Manager ou dirigeant, 1 an

Name: Manager or executive, 1 year
Creator: Projet Celsius
License: https://creativecommons.org/licenses/by/4.0/

Projet Celsius

Database entryUsesLevel 4 of 6 occupation or activityOccupation or activity, level 4 of 6Used to calculate 9 entries, levels 5 and 6Calculated with 11 entries, levels 1 and 3Database structure

Manager or executive, 1 year

One year of work for a manager or executive who uses AI: about 700 emails, 33 documents of 30 pages summarised, 9.2 presentations of 10 slides, and 7 other tasks. Projet Celsius estimate.

Text and conversation account for 100% of emissions.
About 1,200 requests to AI over the year, or 5.4 per working day.
1.6 times lower than the AI footprint of an office employee, for the same use.
An estimate: the frequencies and the share done with AI are our assumptions.

Compare

Calculated with

Used to calculate

Breakdown of emissions by item

370 gCO2e in total: 810 Wh of electricity × 458 gCO2e/kWh (servers worldwide).

Inputs of the calculation 8

PUE of the data centre, Read 1,000 tokens, Write, with reasoning, Write 1,000 tokens, Generate an image, Write an email, Summarise 30 pages, Presentation of 10 slides

How the calculation is built

One year of work for a manager or executive who uses AI, task by task: emails, reports to read, decisions to prepare, presentations, meetings. The frequencies come from the content of the occupation, the use of AI from usage surveys (OpenAI, Microsoft, Crédoc); these are assumptions, adjustable.

Comparison with other occupations and activities

See the full list

Secondary school pupil, 1 year150 gCO2e

Field-based occupation, 1 year220 gCO2e

Teacher, 1 year260 gCO2e

Human resources, 1 year330 gCO2e

Free personal use, 1 year340 gCO2e

Manager or executive, 1 yearthis entry370 gCO2e

Accountant, 1 year380 gCO2e

Student, 1 year490 gCO2e

Lawyer, 1 year520 gCO2e

Office worker, 1 year590 gCO2e

Logarithmic axis: each tick is 10 times the previous one.

Characteristics

Unit	1 person, 1 year : 220 days
Default location	world, 458 gCO2e/kWh (Ember 2025, life cycle); 11 other locations to choose from
Counted	the tasks done with AI over a working year, including data centre cooling
Not counted	server manufacturing, model training, the person's device and the tasks done without AI
Status	estimated

All the characteristics identifier, version, rule for the value, sources, licence

Identifier	IA-0304, projetcelsius.com/en/ai/database/IA-0304/
Version	0.1 of 2 November 2026
Category	Occupations and activities, one person, one year
Choice of the selected value	database calculation from the entries this one is made of, at the quantities of the unit
Sources	calculation of the database
Licence	CC BY 4.0

Emissions by server location

Emissions associated with electricity (CO2e)

Poland480 gCO2e
World370 gCO2e
United States310 gCO2e
European Union170 gCO2e
France34 gCO2e

× 14 between Poland and France, for the same electricity

Ember 2025, over the life cycle. The same use consumes 810 Wh in each location.

Manager or executive, 1 year · result per person per year · estimated · typical use · reference year: 220 days · one server location per row.Counted: the tasks done with AI over a working year, including data centre cooling. Out of scope: server manufacturing, model training, the person's device and the tasks done without AIFind this case: projetcelsius.com/en/ai/database/activities/manager-or-executive-1-year/Projet Celsius calculation, AI Impact Database, version 0.4 of 13 November 2026 · IA-0304, entry in version 0.1 · engine 0.3.1 · CC BY 4.0.

