---
title: "H100 rental in Uzbekistan: early access | UZ Dynamics"
description: "NVIDIA H100: early access by request; availability, form factor pending. 80 GB HBM3 · 3.35 TB/s · NVLink. LLMs up to 32B in BF16 on one GPU (estimate)."
url: "https://uzdynamics.ai/en/gpus/h100"
lang: "en"
image: "https://uzdynamics.ai/en/gpus/h100/opengraph-image/card?f08c4f789845417d"
alternates:
  ru: "https://uzdynamics.ai/ru/gpus/h100"
  uz: "https://uzdynamics.ai/uz/gpus/h100"
  en: "https://uzdynamics.ai/en/gpus/h100"
---

> For an index of this site's pages, see [llms.txt](https://uzdynamics.ai/llms.txt). A markdown copy of each page is at its address with `.md` appended.

# NVIDIA H100 rental: early access

Hopper · SXM (being confirmed)

We’re collecting early-access requests. We’re placing our capacity in Uzbekistan. We don’t publish prices yet. We’ll confirm availability in our reply.

H100

[Request H100](https://uzdynamics.ai/en/request?source=gpu-page%3Ah100&gpu=h100)

We’ll reply with terms for your workload. Nothing to pay before the contract.

- **Memory per GPU:** 80 GB HBM3
- **Memory bandwidth:** 3.35 TB/s
- **GPU interconnect:** NVLink 900 GB/s

H100 diagram: 5 HBM3 stacks, 80 GB total; The full GH100 chip is designed for 6 stacks; NVLink 900 GB/s

Simplified diagram, not to scale

## NVIDIA H100 specifications

- **Architecture:** Hopper
- **Announced:** 2022
- **Memory:** 80 GB
- **Memory type:** HBM3
- **Memory bandwidth:** 3.35 TB/s
- **FP32:** 67 TFLOPS
- **FP16/BF16, dense:** 989.4 TFLOPS
- **FP8, dense:** 1,978.9 TFLOPS
- **FP4, dense:** none
- **FP64:** 34 TFLOPS
- **FP64 Tensor Core:** 67 TFLOPS
- **RT cores:** no
- **NVENC/NVDEC:** NVENC 0, NVDEC 7
- **MIG:** up to 7 instances
- **GPU interconnect:** NVLink 900 GB/s
- **Power (TDP/TGP):** up to 700 W
- **Form factor:** SXM (being confirmed)

Memory is specified per GPU. Multiple GPUs have separate memory: using a larger model across them requires compatible software and model partitioning. NVLink does not make them one GPU.

Other variants: [H100 NVL](https://www.nvidia.com/content/dam/en-zz/Solutions/Data-Center/h100/PB-11773-001_v01.pdf), [H100 PCIe](https://www.nvidia.com/content/dam/en-zz/Solutions/gtcs22/data-center/h100/PB-11133-001_v01.pdf)

The H100 form factor is still being confirmed; memory, its bandwidth and other specifications depend on it. Specs from NVIDIA’s official materials, checked 26 September 2026.

## The H100 by workload

- Inference: LLMs up to 32B in BF16 on one GPU
- Fine-tuning: QLoRA up to 70B, LoRA up to 14B on one GPU
- Training from scratch: models up to 3B on one GPU
- Rendering: no RT cores
- FP64 computing: 34 TFLOPS, Tensor Core 67 TFLOPS

An estimate from open formulas, not a benchmark or an invoice.

## Servers with the NVIDIA H100

Server configuration (CPU, RAM, disk, network) is being confirmed.

Sizing assumption: up to 8 GPUs per server. We’ll confirm the actual configuration in our reply.

[More than 8 GPUs? Discuss a cluster](https://uzdynamics.ai/en/enterprise?source=gpu-page-cluster%3Ah100&gpu=h100&count=more)

## Sources

- [NVIDIA H100](https://www.nvidia.com/en-us/data-center/h100/)
- [NVIDIA H100 Tensor Core GPU Architecture (whitepaper V1.04)](https://dam-cdn.nvd.orangelogic.com/AssetLink/705n6ur546g0uk43w0117r17n8042d73.pdf)
- [NVIDIA Video Encode and Decode Support Matrix](https://developer.nvidia.com/video-encode-decode-support-matrix)
- [NVIDIA Newsroom, March 22, 2022](https://nvidianews.nvidia.com/news/nvidia-announces-hopper-architecture-the-next-generation-of-accelerated-computing)

## Similar GPUs

- [NVIDIA A100](https://uzdynamics.ai/en/gpus/a100)
  80 GB HBM2e · 2.04 TB/s · NVLink
  H100 leads on 5 of 6 metrics. H100 has more FP32 (67 vs 19.5 TFLOPS) and FP64 Tensor Core throughput (67 vs 19.5 TFLOPS). Equal: memory size.
- [NVIDIA H200](https://uzdynamics.ai/en/gpus/h200)
  141 GB HBM3e · 4.8 TB/s · NVLink
  H200 leads on 2 of 7 metrics. H200 has more memory (141 vs 80 GB) and memory bandwidth (4.8 vs 3.35 TB/s). Equal: FP32, FP16/BF16, FP8, FP64 Tensor Core, NVLink.

Dense FP16/BF16 and FP8 from the H100 SXM5 column of NVIDIA’s whitepaper: for the H200, NVIDIA lists only with-sparsity figures, the same as the H100 SXM’s.

## All GPUs in the catalogue

- [B300](https://uzdynamics.ai/en/gpus/b300)
- [B200](https://uzdynamics.ai/en/gpus/b200)
- [H200](https://uzdynamics.ai/en/gpus/h200)
- [A100](https://uzdynamics.ai/en/gpus/a100)
- [RTX PRO 6000](https://uzdynamics.ai/en/gpus/rtx-pro-6000)
- [L40S](https://uzdynamics.ai/en/gpus/l40s)
- [L4](https://uzdynamics.ai/en/gpus/l4)
- [RTX 6000 Ada](https://uzdynamics.ai/en/gpus/rtx-6000-ada)
- [RTX A6000](https://uzdynamics.ai/en/gpus/rtx-a6000)
- [RTX A5000](https://uzdynamics.ai/en/gpus/rtx-a5000)

[Request H100](https://uzdynamics.ai/en/request?source=gpu-page-end%3Ah100&gpu=h100)

We’ll reply with terms for your workload. Nothing to pay before the contract.

```json
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"UZ Dynamics","item":"https://uzdynamics.ai/en"},{"@type":"ListItem","position":2,"name":"GPU catalogue","item":"https://uzdynamics.ai/en/gpus"},{"@type":"ListItem","position":3,"name":"H100","item":"https://uzdynamics.ai/en/gpus/h100"}]}
```