AI chip · NVIDIA
NVIDIA Jetson AGX Orin
Ampere-based embedded AI compute module (32-64 GB), the dominant robotics-side Nvidia SKU 2022-2025 before Jetson Thor ramp.
Powers 9 tracked robots.
Market position
AGX Orin is the humanoid + delivery-robot standard chip of 2023-2025: 275 TOPS in a 60 W automotive module. Figure, Apptronik Apollo, Agility Digit and dozens of Chinese humanoids shipped on it before Thor became available.
What fits in 64 GB
Which open-source LLMs run on one NVIDIA Jetson AGX Orin, by precision. Weights only: add roughly 15% headroom for real serving. If a model does not fit at FP16, try INT8 or INT4 (smaller quality trade-off than most people expect).
| Model | Params | FP16 | INT8 | INT4 |
|---|---|---|---|---|
| Llama 3.1 8B dense | 8 B | 16 GB ✓ | 8 GB ✓ | 4 GB ✓ |
| Llama 3.1 70B dense | 70 B | 140 GB ✗ | 70 GB ✗ | 35 GB ✓ |
| Llama 3.1 405B dense | 405 B | 810 GB ✗ | 405 GB ✗ | 203 GB ✗ |
| Llama 3.3 70B dense | 70 B | 140 GB ✗ | 70 GB ✗ | 35 GB ✓ |
| DeepSeek V3 MoE (671B total, 37B active per token) | 671 B | 1342 GB ✗ | 671 GB ✗ | 336 GB ✗ |
| DeepSeek R1 MoE (671B total, 37B active per token) | 671 B | 1342 GB ✗ | 671 GB ✗ | 336 GB ✗ |
| Qwen 2.5 7B dense | 7 B | 14 GB ✓ | 7 GB ✓ | 4 GB ✓ |
| Qwen 2.5 72B dense | 72 B | 144 GB ✗ | 72 GB ✗ | 36 GB ✓ |
| Mixtral 8x7B MoE (46.7B total, 12.9B active per token) | 46.7 B | 93 GB ✗ | 47 GB ✓ | 23 GB ✓ |
| Mixtral 8x22B MoE (141B total, 39B active per token) | 141 B | 282 GB ✗ | 141 GB ✗ | 71 GB ✗ |
| Gemma 2 27B dense | 27 B | 54 GB ✓ | 27 GB ✓ | 14 GB ✓ |
| Command R+ dense | 104 B | 208 GB ✗ | 104 GB ✗ | 52 GB ✓ |
| Kimi K2 MoE (1T total, 32B active per token) | 1000 B | 2000 GB ✗ | 1000 GB ✗ | 500 GB ✗ |
Math: FP16 = params × 2 bytes; INT8 = params × 1 byte; INT4 = params × 0.5 bytes. MoE models sum every expert (full weights on disk), not the per-token active subset.
Common questions
How much does NVIDIA Jetson AGX Orin cost?
NVIDIA Jetson AGX Orin doesn't have a public launch list price. Vendors like this one usually price through direct sales rather than a public sheet. See list price per TFLOP for chips that do publish.
How much memory does NVIDIA Jetson AGX Orin have?
64 GB of LPDDR5, running at 204.8 GB/s. Straight from the vendor datasheet. See chips with the most memory for context.
How much power does one NVIDIA Jetson AGX Orin draw?
60 W at the chip. A full server draws more once you add CPU, memory, networking and cooling: see the rack power number above. Compare to other chips on perf-per-watt.
What runs on NVIDIA Jetson AGX Orin?
9 robot models we track use this chip, including Spot, Serve Gen 3 and Unitree G1, plus 6 more. Full list below.
Which open-source LLMs fit on one NVIDIA Jetson AGX Orin?
Of 13 open-source LLMs we track, 3 fit at FP16, 4 at INT8, and 8 at INT4 (for example Llama 3.1 8B, Llama 3.1 70B and Llama 3.3 70B). Full table above with each model's memory need.
Key facts
- Class
- Compute SoC (AI accelerator)
- Designer
- NVIDIA
- Safety-critical?
- Yes (embedded / automotive)
- Record as of
- 2026-09-27
Specifications (10 fields, click to expand)
Straight from the vendor datasheet. Dense throughput shown first; sparse (2:4) numbers in parentheses where the vendor publishes them. Full datasheet linked below.
- Process node
- Samsung 8N
- CUDA cores
- 2,048
- Tensor cores
- 64 (3rd gen (Ampere))
- TDP
- 60 W
- Memory
- 64 GB LPDDR5
- Memory bandwidth
- 204.8 GB/s
- INT8 (dense)
- 275 TOPS
- Form factor
- Module (AGX Orin 64GB)
- Announced
- 2022-03-22
- Released
- 2022-07-01
Source: vendor datasheet
Generation
Foundry & process
- Fabbed at
- TSMC (all chips TSMC makes →)
- Process node
- Samsung 8N
Compare with
Robots running NVIDIA Jetson AGX Orin
Every robot model publicly known to ship NVIDIA Jetson AGX Orin inside it. Ranked by how widely each has been deployed.
| Model | Maker | Form factor | Role | Source | Deployments |
|---|---|---|---|---|---|
| Spot | Boston Dynamics | quadruped | onboard AI compute (CORE I/O) | reported | 29 |
| Serve Gen 3 | Serve Robotics | sidewalk | primary inference compute | reported | 15 |
| Unitree G1 | Unitree Robotics | humanoid | developer-kit inference option | reported | 10 |
| May Mobility Autonomous Sienna | May Mobility | av | onboard AV compute | reported | 9 |
| Cruise AV (Chevrolet Bolt) | Cruise | av | primary inference compute | reported | 6 |
| Unitree H1 | Unitree Robotics | humanoid | developer-kit inference option | reported | 5 |
| Digit v5 | Agility Robotics | humanoid | primary inference compute | reported | 4 |
| Booster T1 | Booster Robotics | humanoid | onboard humanoid compute | reported | 2 |
| Cassie | Agility Robotics | humanoid | onboard bipedal compute | inferred | 0 |
Sources
- Jetson OrinNvidia
Adoption rows appear as we confirm each chip-in-robot pairing from a public source. See every chip we track for the full catalog or the NVIDIA page.