Home › Chips › NVIDIA H100

NVIDIA H100

Chip GPU Verzending

The GPU of the ChatGPT era — the unit of account for the 2023–25 AI build-out.

U bekijkt een chip-record: wat deze processor is, de belangrijkste specificaties, wie hem ontwerpt, produceert en van onderdelen voorziet, en waar hij wordt ingezet.

Hopper's H100 became the defining commodity of the AI boom: clusters were sized in H100s, startups raised funding denominated in them, and export policy was written around them. Its Transformer Engine pairs fourth-generation tensor cores with FP8 precision, roughly tripling A100 training throughput per chip.

xAI's Colossus assembled roughly 100,000 of them in 122 days; Meta, Microsoft, OpenAI, Google and every major cloud deployed them in the hundreds of thousands. Blackwell succeeded it in 2025, but H100/H200 fleets still carry a large share of world AI compute.

Specificaties & kernfeiten

Elk cijfer is voorzien van een bronvermelding: vendor spec, vendor claim, reported, estimate of official. "Not publicly disclosed" betekent precies dat — wij verzinnen nooit waarden.

Overview

Designer NVIDIA

Architecture Hopper (GH100)

Announced March 2022 (GTC); shipping volume from early 2023

Status Shipping — succeeded by Blackwell at the frontier

Manufacturing

Foundry / process TSMC 4N (custom 5 nm-class)Vendor spec

Transistors 80 billionVendor spec

Die size 814 mm²Vendor spec

Packaging CoWoS-S with 5 or 6 HBM stacksVendor spec

Geheugen

Memory 80 GB HBM3 (SXM); 94 GB HBM3 (NVL); 80 GB HBM2e (PCIe)Vendor spec

Memory bandwidth 3.35 (SXM) / 3.9 (NVL) TB/sVendor spec

Rekenkracht

TF32 tensor (dense) 495 TFLOPSVendor spec NVIDIA datasheets often headline throughput with structured sparsity; figures here are dense unless noted.

BF16/FP16 tensor (dense) 990 TFLOPSVendor spec

FP8 tensor (dense) 1,979 TFLOPSVendor spec FP8 via the Transformer Engine was Hopper's signature feature

INT8 tensor (dense) 1,979 TOPSVendor spec

FP64 tensor 67 TFLOPSVendor spec

AI Performance

MLPerf Set the reference results for training and inference benchmarks across 2023–24 submissionsMLPerf

Vermogen

TDP up to 700 (SXM) / 350 (PCIe) WVendor spec

Physical

Form factors SXM5, PCIe, NVL (paired PCIe)

Interconnect NVLink 4 — 900 GB/s; PCIe 5.0Vendor spec

Koers

Launch price Not publicly disclosed

Market price ~$25,000–40,000 per GPU at the 2023–24 peakReported

Cloud rental Roughly $2–7 per GPU-hour across major and specialist clouds during 2025 — falling steadilyReported varies with commitment and provider

Hoe dit aansluit op de rest van de stack

De koppelingen van dit record in één oogopslag — klik op een groep om er naartoe te springen. +11 meer groepen hieronder.

Ontworpen door

NVIDIA

Geproduceerd door

TSMC

Procesnode

TSMC N5 / 4N (5 nm class)

Behuizing

CoWoS

Behuizingsleverancier

TSMC

Geheugentechnologie

HBM3

Geheugen geleverd door

SK hynix

Interconnect & netwerken

NVLink & NVSwitch InfiniBand

Aangetroffen in

Dell PowerEdge XE9680/XE9712 Supermicro NVL72 & 4U Liquid-Cooled Systems NVIDIA HGX H100/H200

Runs

Triton (OpenAI) PyTorch Megatron-Core vLLM

Ondersteund door

CUDA

Gebruikt om te trainen

Llama 3.1 405B Llama 4 Grok 3

Beschikbaar vanaf

Amazon Web Services Microsoft Google (Alphabet) Oracle CoreWeave Lambda

Ingezet bij

xAI Colossus

Basis voor

NVIDIA H800

Opgevolgd door

NVIDIA H200

Opvolger van

NVIDIA A100

Concurreert met

AMD Instinct MI300X Google TPU v5p Groq LPU AWS Trainium2 Intel Gaudi 3 Huawei Ascend 910C

Gerelateerd

Export Controls & the Chip War Training & Pre-training MLPerf

Bronnen

NVIDIA H100 page ↗ · Hopper architecture whitepaper ↗

Record beoordeeld per 2026-01-15 — bekijk Methodologie.