"Where's My Discount?": As AI Speeds Up Legal Work, Clients Are Coming for the Billable Hour

The New York Times reports clients are demanding discounts as AI makes law firms faster — while Goldman Sachs, Morgan Stanley, and Citigroup are already squeezing their outside lawyers to share the savings.

Published September 29, 2026 · By Daniel Rao

The U.S. Supreme Court building in Washington, D.C. Big Law's billable-hour model faces its biggest challenge yet as artificial intelligence compresses the hours behind every matter. Photo: Carol M. Highsmith / Library of Congress.

The billable hour survived the fax machine, the BlackBerry, and the cloud. It may not survive the large language model. The New York Times reported Saturday that corporate clients are increasingly asking their law firms a question that strikes at the foundation of the legal industry's economics: as artificial intelligence makes legal work faster and cheaper to produce, where is the client's discount?

The Times' reporting lands on top of an aggressive campaign that began earlier this month, when the Financial Times revealed that three of Wall Street's biggest banks — Goldman Sachs, Morgan Stanley, and Citigroup — are pressing their elite outside lawyers to cut costs, arguing that AI is sharply reducing the hours behind routine legal work like research, document review, and contract analysis. The banks are not asking politely. They are demanding breakdowns of AI savings, moving work to competitive bidding, and shifting toward fixed fees. And they are some of Big Law's most lucrative customers.

What is unfolding is not a negotiation over a few percentage points. It is a fight over who captures the productivity gains of AI — the firms that deploy it, or the clients who pay for the work it accelerates. For a century, the answer in American law was settled: the firms kept the gains, because clients bought hours and firms controlled how many hours a matter took. AI is breaking that settlement, and both sides know it.

The revolt starts with the banks

Wall Street's banks are the ideal spearhead for this fight. They are among the largest consumers of outside legal services in the world, spending on everything from financing transactions and mergers to litigation and white-collar investigations. When they move in concert, the market notices.

"If the number of hours they're working on a matter has come down because of AI… our expectation is for costs to come down significantly per transaction," Adam Meshel, Citigroup's global head of legal, told the Financial Times. Citi has begun asking law firms competing for its business to spell out exactly how much they are saving by using AI — a demand that turns the firms' own efficiency marketing against them. Every press release about an AI-powered document-review platform is now, in effect, a confession that the old price was too high.

Morgan Stanley is going further. Eric Grossman, the bank's general counsel, told the FT that most of the bank's outside legal work will be put out for competitive bidding by the end of this year, with alternative arrangements such as fixed fees replacing hourly billing. His diagnosis of the industry was unsparing: top lawyers have "for a long time been compensated on the foundation of [associates billing for long hours]," he said. "Their compensation model is now extraordinarily unstable." The technology's ability to compress legal work, he added, could amount to "a fundamental altering of the revenue foundation for these mega firms."

Goldman Sachs is pressing the same case. The message from all three banks is identical in substance: we can see the productivity gains, we are sophisticated buyers, and we intend to be paid our share.

The numbers behind the anger

The banks' demands arrive at a moment of unusual pricing tension. Legal bills from the nation's largest firms have soared in recent years. According to Persuit data reported by the FT, Big Law associate rates have climbed 33% since 2023 to an average of $798 an hour this year, while partner rates are up 29% over the same period. Clients were already grumbling about the trajectory. AI handed them the argument they were missing.

The efficiency data, meanwhile, is becoming hard for firms to wave away. Artificial Lawyer's analysis of BigHand survey data this month found that 31% of law firms now say AI has improved their profitability, 27% report reductions in billable hours for certain tasks, and 31% report reduced reliance on manual tasks. Separately, nearly half of the large firms surveyed this year by Citi's own law firms group said AI had already affected how they price their work. The disruption to pricing is still limited — but the direction of travel is not in dispute, even inside the firms.

Related: OpenAI Launches Astra for Law — our report on the AI legal tools now sitting on both sides of this pricing fight.

"Where's my discount?"

The New York Times' Saturday piece crystallized the client-side mood into a single question. As AI makes firms more efficient, clients are asking what the benefit means for them — and firms, the Times found, are not ready to blow up their trusted business model. Ropes & Gray vice chairman Neill Jakobe described clients recognizing that AI can make work more efficient and pressing the pricing question directly.

