Short answer: not yet, and not in the way the law firm press releases suggest. Across 8,028 law firm deal announcements published on Legal Desire between January 2023 and September 2026, the firms with the loudest AI programmes did not take a measurably bigger slice of announced deals once their tools went live. What did not happen is just as telling: deal teams did not get smaller.
Big Law has spent the last three years announcing AI. A&O Shearman went first with Harvey in February 2023. Clifford Chance rolled out Microsoft Copilot to its whole workforce in February 2024. Latham & Watkins licensed Harvey firmwide in August 2025, Linklaters rolled out Legora a month later, and in May 2026 the Financial Times reported that Kirkland & Ellis has set aside around US$500 million to build its own AI platform. Thomson Reuters puts law firm technology spending growth at 9.7 per cent in 2025, probably the fastest real growth on record.
The question every managing partner is being asked by their partnership is simple: is any of this winning work? Deal announcements are one of the few public places where you can look for an answer.
What we measured
Every deal announcement published on legaldesire.com is coded field by field into DealDatabase: the firms, the named lawyers on each team, the client, the value, the jurisdictions. For eight firms with a dated, public, firmwide AI announcement, we compared two things in the 12 months before the announcement and the 12 months after (or up to today, where 12 months have not yet passed):
- Share of announcements: the percentage of all deal announcements in that window that name the firm in any adviser role.
- Named team size: the average number of that firm’s lawyers named on each of its deals.
An important caveat, which we repeat in every DealDatabase analysis: this is published announcements, not market activity. A firm appears more often when more of its deals are announced and reach our newsroom. That limitation turns out to be part of the story.
The numbers
| Firm and AI move | Share before | Share after | Named team before | Named team after |
|---|---|---|---|---|
| Clifford Chance: Copilot firmwide and CC Assist (Feb 2024) | 12.5% | 13.3% | 9.9 | 10.0 |
| Latham & Watkins: Harvey firmwide (Aug 2025) | 17.0% | 31.9% | 12.4 | 13.8 |
| Linklaters: Legora firmwide (Sep 2025) | 7.9% | 9.3% | 8.1 | 8.5 |
| Herbert Smith Freehills Kramer: Legora firmwide (Mar 2026) | 5.3% | 5.5% | 7.5 | 7.4 |
| DLA Piper: Harvey firmwide (Mar 2026) | 4.6% | 5.5% | 7.8 | 7.3 |
| Mayer Brown: firmwide GenAI curriculum (Apr 2026) | 2.5% | 1.9% | 7.3 | 5.8 |
| Baker McKenzie: Legora global rollout (May 2026) | 6.6% | 5.1% | 5.7 | 7.6 |
Share = percentage of all deal announcements in the window naming the firm. Named team = average number of that firm’s lawyers named per deal. Windows run 12 months either side of the announcement date, or up to 30 September 2026.
Reading the table honestly
Most firms barely moved. Clifford Chance, Linklaters, Herbert Smith Freehills Kramer and DLA Piper each shifted by around a point or less. Mayer Brown and Baker McKenzie slipped. If firmwide AI were a decisive edge in winning mandates, you would expect a clearer pattern across firms that adopted it at different times. There isn’t one.
Latham is the exception, and it is probably not about Harvey. Latham’s share of our announcements nearly doubled after its August 2025 rollout. It would be tempting to credit the tool. The more likely explanation is that Latham announces a great deal of its work, consistently, and its releases reached our newsroom in higher volume over that period. Latham was already the most-announced firm in our data in 2023 and 2024, at about 22 per cent.
The biggest AI spender is almost invisible. Kirkland & Ellis, with the reported US$500 million budget, appears on just two of the 1,928 announcements we published in 2026, down from about 12 per cent of announcements in 2023 and 2024. Kirkland did not stop doing deals. Its announcements simply stopped reaching us. Freshfields shows the same gap after early 2025. That is the most useful lesson in the dataset: share of announcements tracks how firms publish, not how they work. AI spend sits a long way upstream of either.
The finding that matters: deal teams did not shrink
The fear, and for some the promise, of legal AI is that it does the junior work, so fewer lawyers are needed on each matter. If that were already happening on transactions, the number of lawyers firms name on each deal should be falling.
It isn’t. Named teams held steady or grew at Clifford Chance (9.9 to 10.0), Latham (12.4 to 13.8) and Linklaters (8.1 to 8.5) after their rollouts. Across the whole dataset, the average number of lawyers named per announcement rose from 9.3 in the second half of 2024 to 10.5 by mid 2026.
Named teams are a proxy. Firms choose who to name, and junior associates are often left off. But the direction is clear enough: two and a half years into Big Law’s AI era, firms are still putting as many, or more, people in front of clients on announced deals.
So where is the return going?
