AI Usage ESG Reporting
From personal awareness to organisational accountability.
The Challenge
You understand AI's per-query environmental impact (thanks to our free calculator). But your organisation needs something different: a defensible, audit-ready line item in your Scope 2 / Scope 3 emissions inventory that survives regulatory scrutiny.
Generic spend-based factors (EEIO) overestimate AI emissions by 10–40×. Your CFO and assurer deserve better.
The Solution: Reporting Options
All tiers use the same methodology as our free calculator, anchored to first-party operator data, but tailored for organisational reporting. Public prices are starting points; exact scope, volume, add-ons and commercial rights are confirmed privately after a short fit review.
Your AI usage is split by deployment mode: self-hosted models (Scope 2) vs. purchased third-party AI services (Scope 3, Category 1). The methodology statement is the product.
Generic spend-based (EEIO) factors overestimate AI emissions by 10–40×. This annex is anchored to first-party operator energy disclosures, the same methodology as the free calculator.
Mapped to Australia's NGER scheme and the GHG Protocol, with confidence bands and a versioned methodology statement your assurer can review.
Founding Pilot: Concierge Annex (AU$990)
- Deployment-mode split (Scope 2 self-hosted vs. Scope 3 Cat 1 purchased AI services)
- Physically-derived energy, water and carbon factors with confidence bands
- NGER + GHG Protocol-aligned methodology statement
- Delivered as a PDF annex, typically within 5 business days
Best for: Organisations reporting this year who want a quick, defensible baseline. Availability is capacity-limited and may close as the standard annex offer opens.
Pricing Structure
Enough detail to budget, not enough to publish the full pricing model.
Exact quotes depend on reporting scope, usage evidence quality, integration needs, portfolio size and whether outputs are used internally, with clients, or inside another product. Reporting workspaces are in build and expected to open in Q4 2026.
Report once
A prepared AI-usage annex for one reporting period.
From AU$1,950
per reporting period
Best for: Teams that need a defensible baseline without adopting a new platform yet.
- AI usage classification by deployment mode
- PDF annex and methodology statement
- Customer-specific factor lineage for assurer review
- Confidence bands and limitations documented
Operate continuously
Recurring reporting for organisations with review history.
From AU$4,800/yr
scales by scope, integrations and review needs
Best for: Organisations that need repeatable reporting across periods and teams.
- Reporting workspace with period history
- Framework-ready exports for local and global reporting
- Controlled imports for AI usage evidence
- Append-only methodology trail
Serve clients
A licensed workflow for consultants and assurance teams.
From AU$12,000/yr
priced by active client portfolio
Best for: Sustainability practices that need repeatable, co-branded client annexes.
- Client workspaces with separated evidence trails
- Co-branded annexes issued under your practice
- Reusable review workflow across reporting periods
- Commercial use rights for client delivery
Embed or license
For platforms that need licensed AI-impact calculations.
From AU$15,000/yr
commercial terms quoted after scope review
Best for: Carbon platforms, data products and enterprise teams embedding AI-impact outputs.
- Licensed calculation outputs for your product or workflow
- Usage-based access controls and audit logs
- Methodology version notices for downstream customers
- Custom commercial and redistribution terms
Prices are starting points in Australian dollars. We do not publish volume thresholds, discount schedules, add-on rates or redistribution terms because those depend on scope and licence rights. Joining the waitlist creates no obligation and reserves no charge.
Every paid option is built on the same engine
- Physically-derived factors, never spend-based (EEIO) proxies
- Append-only methodology audit trail: the assurance premium
- Confidence bands shown on every figure, never hidden
- AI water reporting mapped to GRI 303
- Customer-specific factor lineage available in licensed reports
- Public methodology summary with proprietary calibration kept confidential
Isn't AI usage too small to bother reporting?
Materiality is not the point: defensibility is. Once AI appears anywhere in your operations, an assurer will ask how the figure was derived. A spend-based (EEIO) estimate overstates AI emissions by 10 to 40 times and cannot be defended line by line.
These tiers give you a physically-derived figure with visible confidence bands and an append-only methodology trail, so the number survives review whether it turns out large or small. A small, defensible line item beats a large, indefensible one.
Join the Q4 2026 waitlist
Why Physically-Derived Beats Spend-Based
| Spend-based (EEIO) | Physically-derived (Ours) |
|---|---|
| ~1–2 kg CO₂/$ spent | 0.367 kg CO₂/kWh actual |
| Used as proxy across all software | Anchored to first-party operator disclosures |
| Overestimates AI by ~20–40× | Order-of-magnitude guide |
Result: Your assurer can defend the numbers.
If you operate Australian data centre facilities, you're managing both sides of the equation: your facility's energy, water, and emissions (supply side), and your customers' AI usage impact (demand side).
The Data Centre Register is purpose-built for your reporting: publish your PUE, metered energy, water use, and grid location once, then generate NGER-aligned ESG reports for each reporting cycle. Connect it with this calculator to show your customers their AI usage impact on your facility.
Complementary approach: Use dcreg.online for facility-level NGER Scope 2 reporting, and aixhuman for your customers' Scope 3 (purchased AI services) annex.
Questions?
See our Methodology page for sources and assumptions.
Or email us to discuss your organisation's reporting timeline.
Every figure is an estimate based on available disclosures and modelled assumptions, not a metered actual, nor an assurance opinion. See the methodology page for sources and limitations.