The firms have counterarguments, and they are not frivolous. AI tools may not yet save as much time in practice as vendors promise; the hours saved on research and review can be offset by lawyers spending more time on higher-value analysis; and measuring the precise AI-attributable savings on any given matter is genuinely difficult. As one line of industry analysis puts it, announcing efficiency can sharpen the pricing conversation rather than settle it — because efficiency you cannot measure is efficiency you cannot invoice, or discount, with confidence.

But the firms' discomfort is telling. The same BigHand data found that 35% of partners remain "uncomfortable" discussing AI use with their clients. That is an extraordinary admission: more than a third of the people who own these businesses do not want to talk to customers about the technology reshaping their product. The likely reason is not hard to find. Every candid conversation about AI-driven efficiency is a conversation that ends with the client asking for money back.

"Where's My Discount?" FAQ

Which banks are pressuring Big Law over AI pricing? Goldman Sachs, Morgan Stanley, and Citigroup, per a Financial Times report from early September 2026. Citigroup global head of legal Adam Meshel is asking firms competing for Citi's business to spell out exactly how much they save with AI, while Morgan Stanley general counsel Eric Grossman plans to put most of the bank's outside legal work out for competitive bidding by the end of 2026, with fixed fees replacing hourly billing.

How much have Big Law rates risen? Per Persuit data reported by the FT, associate rates are up 33% since 2023 to an average of $798 an hour, and partner rates are up 29% — rising bills that set up the clash with AI-driven client demands for discounts.

What do the efficiency surveys show? Artificial Lawyer's analysis of BigHand survey data found 31% of law firms say AI improved profitability, 27% report billable-hour reductions on some tasks — and 35% of partners are "uncomfortable" discussing AI use with clients. Separately, nearly half the large firms surveyed by Citi's law firms group said AI had already affected how they price work.

What is the firms' counterargument? That AI tools may not save as much time in practice as vendors promise; that hours saved on research can be offset by more time on higher-value analysis; and that precisely measuring AI-attributable savings on any given matter is genuinely difficult.

Our analysis: from hours to throughput

What follows is analysis. The deeper story here is not about discounts; it is about what replaces the billable hour when hours stop being the scarce resource. Grossman's "fundamental altering of the revenue foundation" points at the answer: a shift from selling time to selling throughput.

Consider the economics. A firm that bills 20 hours for work AI lets it complete in 5 faces three choices: bill 20 anyway (and lose the client), bill 5 (and lose 75% of the revenue), or handle four times as many matters at the lower price (and keep revenue while transforming its operations). The third option is the one the banks are implicitly offering — Morgan Stanley's fixed-fee, competitive-bidding model works for firms only if they can process dramatically more work per lawyer. The profit driver moves from hours billed to matters completed. That is a different business: more like a factory, less like a guild.

This is why the transition is so painful. The billable hour is not just a pricing mechanism; it is the organizing principle of the modern law firm — how associates are evaluated, how partners are compensated, how prestige is allocated. Fixed fees and throughput pricing reward different things: process engineering, technology investment, and the willingness to let junior lawyers' judgment be augmented, then replaced, by systems. The firms that thrive will be the ones that reorganize around that reality. The firms that cling to the hour will find their best clients — the banks, the tech companies, the sophisticated repeat buyers — simply routing work elsewhere.

There is a historical echo worth noting. Every previous wave of legal technology — word processing, e-discovery, legal research databases — was absorbed by the billable hour without disturbing it, because none of them changed who did the work. AI changes who does the work. When the associate army that justified the pyramid stops being necessary, the pyramid's economics stop working. Meshel's "different working model" within a year may be optimistic as a timeline. As a direction, it is almost certainly right.

Watch the next round of bank panel reviews and the year-end billing data. If fixed-fee arrangements and competitive bidding keep spreading from the banks to other sophisticated buyers — and if the firms' own surveys keep showing AI-driven hour reductions — the billable hour's long reign will not end with a dramatic collapse. It will end the way most business models end: as a premium option for bespoke work, while the volume business moves on without it.

Sources: The New York Times ("As A.I. Makes Law Firms More Efficient, Clients Ask: 'Where's My Discount?'", September 26, 2026); Financial Times via the New York Post (September 1, 2026); Persuit data (via FT); Artificial Lawyer / BigHand survey analysis (September 16, 2026); AINave (Neill Jakobe / Ropes & Gray remarks).