Use case
What to compare when a workload, battery or building is scheduled against grid carbon intensity rather than a fixed emissions factor.
5 tools in this scenario · verified 2026-10-27
Carbon-aware scheduling only works as well as the intensity signal behind it. A feed that reports a national daily average will not move a training job to a cleaner hour, while a five-minute regional signal can be too noisy to act on and too sparse to cover the sites that matter. The choice of feed therefore decides what an organisation can claim: a location-based hourly series supports operational emissions reporting under the GHG Protocol, but a marginal-emissions measure is what actually answers whether shifting a load reduces emissions. Feeds also differ on geography, history, licensing, latency and whether the data can be redelivered through an API, a webhook or a batched file. Teams running several data centres usually end up combining two sources: a public grid average for coverage, and a commercial marginal signal for the regions where most of the load sits. The criteria below are the ones that change that decision rather than the ones that sound thorough.
1
Signal type: average versus marginal
Average intensity describes the grid as a whole; marginal intensity describes the generator that would respond to a small change in demand. Shifting load reduces emissions only if the marginal signal is used, while both are defensible for reporting.
2
Temporal resolution and lag
Hourly is enough for batch scheduling, five-minute windows are needed for short jobs and interruptible workloads. Check how long after the interval a value is published, because a signal delivered late is only useful for planning, not for control.
3
Geographic granularity
Country-level data hides the difference between bidding zones. Confirm which exact zones are covered for your facilities rather than which countries appear in a demo, and whether sub-zones exist where your load is.
4
Methodology and audit trail
Ask for the published methodology, the data sources behind it, and whether historical values are revised. A feed that cannot explain how a number was produced will not survive an assurance review.
5
Delivery and cost model
Compare an API with quotas, a batch export and a webhook push. Pricing tiers often depend on historical depth and zones, so price the two or three zones you actually run in rather than the headline coverage.
Carbon-Aware Computing
Per-country grid signals sold by the year: carbon intensity, carbon-free energy share and wholesale prices, hourly.
Carbon-Aware Computing
Regional forecast API for Great Britain's grid carbon intensity and generation mix.
Carbon-Aware Computing
JavaScript library for estimating the carbon emissions of digital services and website visits.
Sustainable Cloud
Open-source Web API and CLI for shifting software to lower-carbon times and regions.
Carbon-Aware Computing
Estimate and monitor the carbon footprint of web application user scenarios, including in CI.
Last verified 2026-10-27. Figures on the linked tools are checked against the sources above; confirm pricing and compliance coverage before signing.