Usage parameters

AI use

Advanced settings tooling, model class, reasoning, image size, calibration, duration

ToolingModel classReasoningImage sizeCalibrationDuration

Tasks and frequencies

Frequency and share done with AI, task by task · typical use

Adjust each task 12 tasks, frequency and share done with AI

Task	Frequency	Share with AI	Per year	Electricity	Share
Write an email	per day	8 %	704	160 Wh	20 %
Quick question	per day	25 %	275	83 Wh	10 %
Proofread a page	per day	21 %	92.4	55 Wh	7 %
Summarise 30 pages	per week	25 %	33	160 Wh	20 %
Conversation of 10 exchanges	per week	16 %	21.1	74 Wh	9 %
Write meeting minutes	per week	25 %	33	47 Wh	6 %
Write a page	per week	16 %	14.1	6.4 Wh	1 %
Analyse a spreadsheet	per week	12 %	10.6	29 Wh	4 %
Presentation of 10 slides	a year	21 %	9.24	110 Wh	13 %
Deep research	a year	8 %	0.88	87 Wh	11 %
Summarise a meetingdoes not apply with this tooling	per week	0 %
Browser agentdoes not apply with this tooling	per week	0 %

Result for the chosen case

Location of the servers

Quantity and duration

Days

220 days, servers worldwide: 370 gCO2e, 810 Wh.

For a carbon footprint

The row for a carbon footprint

Item	purchases of services (GHG Protocol scope 3, category 1); regulatory BEGES, category 4
Quantity	1 role over one year
Factor	370 gCO2e per item over one year, IA-0304 version 0.1, according to the local grid (location-based), servers worldwide
Total	370 gCO2e · not counted: server manufacturing, model training, the person's device and the tasks done without AI

Frequency and emissions of each task of the occupation "Manager or executive"

Reading: "Write an email", the most frequent task, is also the one that weighs most over a year. Carbon footprint (CO2e), typical use. Each dot is a task: the most frequent on the right, the heaviest each time at the top; diagonals link what weighs the same over a year. In grey, the tasks of the 19 other occupations, at their default settings. Servers worldwide. Calculation from the Projet Celsius AI Impact Database.

Typical usage profile

Projet Celsius assumption

The typical manager gives AI 16 of their 200 emails, 6 questions and 2 texts to rework; has it summarise one report in 4 and prepares a decision or an interview with it every 2 weeks. About 27 requests a week. Over the year, about 810 Wh. Frugal, 270 Wh; intensive, 2.9 kWh. With the office suite, their 7 meetings summarised each week outweigh everything else and the year rises to 2.6 kWh.

Breakdown by item

Quantity, unit value, contribution and nature of the data for each item

Item	Quantity	Unit value	Contribution	Share	Nature of the data
Write an email	700 times	0.23 Wh	160 Wh	20 %	assumption · Microsoft Research
Summarise 30 pages	33 times	4.9 Wh	160 Wh	20 %	assumption · O*NET
Presentation of 10 slides	9.2 times	11 Wh	110 Wh	13 %	assumption · Microsoft Research
Deep research	0.88 times	99 Wh	87 Wh	11 %	assumption · Microsoft Research
Quick question	280 times	0.3 Wh	83 Wh	10 %	assumption · Crédoc
Conversation of 10 exchanges	21 times	3.5 Wh	74 Wh	9 %	assumption · Microsoft Research
Proofread a page	92 times	0.6 Wh	55 Wh	7 %	assumption · Microsoft Research
Write meeting minutes	33 times	1.4 Wh	47 Wh	6 %	assumption · Microsoft Research
Analyse a spreadsheet	11 times	2.8 Wh	29 Wh	4 %	assumption · Microsoft Research
Write a page	14 times	0.45 Wh	6.4 Wh	1 %	assumption · Microsoft Research

Entries that use it (9)

Very small business (5), SME of 50 people, Mid-cap company of 500 people, Large company (5,000), France, inhabitants, Office workers, Public employees, Île-de-France, inhabitants and

Sources of the calculation

17 publications, for this entry and those used to calculate it, with the chosen parameters.

The result is calculated by Projet Celsius. The sources below support its uses, its factors and its conversion; they do not all measure this total.

Uses and assumptions

These publications shed light on AI tasks and uses. They do not measure the carbon footprint of the case calculated.

