accessed November 11, 2026.
The cluster tier: where most B200s actually live
Unlike the H100, which rents happily one pod at a time, most B200 capacity is sold as clusters — and cluster prices sit above the single-instance rates, because managed multi-node networking and support are bundled in:
| Provider | Cluster product | $/GPU/hr | Terms |
|---|---|---|---|
| Lambda 1-Click Clusters | 16 GPUs | $9.86 | 2 weeks – 1 year, via sales team |
| Lambda 1-Click Clusters | 64 GPUs | $9.36 | 2 weeks – 1 year, via sales team |
| Lambda 1-Click Clusters | 256+ GPUs | $8.87 | 2 weeks – 1 year, via sales team |
| RunPod Instant Clusters | B200 | $9.59 | Self-serve cluster product |
Two lessons here. First, scale discounts inside the cluster tier are modest — going from 16 to 256+ GPUs on Lambda saves about 10%, not the steep curve buyers often expect. Second, comparing a cluster rate against a single-pod rate is a category error: the cluster buys you the interconnect that makes 64 GPUs train like one machine. Compare clusters against clusters, pods against pods.
What a full month of B200 costs
Run one B200 continuously for a 730-hour month at on-demand rates and the table above becomes:
| Option | $/GPU/hr | One GPU, 24/7 for a month (730 hrs) |
|---|---|---|
| Lambda HGX B200 (instance) | $7.69 | ~$5,614 |
| RunPod B200 (Secure Cloud) | $7.99 | ~$5,833 |
| Nebius HGX B200 | $8.50 | ~$6,205 |
| CoreWeave HGX B200 (per-GPU equivalent) | $8.60 | ~$6,278 (node: ~$50,224) |
At these numbers, a single always-on B200 costs roughly what two always-on H100s cost on the same class of provider — which is exactly the trade-off to price before you commit. Run your own hours through the GPU Cost Calculator, and for steady workloads ask about committed rates first; see how GPU cloud pricing works for the on-demand vs. spot vs. reserved mechanics.
B200 vs H100: is the premium worth it?
RunPod lists both chips as single self-serve GPUs, which makes it the cleanest like-for-like comparison available: B200 at $7.99/hr vs. H100 SXM at $3.99/hr — almost exactly 2x. For that premium the B200 brings 180GB of HBM3e memory against 80GB, plus the Blackwell architecture's low-precision formats. Our full H100 rental price comparison has the H100 side of the ledger.
- Pay the premium when memory is the wall: models, batch sizes, or context lengths that do not fit in 80GB either run on a B200 or get split across multiple H100s — and splitting has its own cost in complexity and interconnect.
- Pay it when your stack uses Blackwell formats: workloads tuned for Blackwell's low-precision tensor paths can finish enough faster that the cost per finished job drops even at twice the hourly rate. Cost per hour is the wrong unit; cost per completed run is the right one.
- Skip it when the H100 already fits and finishes: if your job fits in 80GB and runs at good utilization on Hopper, two H100-hours still cost about the same as one B200-hour — and parallel H100s are easier to find.
The fine print worth knowing
- Lambda's rates are dated. The $7.69/GPU/hr instance rate and the cluster rates on Lambda's pricing page are posted as taking effect November 16, 2026; the page directs logged-in users to current pricing until then. We list the posted rates because they are the only public figures, and flag the date so you can verify.
- Nebius's preemptible floor is not a price. B200 preemptible capacity is listed "from $0.99/GPU/hr," but Nebius states its spot prices are dynamic and can change as often as every 15 minutes. Budget against the $8.50 on-demand rate; treat anything below it as a bonus for checkpointable work.
- CoreWeave's unit is the node. $68.80/hr buys a complete 8-GPU HGX B200 node; the $8.60/GPU figure is a conversion for comparison, not a product you can buy one of. Its spot node at $34.11/hr works out to $4.26/GPU — the cheapest Blackwell entry in this table, for work that can be interrupted.
- Scarcity is part of the price. B200 is the current flagship and the tightest-supplied GPU in these catalogs; listed rates assume the capacity is actually available in your region when you need it. Our provider directory tracks where to check each cloud's current numbers.
- Storage and egress sit outside the GPU hour. Every provider here charges separately for storage. Our guide to reading a GPU cloud invoice walks through every line beyond GPU-hours.
Run the numbers for your workload: multiply these rates by your real GPU-hours in our GPU Cost Calculator — and always verify the live rate on the provider's pricing page before you commit.
Frequently asked questions
How much does it cost to rent a B200 GPU per hour?
As of November 11, 2026, self-serve on-demand B200 rates run from $7.69/GPU/hr on Lambda and $7.99/hr on RunPod to $8.50/GPU/hr on Nebius and $8.60/GPU/hr on CoreWeave (priced as part of an 8-GPU node at $68.80/hr). Every rate was read from the provider's official pricing page that day.
How much does a B200 cost for a full month?
Running one B200 around the clock (730 hours) at on-demand rates costs roughly $5,614/month on Lambda, $5,833/month on RunPod, $6,205/month on Nebius, and $6,278/month at CoreWeave's per-GPU rate — about $50,224/month for a full 8-GPU CoreWeave node. Cluster contracts and reservations usually beat pure on-demand for steady work.
Is a B200 worth roughly twice the price of an H100?
On RunPod, a B200 at $7.99/hr costs almost exactly twice an H100 SXM at $3.99/hr, and it brings 180GB of memory against the H100's 80GB plus the newer Blackwell architecture. It is worth it when your model or batch size is memory-bound or when Blackwell-specific low-precision formats speed your job up; for work that already fits and runs well on an H100, the H100 is usually the cheaper hour.
Why do Lambda's B200 cluster prices differ from its instance price?
Lambda sells B200s two ways: self-serve instances at $7.69/GPU/hr, and 1-Click Clusters — production clusters of 16 to 2,000+ GPUs on 2-week to 1-year terms — at $9.86/GPU/hr for 16 GPUs, $9.36 for 64, and $8.87 for 256 or more. The cluster product bundles managed multi-node networking and support, so it prices above the raw instance.
How current are these B200 prices?
Every rate in this guide was read from the provider's official pricing page on November 11, 2026. GPU cloud prices change frequently — Lambda's listed rates, for example, are posted as effective November 16, 2026 — so always confirm the live rate on the provider's pricing page before budgeting.
Related guides
- H100 Rental Prices Compared: What One H100 Hour Actually Costs Across Providers
- CoreWeave vs Lambda Labs: Pricing and Differences Explained
- A100 vs H100 vs L40S, Explained for Beginners
- How GPU Cloud Pricing Works: On-Demand vs Spot vs Reserved
- GPU Cloud Provider Directory: Where to Check Real Prices