compute · United States
NVIDIA
The dominant supplier of accelerators for frontier AI training and inference, spanning GPUs, NVLink interconnect, networking and the CUDA software stack.
- Founded
- 1993
- Chips tracked
- 21
- Market cap
- $5.30T
- Revenue growth
- +85% YoY
Stock Price
NVDA by TradingView ↗Landscape
AI chips & accelerators| Chip | Architecture | Launch date | Memory capacity | Memory bandwidth |
|---|---|---|---|---|
| GB300 NVL72 (per GPU) | Blackwell Ultra + Grace | 2025-H2 | 288 GB HBM3E | 8 TB/s |
| Blackwell Ultra GB300 | Blackwell Ultra | 2025-H2 | 288 GB HBM3E | 8 TB/s |
| RTX PRO 6000 Blackwell Server | Blackwell (RTX PRO) | 2025-03 | 96 GB GDDR7 | 1.6 TB/s |
| GH200 Grace Hopper 144GB | Hopper + Grace | 2024-Q2 | 144 GB HBM3E | 4.9 TB/s |
| H200 NVL | Hopper | 2024-11 | 141 GB HBM3E | 4.8 TB/s |
| B200 | Blackwell | 2024-03 | 192 GB HBM3E | 8 TB/s |
| GB200 Grace Blackwell | Blackwell + Grace | 2024-03 | 384 GB HBM3E (2 GPUs) | 16 TB/s (2 GPUs) |
| H200 | Hopper | 2023-11 | 141 GB HBM3E | 4.8 TB/s |
| L40S | Ada Lovelace | 2023-08 | 48 GB GDDR6 | 864 GB/s |
| H100 NVL | Hopper | 2023-03 | 188 GB HBM3 (2 GPUs) | 7.8 TB/s (2 GPUs) |
| L4 | Ada Lovelace | 2023-03 | 24 GB GDDR6 | 300 GB/s |
| L40 | Ada Lovelace | 2022-10 | 48 GB GDDR6 | 864 GB/s |
| H100 PCIe | Hopper | 2022-09 | 80 GB HBM2e | 2.0 TB/s |
| H100 SXM | Hopper | 2022-09 | 80 GB HBM3 | 3.35 TB/s |
| A30 | Ampere | 2021-04 | 24 GB HBM2 | 933 GB/s |
| A10 | Ampere | 2021-04 | 24 GB GDDR6 | 600 GB/s |
| A100 80GB | Ampere | 2020-11 | 80 GB HBM2e | 2.0 TB/s |
| A100 40GB PCIe | Ampere | 2020-05 | 40 GB HBM2 | 1.56 TB/s |
| T4 | Turing | 2018-09 | 16 GB GDDR6 | 320 GB/s |
| V100 SXM2 32GB | Volta | 2018-03 | 32 GB HBM2 | 900 GB/s |
| P100 SXM | Pascal | 2016-04 | 16 GB HBM2 | 732 GB/s |
Datacenter-class accelerators as published by the vendor · memory figures are per-package unless stated.