Anthropic · June 2025
How we built our multi-agent research system ↗ (new tab)
Re-read from cache
Role and limits : Anthropic
What the publication covers. an agent uses about 4 times the tokens of a conversation, a multi-agent system about 15 times
What it does not cover. electricity
Reading by the database. Publisher orders of magnitude, with no absolute value
- Deep research → : Re-read from cache →
  The context re-read from cache at each of the 40 steps, 40,000 tokens on average
  Scope of the figure used. server, convention C4
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
Apec · June 2025
Les cadres et l'IA ↗ (new tab)
Manager or executive, 1 year
Role and limits : Apec
What the publication covers. frequency of generative AI use at work, 2,000 private-sector managers surveyed in March 2025
What it does not cover. non-managers and the number of requests
Reading by the database. Recent French survey on frequency of use at work; copy read on a specialist press site
- Manager or executive, 1 year →
  Support for the "Typical use" setting
  Scope of the figure used. The median user among those who use AI, end of 2026: they turn to it every working day for part of their questions, writing and reading, about 5 to 15 requests per day in an office job. In an organisation or a territory, it also sets the share of employees who use it (25%).
  Limit. This setting remains an assumption of the database, not a measurement of the occupation.
Crédoc · 2026
Baromètre du numérique, édition 2026 ↗ (new tab)
Quick question
Role and limits : Crédoc
Author. Crédoc for Arcep, Arcom, the CGE and the ANCT
What the publication covers. use of generative AI by people aged 12 and over in France in 2025: who, for what, how often, paid or not
What it does not cover. the number of requests
Reading by the database. The reference survey in France, 4,145 people
- Manager or executive, 1 year → : Quick question →
  About 5 questions per day, assumption; share done with AI based on the Crédoc Baromètre du numérique (2026 edition, 2025 data), adjusted to the occupation.
  Scope of the figure used. computation of the response, data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
EnronSent
EnronSent Corpus, notice ↗ (new tab)
Write, with reasoning · Read 1,000 tokens
Role and limits : EnronSent
Author. Will Styler
What the publication covers. 96,106 messages sent by Enron employees, 13,810,266 words
What it does not cover. the messages received
Reading by the database. Professional emails from 2000 and 2001: an order of magnitude, not a norm
- Write an email → : Write, with reasoning →
  The email written, 150 words
  Scope of the figure used. server, excluding the data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
- Write an email → : Read 1,000 tokens →
  Length convention: an email
  Scope of the figure used. A typical email: the sent messages of the Enron corpus average about 144 words; figure read in an excerpt, notice not verified
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.
- Write an email → : Write, with reasoning →
  Length convention: an email
  Scope of the figure used. A typical email: the sent messages of the Enron corpus average about 144 words; figure read in an excerpt, notice not verified
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.
Epoch AI · February 2025
How much energy does ChatGPT use? ↗ (new tab)
Quick question · Re-read from cache
Uses and assumptions · Energy and factors
Role and limits : Epoch AI
What the publication covers. the calculation from the hardware and the assumed size of the model
What it does not cover. actual measurements
Reading by the database. A public, transparent calculation
- Quick question →
  Reference of the selected value
  Scope of the figure used. GPT-4o, 500 tokens written, full scope
  Limit. A public calculation, from the hardware and the assumed size of the model
- Conversation of 10 exchanges → : Re-read from cache →
  The context re-read from cache at each turn: the previous exchanges, and the document already in the conversation if there is one
  Scope of the figure used. server, convention C4
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.

