TotalEnergies AI Strategy: Copilot for the Office, Data for the Plant

TotalEnergies is building AI from the data up. It gave Microsoft 365 Copilot to about 30,000 employees, is wiring its plants for real-time data so AI has something to work with, and has launched a three-year, €100 million-plus program with Mistral AI to build models for finding and developing oil and gas reservoirs. Outside the company, AI is also a market: TotalEnergies supplies power to data centers.

The CEO points AI at the core business. Announcing the Mistral program in September 2026, Chairman and CEO Patrick Pouyanné said exploration and reservoir engineering are among the areas where “artificial intelligence can create the greatest value for our activities” (company release, September 2026).

AI is also part of the formal strategy. At the 2025 results and 2026 objectives presentation in February 2026, Stéphane Michel, President of Gas, Renewables & Power, presented “how we plan to boost our operations by using AI.” The 2025 Universal Registration Document, published in March 2026, has a section titled “Digital acceleration as a performance lever,” and its strategy overview names digital as a way to lower operating costs.

Four priorities set the focus. TotalEnergies set out its approach in a Summer 2026 shareholder newsletter article, “AI as a Performance Driver,” which lists four priorities:

• Health, safety and environment: improving safety and reducing emissions
• The digital plant: improving the performance and availability of industrial sites
• Geosciences: AI for subsurface understanding, seismic interpretation and reservoir development
• Integrated Power: AI forecasting, especially of weather, and electricity trading

Outside, AI is a growth market. Michel described “how we are going to power the AI revolution with our Integrated Power supply,” and CFO Jean-Pierre Sbraire said the power business “surfed, I would say, on the wave created by datacenters and AI” (February 2026 presentation).

Data comes first. Michel put it plainly: “there is no AI programme if you don’t have a strong data platform” (February 2026 presentation). TotalEnergies plans to multiply by ten the data points it collects across its industrial assets, and it is deploying AspenTech’s Inmation data software on all its sites. Namita Shah, President of OneTech, the company’s engineering and technology unit, said the aim is to connect more than 70% of equipment so there are fewer breakdowns. In September 2025, TotalEnergies agreed to roll out Cognite’s industrial data platform across all its operated upstream assets over three years.

Copilot went wide. After a 300-person pilot in September 2023, TotalEnergies announced in February 2024 that it would deploy Copilot for Microsoft 365 to its employees, with Power Platform licenses so teams could build their own low-code apps. It committed to train every employee on the new tools that year. By October 2025, about 30,000 people had Copilot licenses (Microsoft case study).

Agents pass a gate before they scale. An internal validation committee checks each proposed agent for business value, technical viability, cybersecurity and overlap with existing agents. More than 30 applications were deployed or in development by October 2025, including:

• MARGE: a maintenance agent that analyzes and classifies intervention reports
• BuyerCompanion: a procurement agent for purchases up to €50,000, live in France

Executives are expected to use agents too. In February 2026, Pouyanné said that “each executive of the Company will be trained to have at least one agent with him” the following week. Training leaders to work with agents is one way to close the chat-to-agent gap.

An engineering hub supplies the people. TotalEnergies is building a Global Competence Centre in India, staffed by its own employees, to develop its digital and AI programs. Michel set a target of at least 500 engineers by 2027.

A European partner handles the hardest science. Mistral AI first joined in June 2025 through a joint innovation lab, starting with an assistant for TotalEnergies’ 1,000 researchers. In September 2026, the two agreed a three-year program, with an investment of more than €100 million, to build frontier AI models for reservoir exploration and engineering. Pouyanné said the program reflects a commitment to a “high-performing European digital ecosystem.” Seven months earlier, Shah had said that on the subsurface side, AI was “probably in the very early stages.”

The workforce is stable. Headcount has stayed at about 101,000 to 103,000 since 2022. TotalEnergies has not linked AI to job cuts.

Governance is described in part. Besides the validation committee, a four-person AI Trust team in the group IT department includes an AI compliance unit that promotes compliance with the EU AI Act (job posting, for an October 2026 start). The sources reviewed contain no published AI principles.

AI fluency, coded against the Lead with AI Fluency Matrix (September 2026 edition):

• Fluency floor: Published in part. In 2024, TotalEnergies committed to train every employee on its generative AI tools, and in 2026 it said every executive would be trained to work with an agent. No completion figures have been published.
• Highest level evidenced: Level 3 (Integrating), company-claimed. Agents are built into maintenance and procurement work. The sources reviewed do not show AI running a process end to end, so Level 4 is not evidenced.
• Portfolio design: A broad base plus industrial specialists. Copilot gives about 30,000 employees a general assistant, agents pass through a validation committee, and OneTech, the India centre and partners such as Cognite and Mistral build industrial AI.

