How do I get Scope 3 emissions for my AWS, Azure or Google Cloud usage in India?

Download the figure from each provider's carbon dashboard: the AWS Customer Carbon Footprint Tool, Microsoft's Emissions Impact Dashboard, or Google Cloud Carbon Footprint. Report it as Scope 3 Category 1, purchased services, and note whether it is location-based or market-based. Where no dashboard figure exists, estimate from usage and label it.

Why public cloud is Scope 3, not Scope 2

You do not buy the electricity that runs the servers; the provider does. Their Scope 1 and 2 for the share of capacity you use becomes your Scope 3 Category 1, purchased goods and services, under the GHG Protocol Scope 3 Standard.

Servers you own in a colocation facility are different: if you pay for the power they draw, that electricity can be your Scope 2. Check your colocation contract before you choose the category.

Where to find the number for each provider

Export the same twelve months as your financial year. Most dashboards run calendar months with a lag, so April to March may need two downloads.

• AWS: the Customer Carbon Footprint Tool in the billing console. AWS revised its method in 2025 and now shows location-based and market-based figures and some Scope 3 categories, such as server hardware; read the current methodology note for exactly which scopes and categories are in.
• Microsoft Azure: the Emissions Impact Dashboard, which needs a one-time connection to your tenant and reports by service and region.
• Google Cloud: Carbon Footprint in the console, with location-based and market-based figures and an export to BigQuery for monthly detail.

Location-based or market-based

Providers that buy renewable power or certificates show a lower market-based figure. Your questionnaire may ask for one or both. Report the basis next to each number, and do not add a market-based AWS figure to a location-based Google figure without saying so.

If a buyer wants a figure comparable with your own India Scope 2, a location-based number for an India region is the closer match.

When there is no dashboard figure

Indian resellers, smaller hosting firms and managed service bills often have no dashboard. Ask them for kWh or their own emissions per invoice before you fall back to spend.

1. Usage-based: vCPU hours, memory and storage times a published power figure per unit, times the data centre's power usage effectiveness (PUE), times the grid factor of the hosting region.
2. Spend-based: annual cloud spend times a spend-based factor for computer services, from a named database. Use it only when usage is not available, because it moves with price, not energy.
3. Label either result as an estimate, with the method, factors and assumptions written down.

Worked example: usage-based estimate for one month in an India region

• Compute: 40 vCPUs running all month, 730 hours
• Assumed average draw: 3 W per vCPU, including memory (illustrative assumption, label it)
• 40 x 730 x 3 = 87,600 Wh = 87.6 kWh at the server
• Assumed PUE: 1.2, so 87.6 x 1.2 = 105.12 kWh at the facility
• Location-based factor: 0.710 kg CO2 per kWh (CEA CO2 Baseline Database, version 21.0)
• 105.12 x 0.710 = 74.6 kg CO2 for the month, about 0.9 tonnes a year

This leaves out the embodied emissions of the hardware. Replace the assumed draw and PUE with provider figures where they are published.

What Senseible does here

Senseible's cloud and data centre emissions calculator turns usage from your billing export, such as vCPU-hours, memory, storage and network transfer, into kWh and location-based and market-based figures by hosting region, for when provider data is missing.

• Cloud and data centre emissions calculator
• Colocation emissions with PUE
• Spend-based or activity-based Scope 3

Sources: GHG Protocol Corporate Value Chain (Scope 3) Standard, Category 1, GHG Protocol Scope 2 Guidance (2015), CEA CO2 Baseline Database

Last reviewed 2026-10-16. Provider carbon tools change method and scope coverage; check each methodology note before each reporting cycle.