Grok 3 training emissions
xAI released Grok 3 on February 19, 2025 (source).
- Our estimate29,000–62,000 tCO2e to train
- About as much as6,800–14,000gasoline cars driven for a year (EPA)
How we got the range
xAI has not published what training Grok 3 emitted. We multiplied the compute it took by what each NVIDIA H100 uses and what the grid emits, at the low and high end of each assumption. The method, in full.
| Input | Value | Source |
|---|---|---|
| Training compute | 3.50 × 10²⁶ FLOP | Epoch AI, Data on AI models, Grok 3 (about 80,000 H100s for about three months on Colossus; Epoch's range 2.1e26 to 1.13e27) |
| H100 peak (FP8, dense) | 1.98 × 10¹⁵ FLOP/s | NVIDIA, H100 specifications: 3,958 teraFLOPS FP8 with sparsity, dense is half |
| H100 rated power | 700 W | NVIDIA, H100 specifications: max TDP up to 700 W |
| Utilization, low | 0.3 share of peak | Epoch AI, estimation methods: typical language-model range 0.3 to 0.5 |
| Utilization, high | 0.5 share of peak | Epoch AI, estimation methods: typical language-model range 0.3 to 0.5 |
| PUE, low | 1.1 ratio | Google, data center efficiency: fleet quarterly PUE 1.10, trailing twelve months 1.09 (Q2 2026) |
| PUE, high | 1.4 ratio | LBNL, 2024 United States Data Center Energy Usage Report: U.S. average PUE 1.4 in 2023, p. 47 |
| U.S. grid, location-based | 384.4 gCO2e/kWh | Ember via Our World in Data, carbon intensity of electricity, United States, 2025 |
- xAI says only that Grok 3 trained on its Colossus supercomputer in Memphis with "10x the compute" of earlier leading models. The compute figure is Epoch AI’s estimate, and Epoch’s own range runs from 0.6 to 3.2 times it.
- xAI publishes no emissions or electricity figures, so both ends use the U.S. grid average. Colossus has also run on gas turbines, which this does not count separately.
Also cited: xAI, Grok 3 Beta announcement
Last checked October 17, 2026. Every model