AI's Invisible Invoice

Artificial intelligence rarely enters consulting as a revolution. It arrives through ordinary tasks: cleaning an Excel file, summarising a client document, spotting an inconsistency in a slide deck. For Louis, a junior consultant at Wavestone working on a data-office project for Schneider Electric, the value is immediate. Microsoft Copilot, he says, saves him "two or three hours a day" when cleaning data. Clients want faster, more precise, more personalised work, and AI seems to deliver it. Yet one question remains: do the hours saved justify the electricity and water used to run AI?

A productivity calculation with a blind spot
AI accelerates consulting's work of turning information into advice. Louis uses it to clean data, summarise PDFs, improve presentations, draft meeting notes and search Wavestone's internal knowledge base, describing it as "a kind of consultant" supporting his own work. On Schneider Electric's battery-passport project, AI helps harmonise production and transport data across countries, producing faster, more traceable results. But the productivity calculation begins at the keyboard and ends at the deliverable; the infrastructure behind the answer stays outside the frame.

The weightless prompt
That infrastructure is precisely what a desk-bound consultant rarely sees. A prompt is typed, a response appears, and the process feels weightless. Louis admits as much: "At my scale, I do not see the consequences of AI regarding the environment." This is not indifference; it is distance between use and consequence. What he sees is time saved and errors caught. What stays invisible is the electricity powering the model, the water cooling the servers, and the infrastructure quietly being built to sustain it. The IEA (International Energy Agency) reports that electricity use by AI-focused data centres rose 50% in 2025 (IEA, 2026). Teams typically ask whether a tool is accurate and secure, rarely how much energy a task requires, or whether a simpler solution would do just as well.

At my scale, I do not see the consequences of AI regarding the environment.

Two ambitions, one footprint
The tension sharpens inside companies already committed to sustainability. Schneider Electric is strongly associated with environmental responsibility, yet it also wants to be a leading AI adopter, and these two ambitions are not automatically compatible: deploying AI at scale requires data centres and computing power that add to the very footprint the technology is meant to help manage. Consulting firms face the same tension between AI-led competitiveness and their CSR commitments. Louis thinks consultants probably need AI. The question is whether they use it selectively, not by reflex.

A new hierarchy inside the firm
Making such choices also depends on who has time to learn the tools. Louis points to unequal access to AI competence: some employees have time to experiment and build expertise, while others remain fully absorbed by client deadlines. "There is a gap between people who have time to use and test AI and people who do not have time to do so," he explains. This can create a hierarchy favouring employees who already have time to learn AI. Training offered without protected time risks rewarding people who already had the fewest constraints.

There is a gap between people who have time to use and test AI and people who do not have time to do so.

Governance is not responsibility
Wavestone has begun organising seminars on AI for sales, for business and for individual productivity, with partners setting the firm's direction, a necessary first step. But governance and responsibility are not identical. Governance asks who may use which tools and how outputs are validated. Responsibility asks harder questions: was AI necessary for this task, what resources did it require, could a simpler approach have worked, and who is accountable when the benefits are overstated? Wavestone publishes ESG and CSRD reports and is building tools to measure AI's impact, a genuine start. An annual report cannot show whether AI use on a minor task was proportionate. Each project could record why AI is needed, simpler options, and available impact data.

The end of the junior training ground?
Looking five years ahead, Louis expects AI to reduce the number of consultants needed for analytical, repetitive work: one consultant with a powerful tool could do what several once did. This challenges a model built on large teams of juniors, for whom that repetitive work was also a training ground. Human consultants will stay valuable for relationships and judgement, he argues, but "the human touch" cannot become a slogan used to justify shrinking teams. If firms want empathy and judgement from consultants, they must still make room for younger employees to learn those very skills.

When everyone is accountable
Asked who should ultimately be accountable, Louis answers simply, "everyone": the firm, the technology provider, the client, regulators and society all carry part of the responsibility. That shared model is unavoidable, but it carries a risk. When responsibility is distributed this widely, each actor can assume someone else will act first, while the environmental cost keeps quietly accumulating in the meantime.

Measuring what efficiency costs
AI adoption in consulting is already underway; the real question is whether efficiency remains the only measure of value. Speed is not the same as responsibility. Responsible use requires discernment about when AI genuinely helps, measurement that goes beyond broad ESG communication, and accountability that does not quietly dissolve behind the comforting phrase "shared responsibility."

The consultant of the future will not be the one who uses AI for everything, but the one who can explain why a given use is necessary and proportionate, and who is willing to measure what that efficiency actually costs.

IEA (2026) Key Questions on Energy and AI. Paris: IEA. Available at: https://www.iea.org/reports/key-questions-on-energy-and-ai (Accessed: 6 November 2026).

Article written by BERTINOTTI Hugo, CHAPON Romain, CHENNOUFI Riad —
Organization & Management of Consulting Firms

Interviewee: Louis, junior consultant at Wavestone