AI chip · Tesla
Tesla HW4 (FSD Computer 2)
Tesla's second-generation in-vehicle FSD compute board (2023-present); shipping in every new Model S/3/X/Y as the AV inference substrate. Believed fabricated by Samsung Foundry.
Powers 3 tracked robots.
Market position
The FSD Computer 2 chip Tesla shipped in every Model S/3/X/Y/Cybertruck from 2023 onward: dual Samsung 7nm SoCs, ~300 TOPS. Superseded by HW5 (AI5) in the Cybercab. Every Tesla Robotaxi vehicle on the road today runs HW4.
What fits in 16 GB
Which open-source LLMs run on one Tesla HW4 (FSD Computer 2), 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 Tesla HW4 (FSD Computer 2) cost?
Tesla HW4 (FSD Computer 2) 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 Tesla HW4 (FSD Computer 2) have?
16 GB of GDDR6. Straight from the vendor datasheet. See chips with the most memory for context.
How much power does one Tesla HW4 (FSD Computer 2) draw?
90 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 Tesla HW4 (FSD Computer 2)?
3 robot models we track use this chip, including Tesla Robotaxi (Model Y ADS), Tesla Optimus and Cybercab. Full list below.
Which open-source LLMs fit on one Tesla HW4 (FSD Computer 2)?
Of 13 open-source LLMs we track, 2 fit at FP16, 2 at INT8, and 3 at INT4 (for example Llama 3.1 8B, Qwen 2.5 7B and Gemma 2 27B). Full table above with each model's memory need.
Key facts
- Class
- Compute SoC (AI accelerator)
- Designer
- Tesla
- Safety-critical?
- Yes (embedded / automotive)
- Record as of
- 2026-09-27
Specifications (7 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 7LPP
- TDP
- 90 W
- Memory
- 16 GB GDDR6
- INT8 (dense)
- 300 TOPS
- Form factor
- Auto-board (dual-SoC)
- Announced
- 2023-01-01
- Released
- 2023-02-01
Source: vendor datasheet
Foundry & process
- Fabbed at
- Samsung Foundry (all chips Samsung Foundry makes →)
- Process node
- Samsung 7LPP
Compare with
Robots running Tesla HW4 (FSD Computer 2)
Every robot model publicly known to ship Tesla HW4 (FSD Computer 2) inside it. Ranked by how widely each has been deployed.
| Model | Maker | Form factor | Role | Source | Deployments |
|---|---|---|---|---|---|
| Tesla Robotaxi (Model Y ADS) | Tesla | av | FSD Computer 2 (dual-SoC) | reported | 13 |
| Tesla Optimus | Tesla | humanoid | primary inference compute | reported | 6 |
| Cybercab | Tesla | av | primary inference compute | reported | 1 |
Sources
- Tesla Autopilot hardwareWikipedia
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 Tesla page.