DeepSeek-V3 H800 training cluster
DeepSeek used this cluster to train DeepSeek-V3. Its paper documents the training configuration rather than a current fleet inventory.
DeepSeek-V3 H800 training cluster is a GPU cluster run by DeepSeek. It has 2,048 NVIDIA H800 GPUs installed.
Facts
Source per value- Owner / operator Organization that owns or operates the GPUs. Hosting and facility roles are listed under Partners.
- DeepSeek 1
- GPU maker NVIDIA, AMD or Intel, taken from the GPU model.
- NVIDIA 1
- GPU model Accelerator model or family as disclosed. A GH200 is one GPU on a superchip; an MI300A is an APU that contains a GPU.
- NVIDIA H800 1
- GPUs installed Physical GPUs in the cluster, from the operator or a deterministic sum of disclosed quantities (nodes × GPUs per node). Planned counts, H100-equivalents and unaudited press totals are never counted here.
- 2,048 1
- GPU count notes What the count covers (one partition, one site, a design limit), qualifiers and competing figures.
- Published V3 training cluster, not DeepSeek total GPU holdings. 1
- Location City, region and country where the cluster sits. Empty when the operator does not name the site.
- Not published
- Live power MW Power in service, in megawatts of IT load where the source says so. Company figures are marked in the notes; Epoch AI satellite estimates are labelled as estimates.
- Not published
- Status notes The source’s wording on status, with the date.
- Documented in the DeepSeek-V3 paper as the training configuration; current status is not published. 1
- Workload Main workload as stated by the source.
- DeepSeek-V3 pre-training 1
Sources
- arxiv.org accessed 2026-11-11
Related clusters
All 87 clustersLists
Need GPU capacity at this scale? Request quotes from GPU cloud providers
Tell us what you need once. We send it to matching GPU cloud providers and they reply to you directly.