Do we know the environmental cost of where and with whom we run our AI?
Enhanced: 22 controls
Who it applies to
Enter for every AI system whose letter is E or A (this area has no Baseline controls). Go as deep as your system letter, and implement in your mode.
Why it exists
Make environmental efficiency part of how you choose AI providers and where you run — because the same workload can carry very different carbon costs depending on whose infrastructure runs it and in which location. Providers differ in the efficiency of their infrastructure and in the carbon intensity of the energy behind it; regions differ in how their electricity is generated. These are procurement variables you can ask about and choose on, and they are invisible only because no one puts them in the questionnaire.
What to actually do
Ask providers about the energy and carbon behind their AI service. Add to the supplier due-diligence questionnaire from the Sourcing dimension the questions that surface environmental cost: the efficiency of the infrastructure, the carbon intensity of the energy, the location options. What you do not ask, you cannot choose on.
Choose lower-impact locations where you have the choice. Where the provider offers a choice of region and your data-residency obligations permit, prefer locations with cleaner energy. This is often a free or near-free reduction, forgone only because no one considered it.
Controls
Enhanced2 controls
Control 215
Include environmental questions in AI supplier due diligence.
EnhancedPreventiveall rolesper system
What it is
Energy efficiency, carbon intensity, and location options.
How to implement it · Minimum structure
Infrastructure efficiency: does the provider publish data-centre efficiency figures for this service?
Energy: share of renewable or low-carbon electricity, and its basis (own generation, long-term contracts, certificates)
Carbon intensity: emissions per unit of compute or per request, where available
Location options: regions in which the service runs and which ones the customer can select
Reporting: can the provider give usage-based energy or emissions figures for your account?
Unanswered questions recorded as 'not disclosed', never left blank
How to start
Add these questions as a section of the AI supplier due-diligence questionnaire used in sourcing, and send it at the next new contract or renewal.
Evidence
Completed questionnaires with the environmental answers, on file for each AI supplier.
Mode: light / standard / full
light
Whoever buys the tool emails the questions to the provider and files the reply with the contract.
standard
Procurement includes them in the standard questionnaire and records answers in the supplier register.
full
Procurement scores the answers in supplier selection, reviewed by the sustainability function.
Control 216
Prefer lower-carbon locations where residency and cost permit.
EnhancedPreventiveall rolesper system
What it is
A default consideration in deployment, not an afterthought.
How to implement it · Steps
List the regions the provider offers for the service
Remove those excluded by data-residency obligations or by response-time requirements
Rank the rest by carbon intensity of the local electricity, using provider or public grid figures
Choose the lowest-carbon region unless the cost increase exceeds a threshold set in advance (e.g. 10%)
Record the choice and reason; revisit at renewal or when the provider opens new regions
How to start
For each system with a region setting, check which region it runs in today and which alternatives the provider allows.
Evidence
The location decision per system, with the residency constraints, options considered and reason.
Mode: light / standard / full
light
Whoever configures the tool picks the region once with these steps and notes it in the inventory.
standard
The system owner documents the decision at deployment; IT checks it against residency rules.
full
Region is a mandatory deployment field, with lower-carbon regions as the platform default.
Evidence
The environmental questions in the due-diligence records; the location decisions and the reasoning behind them.