Chip comparison
Google TPU v4 vs Meta MTIA v2
Side-by-side specs
Straight from each vendor's datasheet. Winner in bold: higher throughput, memory, or transistor count; lower power draw or price. Sparse (2:4) throughput roughly doubles the FP8/INT8 numbers shown.
| Field | Google TPU v4 | Meta MTIA v2 |
|---|---|---|
| Process node | TSMC N7 | TSMC N5 |
| TDP | 192 W | 90 W |
| Memory | 32 GB HBM2 | 128 GB LPDDR5 |
| Memory bandwidth | 1,200 GB/s | 205 GB/s |
| BF16 (dense) | 275 TFLOPS | — |
| INT8 (dense) | 275 TOPS | 708 TOPS |
| Form factor | OAM (per-chip in v4 pod) | PCIe (internal Meta) |
| Announced | 2021-05-18 | 2024-04-10 |
| Released | 2022-05-01 | 2024-05-01 |
Robots running Google TPU v4
No robots publicly running it yet.
Robots running Meta MTIA v2
No robots publicly running it yet.
Data centers with Google TPU v4
Data centers with Meta MTIA v2
Common questions
Google TPU v4 vs Meta MTIA v2: which has more memory?
Meta MTIA v2 carries more HBM (Google TPU v4: 32 GB HBM2; Meta MTIA v2: 128 GB LPDDR5). Memory capacity sets the largest model that fits on one chip without partitioning; memory bandwidth sets inference throughput.
See also: every chip comparison · Google TPU v4 full page · Meta MTIA v2 full page.