Law Firms Are Spending Millions on AI for an Audience of None
Procurement budgets are soaring based on vendor promises, while nobody inside the building is tracking whether any of it actually works.
An AI-assisted editorial, reviewed by a human before publishing. It reasons over our own tracker data (and cited context). A point of view, not legal advice.
Ignore the eye-popping announcements about firmwide deployments for a second and look at 0%. That is the percentage of law firms in legal consulting firm Harbor's latest survey that reported having a mature framework to measure the business impact of the artificial intelligence tools they buy.
It is not that firms aren't buying. Harbor found average tech spending at law firms jumped 41% from 2021 to 2025, devouring between 5% and 6% of overall revenue. The industry is effectively throwing six cents of every gross dollar at technology platforms, while admitting in the same breath that it has no reliable way to prove any of it generates a return.
Harbor found average tech spending at law firms jumped 41% from 2021 to 2025—yet zero firms report having a mature framework to measure AI's impact.
The Corporate Defense for Unchecked Budgets
To be fair to the managing partners writing these checks, the corporate pressure to move fast is immense. According to a survey of 528 in-house legal leaders across six countries published by Axiom, every single in-house team currently using AI plans to increase its budget next year, with zero planning cuts. Vendors are flooding the market with numbers to capitalize on that fear of falling behind; published buyer guides from LuMay claim an average first-year return on investment of 410% for their legal agents, while ranking tools like Thomson Reuters' CoCounsel at 340% ROI and Harvey at 310%.
Inside Big Law, the arms race looks like a strategic imperative. Kirkland & Ellis is co-developing an enterprise platform with Palantir as part of a $500 million plan targeted at private equity fund formation. Jackson Lewis recently announced a deployment of Harvey across all 60-plus offices for its 1,100-plus attorneys, collaborating with vendor engineers to build labor-specific tools. Meanwhile, firms like Weil Gotshal, Cleary Gottlieb, Freshfields, and Williams & Connolly are lining up to partner on Google’s new Gemini Enterprise for Legal.
When your peers are putting half a billion dollars into custom platforms or putting algorithms into the hands of thousands of associates, doing nothing feels like professional negligence.
Panic Is Not a Financial Strategy
The problem is that FOMO is not a metric. The exact same Axiom study that logged zero budget cuts also revealed that 83% of in-house legal leaders cannot measure whether their AI spend is working, and only 7% have successfully scaled the technology across their organizations.
Law firms are running the exact same playbook, signing enterprise site licenses based on pitch decks rather than operational proof. We are watching firm management commit massive capital to platforms without setting baseline data for drafting hours, review burdens, error rates, or client billings. When Aderant launches an Agent Center to automate billing and talent evaluation, or Thomson Reuters connects CoCounsel to Everlaw to stream eDiscovery documents directly into research workflows, these tools are dropped into law firms that simply assume efficiency will magically follow.
Even when individual practices attempt to build bespoke tools—such as Weil Gotshal creating its proprietary BenchMark platform to analyze judicial rulings and transcripts—the broader enterprise rarely measures whether these investments reduce client costs or merely add overhead. Outside accounting guidelines and published advice on tracking contract ROI mean nothing if a firm's executive committee treats software procurement as a PR milestone rather than a capital discipline.
The Reckoning on the Billable Hour
Until firms establish baseline metrics to track turnaround time and matter profitability, buying software because everyone else is doing it remains pure theater. You cannot automate a workflow you haven't measured, and you cannot justify a 41% budget expansion based on vendor-supplied claims of 400% returns.
The legal industry's AI shift will eventually demand proof. Until partners demand actual operational data before renewing these multi-million-dollar software contracts, they aren't modernizing their practice. They are just writing blank checks to software companies.