THE AI
RECKONING

Hourly billing did not become the dominant pricing model because clients and firms sat down and agreed it was the best way to price legal work.

It became dominant because each side independently decided it was the most convenient option for them. Firms liked it because it was simple and linked effort to revenue. Clients tolerated it because it felt transparent and there was no obvious alternative that the two parties had built together.

Over time, the problems with that arrangement became clear to everyone. Incentives were misaligned. Costs were unpredictable. Quality was hard to measure. But instead of coming together to design a better model, each side adapted independently. Firms raised rates, hired more associates, introduced legal project management tools, and built their entire financial structure around billable time. Revenue-per-partner, compensation models, promotion criteria, firm valuations: all of it was anchored to the billable hour.

Corporate legal departments, meanwhile, developed their own set of tools to manage the dysfunction: e-billing systems, automated invoice review, outside counsel guidelines, billing rate caps, staffing restrictions. An entire legal operations discipline grew up around the task of controlling outside counsel spend within a pricing model that neither side would have designed from scratch.

Both responses were rational. Neither addressed the root problem. For decades, the industry lived with a pricing model that clients resented and firms depended on, because no one initiated the conversation to build something that worked for everyone. Each side raced to develop technology and tools to outpace the other, turning it into an arms race. And the resulting dysfunction became the water the industry swims in today. Decades of these issues and ninety percent of legal dollars still flow through standard hourly arrangements.[9]

AI has the potential to break this cycle. The economic pressure is certainly there: 61% of in-house counsel plan to push for changes in how legal services are priced.[3] Thomson Reuters found that 44% of legal professionals predict AI will drive a decline in hourly billing.[9] The Wolters Kluwer 2026 survey says 62% believe AI-driven efficiencies will reduce billable hours.[4]

But look at how the shift is actually playing out. Clients are writing new billing guidelines unilaterally. Firms are developing new pricing models on their own. Clients are setting AI use requirements that their firms had no input in designing. Firms are building AI governance policies that their clients have never seen. It is the billable hour story all over again: instead of building a shared model, each side is developing its own set of workarounds, and those workarounds are already starting to conflict.

In-house teams will show up at the next panel review with requirements their firms had no role in shaping. Firms will present capabilities that do not map to what clients actually need. Billing guidelines will clash with pricing proposals. Governance frameworks will overlap in some places and have gaps in others. Anyone who has lived through a contentious rate negotiation knows exactly how this ends.

The financial picture is where the collaboration failure becomes hardest to ignore.

Lawyers using GenAI tools save up to 260 hours per year.[10] Document review is faster and, in controlled tests, more accurate than human-only work. The efficiency gains are real.

And yet, 59% of in-house teams that know their firms use GenAI have seen no cost savings at all. Only 24% are satisfied with how outside counsel are using AI for cost effectiveness.[3] Meanwhile, law firm revenue jumped 13% in 2024, net income rose 17%, and Am Law 50 firms posted rate increases north of 10%.[9][12] In-house departments are, in effect, subsidizing their firms' AI investment without seeing any return.

Work finds other channels

Without a joint conversation about how to share those efficiency gains, the economics push work elsewhere. Nearly two-thirds of in-house teams expect to rely less on outside counsel.[3] Some departments now handle 75% of their work internally.[4] The $28.5 billion alternative legal service provider market is absorbing much of the rest. Work flows to wherever it can be done most efficiently. If firms and their clients were collaborating on AI-enabled workflows, much of this work might stay within the relationship, delivered faster and priced differently. Without collaboration, the work finds other channels. That is not a threat. It is just gravity.

A telling case from outside the legal industry: in February 2026, KPMG International pushed its auditor Grant Thornton UK to cut fees, arguing that AI efficiencies should mean lower costs. Grant Thornton's fee dropped about 14%.[11] That was not a collaboration. It was a power play. If legal follows the same path, we get a decade of adversarial pricing negotiations that leave everyone worse off.

If collaboration is so clearly better, why is almost nobody doing it? Part of it is structural. Part of it is habit. And part of it is a communication breakdown that has been building for years.

Firms are not ignoring AI. They are investing in it heavily. But they are waiting for clearer direction from their clients before committing to a specific model for how AI should change the way they work together. The signals coming from CLDs are contradictory: most are not asking firms to use AI, but many are simultaneously preparing to evaluate firms on their AI capabilities during the next panel review. Firms read those mixed signals and hedge. They build internal capabilities without putting a stake in the ground on pricing, governance, or workflow changes that would require client buy-in. The structural incentives reinforce the hesitation. Partner compensation is still tied to hours billed or revenue originated. A partner who proposes an AI workflow that cuts billable hours by 40% is, under most compensation models, volunteering for a pay cut. Firms also compete with each other for panel slots, and sharing performance data with a client risks having that data used against them in the next negotiation.

In-house teams, meanwhile, are sending the indirect signals that firms find so hard to read. Rather than inviting firms into a joint planning process, departments are writing AI guidelines and issuing them as requirements. Rather than asking firms what they are building and how it could be applied to shared work, they are publishing questionnaires and scoring rubrics. The intent may be collaborative, but the method is unilateral. Legal ops functions are also stretched thin, already running internal AI adoption while managing day-to-day outside counsel oversight, and adding a genuine collaboration workstream with each firm feels like more than the team can absorb.

Underneath all of it sits a trust deficit that predates AI by decades. Hourly billing disputes, opaque staffing decisions, rate negotiations that feel more like trench warfare than partnership: these experiences have left residual tension in many client-firm relationships. Building a joint AI program requires a level of openness and shared investment that some of those relationships have not earned yet. It is hard to co-develop a new operating model with someone you have been arguing with over invoices for ten years.

None of these barriers are permanent. But they explain why, despite the obvious logic of collaboration, the default remains two playbooks in two rooms.