AI generation uses energy. Code generation is lightweight; image and video generation are heavier. We offset what we can and are working toward carbon-neutral operations by 2027.
We estimate energy consumption per generation type based on our infrastructure and AI provider reports. These are approximate figures and vary with model size and infrastructure efficiency:
| Format | Est. Energy/Request | Est. CO₂-eq | Notes |
|---|---|---|---|
| Code | 0.02 kWh | ~8g CO₂ | Lightweight inference on efficient models |
| Image | 0.15 kWh | ~60g CO₂ | GPU-based generation, image models |
| Music | 0.1 kWh | ~40g CO₂ | Audio model inference |
| Video | 0.8 kWh | ~320g CO₂ | Highest energy use due to compute |
| App (code + preview) | 0.05 kWh | ~20g CO₂ | Multiple code generations + image preview |
CO₂-equivalent estimates assume US average grid intensity of ~400g CO₂/kWh. Estimates updated quarterly as model efficiency improves.
We purchase carbon offsets equivalent to our estimated operational energy use (hosting, compute, office operations). Offsets are sourced from verified projects (Gold Standard, Verra). We are committed to transitioning to carbon-neutral operations by Q2 2027 through renewable energy procurement.
Skylina runs on cloud infrastructure that is continuously optimized for efficiency. Our hosting providers use renewable energy for a portion of their operations. We evaluate and select infrastructure partners based on their renewable energy commitments.
We choose AI models that balance capability with energy efficiency. We monitor energy consumption per generation and will report changes as our model stack evolves. Updates to this disclosure will be made within 30 days of any material change to our generation infrastructure.
Energy estimates are derived from: (a) AI provider transparency reports (Anthropic publishes energy usage data); (b) our own infrastructure monitoring (cloud compute billing data); (c) industry benchmarks for comparable model sizes. We review and update estimates quarterly.
Questions about our environmental impact: env@skylina.polsia.app