The 5 other publications in this section

Google · 2026
Gemini Deep Research ↗ (new tab)
Read 1,000 tokens
Role and limits
What the publication covers. up to hundreds of sites browsed and pages processed per search
What it does not cover. electricity
Reading by the database. A presentation page: a ceiling, not an average
- Deep research → : Read 1,000 tokens →
  The web pages read, about a hundred by default: Google announces up to hundreds of sites
  Scope of the figure used. server, excluding the data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
Microsoft Research · July 2025
Working with AI: Measuring the Applicability of Generative AI to Occupations ↗ (new tab)
Manager or executive, 1 year · Write an email · Proofread a page · and 6 other entries
Role and limits
Author. Kiran Tomlinson, Sonia Jaffe, Will Wang, Scott Counts, Siddharth Suri (Microsoft)
What the publication covers. 200,000 conversations with Bing Copilot matched to O*NET activities; applicability score by occupation, in open data (version 7 of October 2026)
What it does not cover. actual adoption by occupation
Reading by the database. Measures what AI is used for and succeeds at, not who uses it
- Manager or executive, 1 year →
  Published observation on uses
  Scope of the figure used. Generative AI applicability score for executives (code 11-1011), 200,000 conversations with Bing Copilot, from 0 to 1
  Limit. Used for the occupation coefficient: square root of the score divided by 0.25, the score of office employees
- Manager or executive, 1 year →
  Published observation on uses
  Scope of the figure used. Generative AI applicability score for general and operations managers (code 11-1021), 200,000 conversations with Bing Copilot, from 0 to 1
  Limit. Used for the occupation coefficient: square root of the score divided by 0.25, the score of office employees
- Manager or executive, 1 year → : Write an email →
  About 40 emails per day, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
  Scope of the figure used. tokens read and written, data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
- Manager or executive, 1 year → : Proofread a page →
  About 2 texts to rework per day, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
  Scope of the figure used. tokens read and written, data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
- Manager or executive, 1 year → : Conversation of 10 exchanges →
  About 3 decisions or interviews to prepare per week, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
  Scope of the figure used. messages, context re-read from cache, responses, data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
- Manager or executive, 1 year → : Write meeting minutes →
  About 3 sets of minutes per week, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
  Scope of the figure used. simple tasks, data centre included
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
- Manager or executive, 1 year → : Write a page →
  About 2 pages per week (note, strategy), assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
  Scope of the figure used. tokens read and written, data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
- Manager or executive, 1 year → : Analyse a spreadsheet →
  About 2 dashboards to analyse per week, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
  Scope of the figure used. tokens read and written, data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
- Manager or executive, 1 year → : 10-slide presentation →
  About 1 presentation per week, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
  Scope of the figure used. simple tasks, data centre included
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
- Manager or executive, 1 year → : Deep research →
  About 1 in-depth question per month (market, competitor), assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
  Scope of the figure used. searches, pages read, context re-read, report, data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
O*NET · 2025
O*NET 30.1 Database, Task Statements and Task Ratings ↗ (new tab)
Summarise 30 pages
Role and limits
Author. National Center for O*NET Development, US Department of Labor
What the publication covers. tasks of each occupation and frequency reported by job holders, in 7 classes from "once a year or less" to "every hour or more"
What it does not cover. occupations without a frequency survey, such as data scientists
Reading by the database. The reference classification of tasks; US job holders
- Manager or executive, 1 year → : Summarise 30 pages →
  About 3 reports to read per week, reviewing results being weekly or more often in O*NET; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
  Scope of the figure used. document read, summary written, data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