Three places, none of which shows up in a league table yet.
- Margin, not market share. Thomson Reuters’ 2026 State of the US Legal Market report shows demand growth of about 2 per cent at Am Law 100 firms in 2025 against about 5 per cent at midsized firms, while rates rose 7.3 per cent and profits 13 per cent. The biggest firms are not winning more work. They are earning more on the work they have, and AI is part of that story.
- Products sold to clients. A&O Shearman sells ContractMatrix and shares revenue on the agents it builds with Harvey. Kirkland has built a fund formation platform with Palantir. Cooley launched an IPO tool with OpenAI in September 2026. That income never appears in a deal announcement.
- Talent and prestige. AI programmes are now a recruiting and pitching signal. Firm Prospects data reported by Law.com shows 46 lawyers moved from Am Law 200 firms to AI companies in the first half of 2026, most of them associates. Firms are spending partly to keep people, not only to win mandates.
My view
Firmwide AI has become table stakes. When Harvey, Legora, Copilot or Claude sits on every desk in the top 30 firms, it stops being a reason a client picks one firm over another. Being without it is a disadvantage; having it is not an advantage.
The firms that turn AI into deal share will be the ones that change something a client can see: a fixed fee they could not offer before, a timetable they could not hit before, or a product the client keeps using after the deal closes. Until then, the deal tables will keep being decided by the old things: relationships, sector depth and who picks up the phone.
The signals I will be watching in our data over the next twelve months are simple. Do named teams finally start to shrink? Does disclosed value per lawyer rise? Do firms that sell AI products to clients start appearing on more deals for those same clients? When those move, the spending will have started to show up.
Frequently asked questions
Which law firms have invested most in AI?
The largest reported figure is Kirkland & Ellis, which the Financial Times reported in May 2026 has budgeted around US$500 million over three to four years for its own AI platform. Other major firmwide moves include A&O Shearman (Harvey, 2023), Clifford Chance (Microsoft Copilot, 2024), Latham & Watkins (Harvey, 2025), Linklaters (Legora, 2025), Freshfields (Google Cloud in 2025 and Anthropic in 2026), Baker McKenzie (Legora, 2026) and Skadden (OpenAI, 2026). Most firms do not disclose what they spend.
Has AI helped law firms win more deals?
Not measurably, on the evidence of announced deals. In DealDatabase’s analysis of 8,028 announcements from 2023 to 2026, firms’ share of announcements did not consistently rise after they rolled out AI firmwide. The clearest return so far appears in profitability and in AI products sold to clients, not in deal share.
Is AI reducing the size of deal teams at big law firms?
Not yet. The average number of lawyers named per deal announcement rose from 9.3 in late 2024 to 10.5 in mid 2026, and named teams at Clifford Chance, Latham & Watkins and Linklaters held steady or grew after their AI rollouts.
What is the difference between Harvey and Legora?
Both are generative AI platforms built for law firms. Among large firms, Harvey is used by A&O Shearman, Latham & Watkins, Slaughter and May, DLA Piper and Paul Weiss, among others. Legora has been chosen by Linklaters, Herbert Smith Freehills Kramer, Baker McKenzie, White & Case and Goodwin.
The deal records behind this analysis, with the firms, lawyers and disclosed values on every announcement, are searchable at DealDatabase.co. Firms can submit their announcements through the Legal Desire Press Desk.
Method: deal announcements published on legaldesire.com from 1 January 2023 to 30 September 2026 (8,028 records) and coded into DealDatabase, with firm name variants merged. AI announcement dates are taken from each firm’s or vendor’s public release. Percentages are share of announcements in each window; firm figures count every announcement naming the firm in any adviser role. Named team counts only lawyers named in the announcement. This measures published announcements, not market activity.
Sources: A&O Shearman and Harvey (Feb 2023); Clifford Chance and Microsoft (Feb 2024); Latham & Watkins and Harvey (Aug 2025); Linklaters and Legora (Sep 2025); HSF Kramer and Legora (Mar 2026); Baker McKenzie and Legora (May 2026); Mayer Brown GenAI curriculum (Apr 2026); Financial Times on Kirkland’s AI budget (May 2026); Kirkland and Palantir (Jun 2026); Skadden and OpenAI (Sep 2026); Cooley GO Public (Sep 2026); Thomson Reuters 2026 State of the US Legal Market, via LawNext; Law.com on lawyers moving to AI companies (Sep 2026).
Where the legal industry reads first.
Enjoyed this article? Get the biggest legal industry updates, deals, appointments, insights and expert interviews in your inbox, free.
No spam. Unsubscribe anytime.From Legal Desire
Is your firm the one being cited, or the one being skipped?
We ran the test. On several everyday legal questions, software companies are answering and law firms are not. We help firms fix that, and we build the technology behind the practice.