Model Benchmarks
data by
| Model | Released | Intelligence | Coding | GPQA | HLE | SciCode | Tau2 | LCR | IFBench | Speed | TTFT | $/1M in | $/1M out | $/1M blended |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Nemotron 3 Ultra 550B A55B (Reasoning) | 2026-06-04 | 22.9 | 49.3 | 86.7% | 28.4% | 40.3% | 83.3% | 79.3% | 81.4% | 158.1 t/s | 1.27s | $0.6 | $2.4 | $1.05 |
| Nemotron 3.5 Lightning | 2026-08-11 | 12.9 | 26.8 | 74.3% | 10.6% | 32.1% | — | 60.3% | — | 287.1 t/s | 0.46s | $0.06 | $0.2 | $0.1 |
| Nemotron 3 Super 120B A12B (Reasoning) | 2026-03-11 | 12.8 | 37.7 | 80% | 20.8% | 36.2% | 67.8% | 65.7% | 71.5% | 199 t/s | 1.18s | $0.19 | $0.65 | $0.31 |
| Nemotron Cascade 2 30B A3B | 2026-03-19 | 11.7 | 25.3 | 75.8% | 12% | — | 53.2% | 39.3% | 80.4% | 0 t/s | 0s | $0 | $0 | $0 |
| Nemotron 3 Nano Omni 30B A3B Reasoning | 2026-04-29 | 10.3 | 13.8 | 46.9% | 4.8% | — | 45.3% | 39.7% | 63.2% | 0 t/s | 0s | $0.09 | $0.36 | $0.16 |
| Llama Nemotron Super 49B v1.5 (Reasoning) | 2025-07-25 | 9 | — | 74.8% | 7.3% | — | 28.1% | — | 37% | 130.1 t/s | 3.43s | $0.4 | $0.4 | $0.4 |
| NVIDIA Nemotron 3 Nano 30B A3B (Reasoning) | 2025-12-15 | 8.9 | 14.4 | 75.7% | 11.4% | 30.6% | 40.9% | 38% | 71.1% | 116.8 t/s | 0.67s | $0.05 | $0.2 | $0.09 |
| Llama 3.3 Nemotron Super 49B v1 (Reasoning) | 2025-03-18 | 8.9 | — | 64.3% | 5.6% | — | 26.9% | 18.3% | 38.1% | 0 t/s | 0s | $0 | $0 | $0 |
| Llama 3.1 Nemotron Ultra 253B v1 (Reasoning) | 2025-04-07 | 7.5 | — | 72.8% | 7.4% | — | 11.4% | — | 38.2% | 0 t/s | 0s | $0 | $0 | $0 |
| NVIDIA Nemotron Nano 12B v2 VL (Reasoning) | 2025-10-28 | 7.5 | — | 57.2% | 5.5% | — | 21.3% | 41% | 31.9% | 113.6 t/s | 4.98s | $0.2 | $0.6 | $0.3 |
Indices on Artificial Analysis 0-100 scale · eval scores are raw accuracy · “—” means the metric is not published for that model.
Release Timeline
last 48 months- 2026-09-19Is Nvidia Or Intel Stock The Better Buy Heading Into Year End?AcquisitionForbes
- 2026-09-18Huawei Speeds Up Its Ascend 960 AI Chip Launch to Take On Nvidia in ChinaProduct LaunchStartup Fortune
- 2026-09-17Cute Critters Come to the Cloud: ‘Aniimo’ Launches on GeForce NOWProduct LaunchNVIDIA
- 2026-09-17Huawei plans Q1 2027 launch of new AI chip as it takes on NvidiaEarningstechcrunch.com
- 2026-09-17Huawei unveils new chip technologies as Chinese firm steps up the AI race with NvidiaProduct LaunchAP News
- 2026-09-17Apple eyes Nvidia NVLink to power its new custom M8 Ultra AI servers — historically bitter rivals reportedly team up for 2029 data center pushProduct LaunchTom's Hardware
Product launches, partnerships, acquisitions, earnings and other milestones from the official newsroom plus major tech press. Source shown on each row.
Newsroom
source ↗- 2026-09-17Cute Critters Come to the Cloud: ‘Aniimo’ Launches on GeForce NOWModel Launch
- 2026-09-16NVIDIA Vera Rubin NVL72 Delivers Leading Performance in MLPerf Inference v6.1 DebutModel Launch
- 2026-09-16Emerald AI, Google and NVIDIA Launch Alliance to Advance Flexible AI Data CentersCompute
- 2026-09-16University of Manchester Uses NVIDIA Earth-2 to Forecast Air Pollution Across the UKModel Launch
- 2026-09-15‘Now We Can Know Everything and Do Anything,’ Jensen Huang Says at DreamforceRelease
- 2026-09-15From Megawatts to Tokens: How NVIDIA Maximizes AI Factory ProductionRelease
- 2026-09-15AI Infra Summit: NVIDIA Vera Rubin and DSX Platform Advancements Showcase Energy Efficiencies of Optimizing Tokens Per Watt for AI FactoriesRelease
- 2026-09-15Heart of the Matter: How a Major Children’s Hospital Uses Open Source NVIDIA AI for Cardiac CareRelease