GPU comparison
Ironwood vs B200: specs and rental prices
Google Cloud TPU Ironwood (TPU7x) against NVIDIA B200: 192 GiB against 180 GB of memory, lowest listed on-demand price $12.00 against $6.00 per GPU-hour or chip-hour. Specs from Google and NVIDIA, prices as of Oct 29, 2026.
In short
- Memory type is HBM on the Ironwood and HBM3E on the B200.
- The B200 lists 7,700 GB/s of memory bandwidth against 7,380 GB/s for the Ironwood: 1.04× higher.
- Not every figure can be compared: at least one vendor page does not publish: TF32 tensor throughput; FP64 throughput; Maximum power.
- The lowest listed on-demand price is $6.00 per GPU-hour for the B200 (Hyperstack) and $12.00 per chip-hour for the Ironwood (Google Cloud).
This summary is generated from the tables below. Prices were last checked Oct 29, 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.
Specifications side by side
| Specification | Ironwood (TPU7x) | B200, per GPU | Difference |
|---|---|---|---|
| GPU memory | 192 GiB | 180 GB | — |
| Memory type | HBM | HBM3E | Different |
| Memory bandwidth | 7,380 GB/s | 7,700 GB/s | B200 1.04× higher |
| FP8 tensor throughput | 4,614 TFLOPS (basis not stated) | 9,000 TFLOPS (with sparsity) | Not compared: the vendors state the figures on different bases |
| FP16 / BF16 tensor throughput | 2,307 TFLOPS (basis not stated) | 4,500 TFLOPS (with sparsity) | Not compared: the vendors state the figures on different bases |
| TF32 tensor throughput | Not published | 2,200 TFLOPS (with sparsity) | — |
| FP64 throughput | Not published | 37 TFLOPS | — |
| Maximum power | Not published | 1,000 W | — |
| GPU-to-GPU interconnect | 1,200 GB/s | 1,800 GB/s | B200 1.5× higher |
Compared versions: Ironwood (TPU7x) and B200, per GPU. The figure in the stronger cell is shaded; power is shown without a winner. Two throughput figures are compared only when both vendors state them on the same basis (both with sparsity, or both dense), and “derived” means a total the vendor publishes divided by the GPU count it publishes. Full tables with sources: Ironwood specifications · B200 specifications.
Rental prices
| Rental price | Ironwood | B200 | Difference |
|---|---|---|---|
| Lowest listed on-demand price | $12.00 per chip-hour · Google Cloud · checked Oct 29, 2026 | $6.00 per GPU-hour · Hyperstack · checked Oct 29, 2026 | Ironwood 2× higher |
| Lowest listed spot price | No fresh listing | $0.99 per GPU-hour · Nebius · checked Oct 29, 2026 | — |
| Price per GB of memory | Not available | $0.033 per GB of GPU memory per hour | — |
| Fresh listings | 4 from 1 providers | 16 from 9 providers | — |
Prices cover every version and instance size the providers list, so they are not tied to the versions compared above; the price per GB of memory divides the lowest fresh on-demand price by the memory of the compared version. Listed prices are not like for like across providers. See every listing: Ironwood prices · B200 prices.
More on these GPUs
Cite this page
Last updated Oct 29, 2026. Figures may change as records are updated — include the date you retrieved them.
| APA | Data Center Census. (2026). Ironwood vs B200. Retrieved from https://datacentercensus.com/gpu-prices/compare/ironwood-vs-b200 |
|---|---|
| MLA | “Ironwood vs B200.” Data Center Census, Oct 29, 2026, https://datacentercensus.com/gpu-prices/compare/ironwood-vs-b200. |
| Link | https://datacentercensus.com/gpu-prices/compare/ironwood-vs-b200 |
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.