OpenAI · September 2025
How People Use ChatGPT (NBER Working Paper 34255) ↗ (new tab)
Manager or executive, 1 year
Role and limits
Author. Aaron Chatterji, Tom Cunningham, David Deming, Zoë Hitzig, Christopher Ong, Carl Shan, Kevin Wadman (OpenAI, Duke, Harvard)
What the publication covers. exact message volumes of the consumer plans; classification of a sample of about 1.1 million conversations by topic, intent and link to work
What it does not cover. enterprise plans, under-18s, logged-out users
Reading by the database. The only published measurement of the volumes and uses of a major assistant; internal data, automatic classification
- Manager or executive, 1 year →
  Published observation on uses
  Scope of the figure used. Share of writing in the work messages of management and leadership occupations
  Limit. Supporting evidence for the share of emails and texts in the manager's task basket
Projet Celsius · November 2026
Tokens per word, per line of code and per cell, measured with 7 public tokenisers
Read 1,000 tokens · Write, with reasoning · Re-read from cache · and 1 other entry
Role and limits
What the publication covers. 479,888 words from 196 blog articles and their English translation, code from the site and the Python library, a table of measurements; tokenisers of GPT-4o, GPT-4, Llama 3, Gemini (via Gemma 3), Mistral, DeepSeek and Qwen
What it does not cover. the Claude tokeniser, not public
Reading by the database. A measurement on technical texts, figures included; to be repeated if publishers change their tokeniser
Projet Celsius measurement, repeated on request with the public tokenisers
- Proofread a page → : Read 1,000 tokens →
  Length convention: one page
  Scope of the figure used. A report page, headings and paragraphs included: between the 1,500-character sheet (250 words) and a full A4 page in single spacing (about 550 words, at 6.1 characters per word measured on our articles)
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.
- Proofread a page → : Write, with reasoning →
  Length convention: one page
  Scope of the figure used. A report page, headings and paragraphs included: between the 1,500-character sheet (250 words) and a full A4 page in single spacing (about 550 words, at 6.1 characters per word measured on our articles)
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.
- Summarise 30 pages → : Read 1,000 tokens →
  Length convention: one page
  Scope of the figure used. A report page, headings and paragraphs included: between the 1,500-character sheet (250 words) and a full A4 page in single spacing (about 550 words, at 6.1 characters per word measured on our articles)
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.
- Summarise 30 pages → : Write, with reasoning →
  Length convention: one page
  Scope of the figure used. A report page, headings and paragraphs included: between the 1,500-character sheet (250 words) and a full A4 page in single spacing (about 550 words, at 6.1 characters per word measured on our articles)
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.
- Conversation of 10 exchanges → : Re-read from cache →
  Length convention: one page
  Scope of the figure used. A report page, headings and paragraphs included: between the 1,500-character sheet (250 words) and a full A4 page in single spacing (about 550 words, at 6.1 characters per word measured on our articles)
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.
- Write a page → : Write, with reasoning →
  Length convention: one page
  Scope of the figure used. A report page, headings and paragraphs included: between the 1,500-character sheet (250 words) and a full A4 page in single spacing (about 550 words, at 6.1 characters per word measured on our articles)
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.
- Analyse a spreadsheet → : Read 1,000 tokens →
  The spreadsheet, 200 rows and 10 columns, and the question
  Scope of the figure used. server, excluding the data centre
  Limit. The quantity and the share entrusted to AI are database assumptions, not a measured footprint.
- Analyse a spreadsheet → : Read 1,000 tokens →
  Length convention: a spreadsheet of 200 rows
  Scope of the figure used. A table of 200 rows and 10 columns of numbers and short labels: 4.7 tokens per cell, measured on a real table
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.
- Deep research → : Write 1,000 tokens →
  Length convention: one page
  Scope of the figure used. A report page, headings and paragraphs included: between the 1,500-character sheet (250 words) and a full A4 page in single spacing (about 550 words, at 6.1 characters per word measured on our articles)
  Limit. This publication supports the quantity of text or content processed, not its carbon footprint.

