Platform footprint
Our energy and emissions, in the open
Qstudy runs on AI models in other companies’ data centres, and that uses electricity. This page estimates how much, live, for the whole platform, so schools, businesses and partners can see what working with us adds to their own footprint.
It is about Qstudy as a service, never about any one person. We do not track individual footprints and never will.
How we estimate it
- 1
Count what was made
Every 15 minutes we tally what the whole platform made, by day and type: notes, quizzes, images, videos, animations, songs. Failed attempts count too; they still used compute. Volumes stay internal; published totals are rounded to two significant figures.
- 2
Convert to electricity
Each type has a low, central and high energy figure per item from published research (table below), including the provider's data-centre overhead. We add our always-on app servers.
- 3
Convert to CO₂e
Electricity is multiplied by a grid emission factor: the U.S. national average from EPA eGRID2023 for the central estimate, with a range from the Canadian national average up to well above the U.S. one. This is the GHG Protocol's location-based method.
CO₂e = (items made × energy per item + server power × hours) × grid intensity
What this covers
In GHG Protocol terms, everything on this page sits in Scope 3, Category 1 (purchased goods and services): the AI providers and cloud hosting we pay for. For a customer, Qstudy’s share of these figures would fall in their own Scope 3. Offices, travel, commuting and the manufacture of hardware are not included.
What keeps it lower
- Everyday text work (notes, quizzes, flashcards) runs on small, fast models. Large models are kept for the tasks that need them, such as writing animation code.
- Animations render on the app servers we already run, one at a time per machine, with hard time and CPU limits, rather than on extra dedicated hardware.
- What you make is saved. Opening it again later reads the saved copy instead of generating it again.
What we will not do
No personal footprints
We do not, and will not, calculate or show any student's or teacher's carbon footprint. Studying should not come with a guilt meter. Every figure here is a platform total.
No neutrality claims
We do not buy offsets and do not call Qstudy green, eco-friendly, carbon neutral or net zero. We would rather show you the estimate and how we got it.
Standards and North American rules
Method
- GHG Protocol Corporate Standard and Scope 2 Guidance (location-based), with all cloud and AI services reported under Scope 3, Category 1, as the Scope 3 Standard describes.
- Ranges rather than single values, because the providers we buy from publish little energy data.
Canada
- Competition Act: since June 2024, environmental claims about a business must be backed by adequate and proper substantiation, and are judged by their general impression as well as their literal meaning. (The separate "internationally recognized methodology" test was removed in March 2026.) We still use the internationally used GHG Protocol, show our working, and make no claim of environmental benefit.
- Competition Bureau guidance on environmental claims and greenwashing: we avoid vague, unqualified terms and state limitations next to the figures.
- Canadian Sustainability Disclosure Standards (CSDS 1 and 2, voluntary): CSDS 2 measures emissions using the GHG Protocol, the same basis as this page.
- The Canadian grid figure is the national average from Environment and Climate Change Canada's National Inventory Report.
United States
- FTC Green Guides (16 CFR Part 260): no unqualified general environmental claims, no offset claims, and qualifications stated clearly and close to the figures.
- California SB 253 and SB 261 apply to companies with more than US$1 billion and US$500 million in revenue respectively. Qstudy is well below both, but the GHG Protocol basis here is the one SB 253 uses.
- The U.S. grid figure is the national average from the EPA's eGRID2023 (revision 2, June 2025).
Limitations
- These are estimates, not measurements. AI providers do not tell us how much energy each request used or where it ran.
- The least certain figures are AI video and songs: their providers publish nothing, so we use the closest published research as a proxy.
- Not yet counted: coding-tutor chat, live lecture transcription and quiz games. They are text workloads and small next to video, but they make the estimate a slight undercount.
- Not included: offices, business travel, commuting, the manufacture of the hardware we run on, and water use.
- Not independently verified. If we get a third-party review, it will be noted here.
For procurement and sustainability teams
The live figures are available as JSON at /api/public/footprint for your own records. If you need a figure for a specific period, our full method, or answers for a supplier questionnaire, get in touch.
Sources
- Google, Measuring the environmental impact of AI inference (Aug 2025)
- Epoch AI, How much energy does ChatGPT use? (Feb 2025)
- Luccioni, Jernite & Strubell, Power Hungry Processing (ACM FAccT 2024)
- MIT Technology Review, We did the math on AI's energy footprint (May 2025)
- U.S. EPA, eGRID summary data (eGRID2023 rev. 2)
- Environment and Climate Change Canada, National Inventory Report 1990-2023, Part 3, Annex 13 (2025)
- U.S. EPA, Greenhouse gas emissions from a typical passenger vehicle
- Cloud Carbon Footprint, methodology (server power coefficients)
- GHG Protocol standards
- Competition Bureau Canada, Environmental claims and greenwashing (updated 2026)
- U.S. FTC, Green Guides
This statement and its factors were last reviewed on 31 October 2026.