Earlier this year one of the largest accounting firms on earth leaned on its own auditor to cut the bill. According to Financial Times reporting, KPMG pushed Grant Thornton’s UK arm for a discount on its audit fee, arguing that Grant Thornton should pass along the savings AI was handing it, and signaling it would look elsewhere if the number did not come down. Neither firm confirmed the talks, but UK Companies House filings show Grant Thornton’s fee for auditing KPMG International’s accounts fell to $357,000 from $416,000, a cut of about 14 percent. Sit with that picture for a second. When a Big Four firm starts squeezing its own auditor over AI efficiency, the pricing reckoning everyone has been predicting is not on its way. It has arrived.

And it does not stay at the top. The same logic runs all the way down to the smallest practice, because it was never really about KPMG. It is about the unit. For roughly a century, accounting sold time. The hour was the product, the invoice, and the mental model all at once. AI compresses the hours. So a firm that still bills by time now sits in the strange position of watching its own revenue shrink precisely as its work gets faster and better. Efficiency, priced by the hour, is a pay cut you hand yourself.

The arithmetic is unforgiving once you write it down. Take a deliverable that used to run twenty hours and let AI do it in five, at the same quality or better. Bill it by time and you have just cut the revenue on that job by about seventy-five percent for equal or superior output. Multiply that across a book of business and the problem stops being hypothetical.

Clients can run that arithmetic too, because they read the same breathless AI headlines you do. A survey of 258 senior leaders at large consulting, accounting, and law firms, conducted by the training company General Assembly, found that 79 percent of professional services firms say AI is already changing their pricing conversations. Among the accounting firms in it, more than a third report clients openly questioning what they charge, and 73 percent say they have changed how they talk about price. PwC’s AI leader Dan Priest put it plainly to Bloomberg last year: clients hear the firm boast about its AI-driven efficiency and, reasonably, want a share of the savings. The KPMG demand was that same instinct with more leverage behind it.

This is where the obvious story falls apart. If AI compresses hours and clients want the savings, you would expect prices to be in free fall. They are not. A 2025 survey of 219 US accounting firms by the practice-management platform Ignition found that 80 percent planned to raise fees in 2026, most of them by five to ten percent. The AICPA’s 2025 survey of more than a thousand firms describes the same shift from the inside: a steady move toward value and fixed pricing, and a continued decline in traditional hourly billing as the primary way firms charge. PwC’s US chief executive Paul Griggs has said the firm will offer alternatives to the hours-based billing that became the profession’s default, pricing some tax and advisory work around AI tools and a different delivery model rather than around time.

Both things are true at once, and the tension between them is the entire story. The hour as a unit is dying, and firms are charging more. That only makes sense if you stop treating price as a function of time and start treating it as a function of value. The firms raising fees are not billing more hours. They are selling something other than hours.

The answer is the part of the work AI cannot do and would not want its name on. Not the reconciliation, which is close to free now, but the judgment that catches the reconciliation the machine got confidently wrong. Not the tax return the software drafts in seconds, but the strategy behind it, the defensibility, and the signature that says a human stands behind the number. Ash Khanna, who runs professional services at General Assembly, described the shift as leadership moving from overseeing projects to owning high-stakes judgment, the human trust and ethical accountability that, in his words, AI cannot replicate. We have made a version of this argument since our first edition. What is new is that it has become a pricing strategy rather than only a job description. The reviewer who signs the work was always the real product. Now that reviewer is the price.

If that sounds like optional strategy, look hard at the alternative. A firm that keeps billing by the hour while AI eats the hours loses twice over. The revenue on each job falls, and the client, watching the work get faster, feels overcharged and starts asking the KPMG question out loud. You end up cheaper and resented at the same time, which is the worst of both worlds. None of this is unique to accounting. Consulting and law are staring down the identical reckoning, and the largest advertising holding companies have already begun reporting a growing slice of revenue tied to outcomes rather than hours. The billable hour is not being retired by regulators or by fashion. It is being made obsolete by a technology that severs the old link between time spent and value delivered.

None of this is a magic word, and the firms selling value pricing as a slogan are skipping the hard part. Quantifying an outcome a client will actually pay a premium for is genuinely difficult, and some work simply commoditizes. Routine compliance and pure data entry are getting cheaper for a reason, and pretending you can hold 2019 rates for work AI now does in minutes is its own kind of denial. Even inside PwC the message splits: while Priest talks about sharing savings, the firm’s Australian assurance leader Sue Horlin has said flatly that services are not cheaper because the AI itself is expensive to run. Both can be true. Some things fall in price, some rise, and telling them apart is the actual job.

The operational reality is humbling too. In that same General Assembly poll, of large and well-resourced firms, roughly half said they had abandoned an AI project in the past year for lack of skills, and more than three quarters were still leaning on manual checks as their main form of AI governance. Repricing around AI you have not learned to run reliably is just a fresh label on the same service. Only about one accounting firm in ten is genuinely moving toward selling proprietary AI tools as a product clients pay to use. It is a real path, and a demanding one.

So the move is not to panic, and not to paste “value pricing” onto your website and call it done. It is more surgical than either.

Start by splitting your book in two. On one side is the commodity work, the compliance filings, the categorization, the rote data entry, where AI will keep pushing the price down and your only sane play is to automate it hard and compete on efficiency rather than defend an old rate. On the other side is the judgment work, the advisory, the planning, the assurance, the reading of a situation, where the value has nothing to do with hours and everything to do with being right. Price that side by the outcome and the accountability, in fixed fees or subscriptions, and stop quoting time for it.

Then set the transparency frame before your clients set it for you. Tell them how you use AI and why your price reflects judgment and a signature rather than keystrokes, so the efficiency becomes your story instead of their leverage. Reinvest the hours AI frees into the advisory capacity you are now charging for, and into what General Assembly’s respondents called an integrity layer, a human explicitly accountable for the accuracy of what the AI produces. Nearly three quarters of firms in that survey said they intend to staff exactly that role within a year. And do not sell an outcome you cannot yet deliver, because the fastest way to discredit value pricing is to charge for judgment you have not actually built.

AI did not just change how accounting work gets done. It broke the unit that work was sold in. Selling hours in an age of automation means selling a thing that shrinks in your hand, faster every quarter. The firms that will be fine, and plenty will be more than fine, are the ones naming a new unit: the outcome, and the human judgment standing behind it. KPMG understood that logic well enough to use it on its own auditor. The only real question for every other firm is whether it reprices on purpose, or waits for a client to do the math first.

— Footnote

Footnote is an independent publication, with no affiliate links and no vendor paying for placement. It is not professional accounting, tax, legal, or business advice. Survey figures are attributed to General Assembly, Ignition, and the AICPA as noted. The KPMG and Grant Thornton fee negotiation is per Financial Times reporting relayed by Accounting Today and was not confirmed by either firm; the audit-fee figures are from UK Companies House filings. All details are current as of July 2026.