Energy and factors

The references of the selected values and of the entries used in this calculation.

Google · Aug. 2025
Measuring the environmental impact of delivering AI at Google scale ↗ (new tab)
Quick question
Role and limits : Google
What the publication covers. chips, host, standby machines, data centre, on-site water
What it does not cover. manufacturing, training
Reading by the database. The only production measurement published by a provider
- Quick question →
  Reference of the selected value
  Scope of the figure used. Median of a text prompt; chips, host, standby machines, data centre
  Limit. The only production measurement published by a provider, not audited by a third party
Google · January 2009
Powering a Google search ↗ (new tab)
Web search (benchmark)
Role and limits : Google
What the publication covers. a search, data centre
Reading by the database. Never remeasured
- Web search (reference) →
  Reference of the selected value
  Scope of the figure used. A search, data centre included
  Limit. Never publicly remeasured; Google's AI answers are not published
Luccioni et al. · October 2024
Power Hungry Processing ↗ (new tab)
Generate an image
Role and limits : Luccioni et al.
Author. Luccioni, Jernite and Strubell
What the publication covers. GPU, processor, memory
What it does not cover. data centre
Reading by the database. Open models only
- Create an image → : Generate an image →
  One image, of the chosen size: standard by default
  Scope of the figure used. GPU, processor, memory; excluding the data centre
  Limit. This figure covers part of the calculation; it does not measure the whole entry.
- Generate an image →
  Reference for the variant "Standard image, from 512 to 1,024 pixels per side"
  Scope of the figure used. Average of the open models tested, 1,000 images; median 1.35 Wh, maximum 11.49 Wh
  Limit. Open models only; closed generators publish nothing
Microsoft Research · 2026
Estimation de l'énergie des requêtes en production (Joule) ↗ (new tab)
Quick question
Role and limits : Microsoft Research
Author. Microsoft Research, Oviedo et al.
What the publication covers. H100 nodes, production
Reading by the database. Calibrated on production
- Quick question →
  Reference of the selected value
  Scope of the figure used. Median, 500 tokens read and 300 written, H100 nodes
  Limit. Not a measurement, but calibrated on production; judges public estimates to be 4 to 20 times too high
OpenAI · June 2025
The Gentle Singularity ↗ (new tab)
Quick question
Role and limits : OpenAI
Author. Sam Altman, OpenAI
What the publication covers. an average query
What it does not cover. method, scope
Reading by the database. One sentence, no method
- Quick question →
  Reference of the selected value
  Scope of the figure used. "An average query", with no published method
  Limit. No published scope; consistent with Google

The other publication in this section

Watershed · Aug. 2026
Estimating GHG Emissions from AI Use: Framework for Corporate-Level Measurement ↗ (new tab)
Read 1,000 tokens · Write 1,000 tokens · Data centre PUE
Role and limits
Author. Bistline, Ulissi, Davis, Glidden, Joyce, Li, Mohsenin and Suh (Watershed)
What the publication covers. factors per 1,000 tokens read and written at the server before PUE (host and standby machines included), PUE separate (1.10 for a large operator, 1.56 for an unknown operator)
What it does not cover. clean measurements, the model and the hardware
Reading by the database. Default factors proposed for review (public data as of June 2026); for a large operator, taken from Google's measurement published in 2025 (Elsworth et al.), which already contains the data centre overhead; constant per token, they underestimate long contexts and agents according to the authors
