What you emit,made visible.

Region-aware carbon estimates for cloud workloads and AI inference, grounded in the Green Software Foundation Software Carbon Intensity spec. Location-based grid intensity - no offset accounting, no certificate maths. Runs in your browser.

Built for honest numbers

Most carbon calculators flatter the provider. This one doesn't.

SCI-aligned methodology

Energy × grid intensity, with PUE applied to rack power. Every coefficient is documented and sourced - Cloud Carbon Footprint, EPA eGRID, EEA, IEA.

Location-based, not market-based

We report the actual grid mix where the workload runs. No provider renewable-energy certificates, no zero-carbon hand-waving.

Cloud and AI together

Compute, storage, egress, and LLM inference in one estimate. Compare a Claude Sonnet workload against an idle GPU instance, side by side.

Region sweep, every region

Hold the workload fixed, sweep across 40+ AWS/Azure/GCP regions, see exactly how much carbon you'd save by moving.

How it works

  1. 01

    Pick a workload type

    Cloud infrastructure, AI inference, or both together. Each mode adapts the inputs to the things that actually move the carbon number.

  2. 02

    Configure the parameters

    Provider, region, instance, hours, storage, egress for cloud. Model, tokens, hosting region for inference. Region matters most - by a wide margin.

  3. 03

    Read the breakdown

    Monthly kgCO₂e, energy in kWh, and the effective grid intensity. Equivalences in km driven and tree-years. Plus a sweep of every region with the same workload so the best move is obvious.

[ - / CTA ]

Spending more
than you should?

Let's find where your cloud and AI spend can work harder.