Kirkland & Ellis
Build, at scale. $500m to create a proprietary, firm-owned AI platform encoding the 'collective intelligence' of its lawyers, while retaining third-party licences alongside it.
- $500m AI platform (unnamed)
Proprietary, firm-owned platform announced via the FT (27 May 2026). Built by outside technology companies plus Kirkland's own engineers and data scientists (180 tech professionals total); vendors barred from reselling it, with Kirkland to 'own, or have the right to own, all of it'. Designed using input from 250 lawyers (including 100 partners) about how they do their jobs; intended to run whole mandates end-to-end rather than serve as a task-level assistant. Name and technology partners due to be disclosed in the coming weeks.
- ·$500m total commitment: >$100m in 2026, hundreds of millions more over 3-4 years, funded from revenues (eating into equity-partner profit short term). The world's highest-grossing firm ($10.6bn revenue last year).
- ·Build is explicitly in addition to continued spend on licences for third parties' AI tools (not named; no public confirmation of which platforms - Harvey/Legora not confirmed as Kirkland customers).
- ·Strategic thesis - Ballis: off-the-shelf tools are 'raising the floor for everyone' but 'we don't get hired for the floor'. Differentiation at the top of the market comes from proprietary, expertise-encoded capability the firm fully owns.
- ·Firm-owned and not resold - the inverse of the A&O Shearman/Harvey revenue-share and the Freshfields/Anthropic co-development models, where the firm helps build something the vendor can commercialise.
- ·Pricing - endorsed value-based pricing displacing the billable hour: 'we already do a number of matters on value-based pricing... it will accelerate... we're going to lean into it'.
The largest publicly-stated 'build' commitment in the market, and the first of the 'Quiet Elite' (Cravath, Davis Polk, Skadden) to go public - emphatically so. The size matters less than the framing: the most profitable firm in the world has concluded that buying off-the-shelf commoditises the baseline and that elite differentiation requires proprietary, fully-owned tooling, while still keeping its third-party licences. Execution risk is real (undisclosed product, budget not result), but it raises the capital bar for 'build your own platform' well beyond what most firms can match.