SCI Score Calculator

Check the carbon your software causes per API call, user or transaction, by the SCI standard.

Multi-cloudCalculator60 seconds

New here? See what this tool does and how to use it.

Before you start: what to have readyShow

How much work your software did in a month: requests, users or runs.
The servers it ran on and their average CPU use, or a measured energy figure.

Start from a scenario

Each one fills in the fields below for a typical workload of that kind. You can change any of them.

R: one unit of work isWhat the score is per. Choose the unit your users or your business count.

API requests a month

E: the energyEstimate it from the servers, or type a measured figure.

Cloud

Instance type

How many

Hours a month each runs730 is around the clock.

Average CPU useFrom your monitoring. Power rises from idle to full with it.

30%

I: where it runsThe grid's location-based figure. SCI does not allow market-based figures or offsets.

Region

SCI, per API request1.89 mg CO₂e

Running it (E x I): 1.43 mg CO₂e
Making the hardware (M): 0.47 mg CO₂e
Energy a month: 42 kWh

Location-based, at 343 g CO₂e per kWh. Same grid as Google's Northern Virginia region (us-east4), whose published figure it uses (2025).

The formula, step by step

SCI = ((E × I) + M) per R, from the Green Software Foundation’s specification, also ISO/IEC 21031:2024. One month of API requests.

E, energy: 4 servers x 730 h x (4 vCPU x 1.568 W + 16 GB x 0.392 W) x PUE 1.135; 41.57 kWh
E x I, operational: 41.57 kWh x 343 g/kWh; 14,260 g
M, embodied: 4 servers x 730 h x 1.595 g an hour (TE x RR/ToR over 4 years); 4,657 g
SCI: (14,260 g + 4,657 g) / 10,000,000; 0.00189 g per unit

How to lower it

Use less energy for the same work: right-size, raise utilization, switch off idle capacity.
Run where the grid is cleaner: the Greenest Cloud Region Finder ranks every region.
Use less hardware for longer: fewer, busier servers, kept in service longer, cut M.
Buying renewable certificates or offsets does not change an SCI score; that is the point of the measure.

What this model assumes

Show

From the servers: Cloud Carbon Footprint’s watts per vCPU, from idle to full by the CPU use you give, plus memory, times the provider’s PUE.
M from the servers is Cloud Carbon Footprint’s embodied coefficient for the instance: the host’s manufacturing carbon, the instance’s share of it, over a four-year life. Its method takes the largest size in a family as the whole host, so small burstable types such as t3 carry a high share.
Only the servers are counted. Add the networks, storage and end-user devices in your software boundary as measured energy if you need them in the score.
The coefficients and their sources: Cloud Carbon Coefficients.

Want this done on your real bill?

Cloud Cost Audit: We go through your real AWS or Google Cloud bill and hand you a ranked list of what to cut first. From $1,500, fixed. Nothing paid upfront.

See the service

Questions

What is an SCI score?

Software Carbon Intensity: the carbon your software causes for one unit of the work it does, such as one API call or one user a month. The Green Software Foundation published it, and it is now an ISO standard, ISO/IEC 21031:2024. It is a rate, not a total, so it can fall while your business grows.

Why can I not use market-based figures or offsets?

The specification forbids them. An SCI score can only fall by using less energy, running on a cleaner grid or using less hardware, so it measures the software, not the purchasing. Report market-based totals separately.

What is M, and why is it per unit too?

The carbon of making the hardware, shared out: the hardware’s total (TE), times the share of its life you use (TiR over EL) and the share of the machine you reserve (RR over ToR). Here it comes from Cloud Carbon Footprint’s coefficient for the instance, or from figures you type in.

Which functional unit should I choose?

The one that grows with the value your software delivers and that your team already counts: requests for an API, monthly active users for a SaaS product, runs for a batch job. Keep it the same over time so the trend means something.

How is this different from the carbon calculator?

The Cloud Workload Carbon Calculator gives a yearly total with a range, to compare regions. The SCI is a single rate per unit of work, by the standard’s rules, to track the software’s efficiency over time.

More tools for: Carbon and sustainability, Compute and utilization

One score is a snapshot. Finitizer keeps it current.

Finitizer joins your usage, cost and carbon by service every day, so the score per unit of work, and what moved it, is there for every release.

See how Finitizer works

SCI Score Calculator