- Read 1,000 tokens →
  Reference for the variant "Medium model, Watershed default factor"
  Scope of the figure used. Default factor for 1,000 tokens read, medium model, large operator, at the server before PUE (Watershed table 2, public data as of June 2026)
  Limit. A default value that Watershed derives from Google's measurement (0.24 Wh for the median text query of the Gemini app), with about 500 tokens per query and a written token costing 3 times a read token; Watershed does not give the split between read and written tokens. The 0.24 Wh already include the overhead of Google's data centre (0.02 Wh, 8%): with 375 tokens read and 125 written, the simplest reading, the factor contains it, and the PUE of 1.10 applied afterwards counts it a second time (about 9%). Values proposed for review, final at the next version of the framework
- Write 1,000 tokens →
  Reference for the variant "Medium model, Watershed default factor"
  Scope of the figure used. Default factor for 1,000 tokens written, medium model, large operator, at the server before PUE (Watershed table 2, public data as of June 2026)
  Limit. 3 times the reading factor, based on the ratio between writing and reading selected by Watershed (Patel et al., Splitwise, 2024). Same derivation and same caveat as for reading: Google's 0.24 Wh already include about 8% of data centre. Values proposed for review, final at the next version of the framework
- Write an email → : Data centre PUE →
  The data centre, convention C3
  Scope of the figure used. building cooling and losses
  Limit. This figure covers part of the calculation; it does not measure the whole entry.
- Data centre PUE →
  Reference of the selected value
  Scope of the figure used. Default factor for a large operator, from the server to the data centre; 1.56 for an unknown operator
  Limit. A default value, not a measurement
- Quick question → : Data centre PUE →
  The data centre, convention C3
  Scope of the figure used. building cooling and losses
  Limit. This figure covers part of the calculation; it does not measure the whole entry.
- Proofread a page → : Data centre PUE →
  The data centre, convention C3
  Scope of the figure used. building cooling and losses
  Limit. This figure covers part of the calculation; it does not measure the whole entry.
- Summarise 30 pages → : Data centre PUE →
  The data centre, convention C3
  Scope of the figure used. building cooling and losses
  Limit. This figure covers part of the calculation; it does not measure the whole entry.
- Conversation of 10 exchanges → : Data centre PUE →
  The data centre, convention C3
  Scope of the figure used. building cooling and losses
  Limit. This figure covers part of the calculation; it does not measure the whole entry.
- Write a page → : Data centre PUE →
  The data centre, convention C3
  Scope of the figure used. building cooling and losses
  Limit. This figure covers part of the calculation; it does not measure the whole entry.
- Analyse a spreadsheet → : Data centre PUE →
  The data centre, convention C3
  Scope of the figure used. building cooling and losses
  Limit. This figure covers part of the calculation; it does not measure the whole entry.
- Create an image → : Data centre PUE →
  The data centre, convention C3
  Scope of the figure used. building cooling and losses
  Limit. This figure covers part of the calculation; it does not measure the whole entry.
- Deep research → : Data centre PUE →
  The data centre, convention C3
  Scope of the figure used. building cooling and losses
  Limit. This figure covers part of the calculation; it does not measure the whole entry.