TotalEnergies reports adoption and use-case results, not enterprise value. All figures below are company-claimed, from the Microsoft case study (October 2025):

• Usage: Copilot is used weekly by 70% of users and daily by nearly half, and 70% would recommend it to colleagues
• Maintenance: MARGE improved the identification of fault causes by 25% to 45%
• Procurement: BuyerCompanion could save 10% on certain purchases and could eventually support 15,000 staff worldwide

The size of the prize is framed, not targeted. Shah said that “even an increase of between 1 to 2% of production on our assets is, of course, just enormous” (February 2026 presentation). TotalEnergies has not set a target for that or published a total value for AI. The €100 million for Mistral is an investment, not a result.

Quarterly results report AI as a deal, not a result. The third-quarter 2026 release on October 29 mentions AI once, as the Mistral partnership among its upstream highlights. Moving from usage figures to a number the board can track is the step from AI adoption to AI value.

Next markers: first results from the Mistral program, whether the India centre reaches 500 engineers by 2027, and whether the 2026 Universal Registration Document reports AI value.

Change log

• September 2023: 300 employees pilot generative AI tools.
• February 2024: TotalEnergies announces Copilot for Microsoft 365 for employees and commits to train every employee that year.
• June 2025: TotalEnergies and Mistral AI set up a joint innovation lab.
• September 2025: Cognite’s data platform is chosen for all operated upstream assets.
• October 2025: Microsoft reports about 30,000 Copilot licenses, MARGE and BuyerCompanion, and the agent validation committee.
• February 2026: AI gets its own segment at the results and objectives presentation; executives are to be trained to work with agents; the India centre targets 500 engineers by 2027.
• March 2026: The 2025 Universal Registration Document includes a section on digital acceleration.
• September 2026: TotalEnergies and Mistral launch a three-year program worth more than €100 million.
• October 2026: The third-quarter release lists the Mistral partnership as an upstream highlight.

• Disclosure: 3 of 4. TotalEnergies answers where it is going, where it will focus and what it will take. How it will know is answered in part, with usage and use-case results but no enterprise value or target.
• Data first. The AI program rests on connecting plants and multiplying data points, not on tools alone.
• Agents are gated. A validation committee checks value, security and overlap before an agent scales.
• Leaders are trained on agents, and every employee was promised training on the generative AI tools.
• The largest disclosed AI investment targets the core business: more than €100 million for frontier models for reservoirs, with a European partner.
• The say-do gap is measurement. A company investing at this scale reports usage, not value.
• Watch the Mistral program. The test is whether AI changes exploration results, not only office productivity.

What the numbers show: Sales fell 28% from 2022 to 2025 as energy prices came down from their 2022 peak, while headcount held at about 101,500, so revenue per employee fell 29%. The decline reflects prices, not staffing, and TotalEnergies has not linked AI to job cuts. These figures do not show that AI caused any change.

Revenue, headcount and revenue per employee, FY2022 to FY2025

• FY2022: $281.0bn revenue, 101,279 employees, $2.77M per employee
• FY2023: $237.1bn revenue, 102,579 employees, $2.31M per employee
• FY2024: $214.6bn revenue, 102,887 employees, $2.09M per employee
• FY2025: $201.2bn revenue, 101,513 employees, $1.98M per employee

Change, FY2022 to FY2025: revenue down 28%, headcount flat, revenue per employee down 29%.

Notes: Revenue is sales, including excise taxes, as reported. Headcount is employees at year-end. Sales move mainly with oil, gas and electricity prices, and include excise taxes collected on fuel sales.

These figures give context on scale and productivity. They do not show that AI caused any change. Sources: TotalEnergies 20-F filings and Universal Registration Documents.

What is TotalEnergies’ AI strategy?

TotalEnergies is building AI on a data foundation. It is connecting its industrial sites for real-time data, giving about 30,000 employees Microsoft 365 Copilot, approving agents through a validation committee and building industrial AI with partners such as Mistral AI and Cognite. Its four priorities are safety and emissions, plant performance, geosciences, and power forecasting and trading.

‍

What is TotalEnergies’ partnership with Mistral AI?

In September 2026, TotalEnergies and Mistral AI launched a three-year program, with an investment of more than €100 million, to build frontier AI models that help geoscientists explore and develop oil and gas reservoirs. It builds on a joint innovation lab set up in June 2025.

‍

How does TotalEnergies use Microsoft Copilot?

About 30,000 employees had Microsoft 365 Copilot licenses by October 2025, and the company says 70% of them use it weekly. It also builds agents on Microsoft tools, including MARGE for maintenance and BuyerCompanion for procurement.

‍

Is TotalEnergies cutting jobs because of AI?

No cuts linked to AI have been reported. Headcount has stayed at about 101,000 to 103,000 since 2022.

‍

How is this profile put together?

Lead with AI profiles use public sources only, such as annual reports, earnings materials, filings and on-record interviews, dated at the time of writing. AI fluency is coded against the Lead with AI Fluency Matrix. Company figures are marked as company-claimed unless they have been independently verified.