Accenture and Concentrix Show Who Keeps the AI Saving
On October 1 Accenture, the world's largest consulting and outsourcing firm, reported quarterly revenue above the top of its own forecast range and said it had seen lower pricing in many areas, which it put down to competition. Two days earlier Concentrix, one of the biggest call-center outsourcers, guided next quarter's revenue down 3% to 5% and named faster AI automation as a cause. The two reports point the same way: the saving from AI is starting to reach the client, and a firm that sells hours has to decide what it sells instead.
The assumption that held until this quarter
The position was simple and, until recently, sound. A services firm adopts AI inside its own delivery, does the same job in fewer hours, and keeps the difference as margin while the client keeps paying for the same volume. The client buys a result and never sees the vendor's cost sheet, so the productivity stays with the vendor.
This week's results show that arrangement coming apart at both ends. The volume clients pay for is shrinking as support and back-office tasks get automated, and at least one large firm says it is handing the productivity to clients.
What Accenture and Concentrix said
This week's Newsletter, OpenAI shelved GPT-6.1 Astra as regulators turned to AI agents, led with models and regulation; the evidence on who gets paid for AI came in two earnings reports. Accenture's fourth fiscal quarter brought revenue of $18.7 billion, up 7% in local currency (that is, excluding currency swings), per its October 1 release. On the earnings call the company said pricing had been broadly stable through the year but was lower in many areas in the fourth quarter, in a highly competitive market. Chief executive Julie Sweet said Accenture is handing clients more productivity from AI and making up for it with new kinds of work and larger scope.
Concentrix runs customer support for large companies, much of it by phone and chat. Its third-quarter revenue slipped 0.5% excluding currency swings, and its fourth-quarter guidance calls for a fall of 3% to 5%, per its September 29 filing. It gave the reasons plainly: faster AI automation, and two large cloud-computing clients ending support arrangements by the fourth quarter instead of the second quarter of 2027, as previously expected. Its margins improved anyway, helped by its own automation.
Both firms are passing the AI saving through. Accenture gives it to clients inside the contract and sells new scope to fill the gap. Concentrix bills for volume, so when automation shrinks the volume of support, its revenue shrinks with it, even where Concentrix runs the automation itself.
The strongest case against
The best reading of the same week is that AI is good for services firms. Accenture grew, and its release says more than 400 clients began advanced AI projects with it during the fiscal year. Sweet's stated view is that the opportunity AI creates is larger than the efficiencies it gives away. Concentrix chief executive Chris Caldwell said on the September 29 call that half its revenue now comes from programs created or substantially reworked in the past three years, much of it with AI.
That reading is right about volume, and the evidence supports it for firms that can resell freed hours as new scope. It does not rescue firms whose product is the hours. Two companies and one quarter set the limits here. Accenture blamed competition for its pricing and did not name AI, and Concentrix also cites offshoring, which it says takes about three points off growth. The volume effect is documented; the price effect is one quarter old.
Europe shows the same split. Teleperformance, the Paris-based group that is the world's largest call-center operator, reported first-half revenue down 1.7% on a comparable basis, per its July results, and said its content-moderation business is shrinking with AI automation and moves offshore. Its margin held at 13.6% on cost control and AI efficiencies, the same combination Concentrix reported: profit protected, revenue shrinking.
Who should be uncomfortable
The exposed firms are agencies, outsourcers and professional firms that bill by the hour or by the seat (a fee per person assigned to the account) for drafting, research, reporting and first-line support, the tasks automation is removing fastest. The exposure is sharpest where a few clients carry the revenue, because one client's decision to end or automate a program removes a whole line, as Concentrix found with two of them.
The move is to raise the saving with the client before its purchasing team does. At the next renewal, show how many hours AI removed from the job, price that part lower, and propose what the freed hours should buy instead: the review, the judgment, the scope the client could not afford before. A firm that offers the cut gets to shape what replaces it. A firm that waits gets the cut and nothing in exchange.
The thing to watch is Accenture's next results in December, and whether "lower pricing in many areas" becomes most areas.
For two years the question in services was whether AI would cut costs. This quarter the firms with the best view of their clients' budgets started reporting what happens next. At Concentrix and Teleperformance, the programs clients pay for are getting smaller while margins hold, so the saving shows up as less billable volume. At Accenture, pricing is lower, for reasons it calls competition, while it hands clients more AI productivity. That is the first sign the price itself is moving. What a services business keeps is what it can sell after the hours are gone: the judgment about what to automate, the accountability when it fails, and the scope no one could afford before. Those that price for that can grow the way Accenture did. Those that price for hours will report the way Concentrix did.