Conversion into carbon

The source of the electricity factor of the selected location.

Worldwide: 458 gCO2e/kWh, life cycle.

Ember · 2026
Global Electricity Review 2026, and Our World in Data based on Ember ↗ (new tab)
Grid electricity
Role and limits : Ember
Author. Ember, Our World in Data
What the publication covers. electricity generation over the life cycle, by country
What it does not cover. guarantees of origin
Reading by the database. The same source for all countries
- Grid electricity →
  Conversion of electricity into carbon: worldwide
  Scope of the figure used. the production of electricity over the life cycle, including power plant construction and fuels
  Limit. An average for the chosen country, not the electricity contract or the data centre of a specific service.

Nature of the data

Published readings, conventions, assumptions

What this figure rests on

Projet Celsius assumptions 100 %

A share counts as an assumption as soon as a quantity set by Projet Celsius (a frequency, a share done with AI, a headcount) enters its calculation, even if the rest rests on published measurements.

Usage assumptions

Frequency, share done with AI according to adoption, source of each assumption

Write an email : 40 per day, 8% with AI in typical use
About 40 emails per day, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
Share done with AI: 3% in light use, 8% typical, 29% intensive · assumption · Microsoft Research · Write an email
Quick question : 5 per day, 25% with AI in typical use
About 5 questions per day, assumption; share done with AI based on the Crédoc Baromètre du numérique (2026 edition, 2025 data), adjusted to the occupation.
Share done with AI: 8% in light use, 25% typical, 86% intensive · assumption · Crédoc · Quick question
Proofread a page : 2 per day, 21% with AI in typical use
About 2 texts to rework per day, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
Share done with AI: 7% in light use, 21% typical, 72% intensive · assumption · Microsoft Research · Proofread a page
Summarise 30 pages : 3 per week, 25% with AI in typical use
About 3 reports to read per week, reviewing results being weekly or more often in O*NET; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
Share done with AI: 8% in light use, 25% typical, 86% intensive · assumption · O*NET · Summarise 30 pages
Conversation of 10 exchanges : 3 per week, 16% with AI in typical use
About 3 decisions or interviews to prepare per week, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
Share done with AI: 5% in light use, 16% typical, 58% intensive · assumption · Microsoft Research · Conversation of 10 exchanges
Write meeting minutes : 3 per week, 25% with AI in typical use
About 3 sets of minutes per week, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
Share done with AI: 8% in light use, 25% typical, 86% intensive · assumption · Microsoft Research · Write meeting minutes
Write a page : 2 per week, 16% with AI in typical use
About 2 pages per week (note, strategy), assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
Share done with AI: 5% in light use, 16% typical, 58% intensive · assumption · Microsoft Research · Write a page
Analyse a spreadsheet : 2 per week, 12% with AI in typical use
About 2 dashboards to analyse per week, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
Share done with AI: 4% in light use, 12% typical, 43% intensive · assumption · Microsoft Research · Analyse a spreadsheet
Presentation of 10 slides : 44 per year, 21% with AI in typical use
About 1 presentation per week, assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
Share done with AI: 7% in light use, 21% typical, 72% intensive · assumption · Microsoft Research · Presentation of 10 slides
Deep research : 11 per year, 8% with AI in typical use
About 1 in-depth question per month (market, competitor), assumption; share done with AI taken from the value common to all occupations for this task, then adjusted to the occupation using the Microsoft study of 200,000 conversations.
Share done with AI: 3% in light use, 8% typical, 29% intensive · assumption · Microsoft Research · Deep research
Summarise a meeting : 2 per week, 70% with AI in typical use · AI in the office suite tooling
About 10 meetings per week, assumption; with AI in the office suite, most are summarised without being asked, assumption; one summarised meeting serves its 5 participants, each carrying a fifth of it.
Share done with AI: 30% in light use, 70% typical, 100% intensive · assumption · Summarise a meeting
Browser agent : 2 per week, 20% with AI in typical use · agents tooling
About 2 online procedures per week, assumption; share delegated to an agent set without a published survey.
Share done with AI: 5% in light use, 20% typical, 60% intensive · assumption · Browser agent

Published figures in support

OpenAI, Chatterji et al. (2025): 52% of work messages
Share of writing in the work messages of management and leadership occupations
Supporting evidence for the share of emails and texts in the manager's task basket · measured · the source ↗
Microsoft Research, Tomlinson et al. (2026): 0.156 applicability score
Generative AI applicability score for executives (code 11-1011), 200,000 conversations with Bing Copilot, from 0 to 1
Used for the occupation coefficient: square root of the score divided by 0.25, the score of office employees · computation · the source ↗
Microsoft Research, Tomlinson et al. (2026): 0.11 applicability score
Generative AI applicability score for general and operations managers (code 11-1021), 200,000 conversations with Bing Copilot, from 0 to 1
Used for the occupation coefficient: square root of the score divided by 0.25, the score of office employees · computation · the source ↗

Methodology conventions

With their alternatives

The full methodology

The working day and year	A working year counts 220 days and 44 weeks, a personal year 365 days; the task lists of occupations give, as a median, about 2 requests to AI per working day at low adoption, 6 at standard adoption, 21 at intensive adoption, excluding code suggestions · alternatives: 365 days for personal use; the profiles of the previous version (5, 20 and 80 requests per day), replaced by task lists
The default location	Without other information, the servers are in the world, at the world average electricity · alternatives: the United States, where most of the large assistants run
Electricity, location-based, on a life cycle basis	The kilowatt-hour is that of the servers' country, on a life cycle basis, from a single source for all countries · alternatives: market-based, with purchase contracts, shown alongside when published
Two significant figures	A calculated value is rounded to two significant figures, a published value is reproduced with the precision of its source; the calculation keeps full precision
The formula	the sum of what makes up the entry, then CO2e: electricity (Wh) multiplied by the carbon intensity of the server location (gCO2e/kWh), divided by 1,000
Calculate a case without an entry	The formula, the default values and a step-by-step example

History

Entry in version 0.1 of 2 November 2026, in the database in version 0.4

The changelog

2 November 2026Version 0.1 of the entry, as of the date of its current value.
31 October 2026Creation of the database, in its first version (0.1).

Citation

Projet Celsius, AI Impact Database, version 0.4 of 13 November 2026, "Manager or executive, 1 year", IA-0304 (entry in version 0.1 of 2 November 2026), projetcelsius.com/en/ai/database/IA-0304/ · CC BY 4.0.

Download

JSON Markdown