Kirkland & Ellis has committed half a billion dollars to build its own AI platform. Most lawyers read that as the giants pulling away. Read with care, it is the opposite — an expensive confirmation that the lawyer who builds his own tools is working from the stronger position.
Kirkland & Ellis has committed half a billion dollars over the next three to four years to build its own artificial intelligence platform. The Financial Times broke the number, and Bloomberg Law confirmed its contours. The firm’s chair frames it as a bet his firm’s scale earns the right to make, a big swing of roughly the one percent of revenue he says a firm generating ten billion a year should put behind new initiatives. One hundred and eighty technologists are building the platform, informed by some two hundred and fifty lawyers describing how they actually do their work. By any measure, it is one of the most ambitious technology bets a law firm has ever placed.
It is not the only one. Fried Frank is rolling out an internally built platform for its private equity funds group. Linklaters has stood up a team of data scientists and lawyers to build bespoke workflows. Littler Mendelson created a new senior post to steer adoption of a tool it built in-house. The wave is real, and the reflex reading is that the giants are pulling away from everyone practicing below that altitude.
For everyone practicing below that altitude, that reading is wrong. Read with care, Kirkland’s five hundred million is not a warning shot. It is a confirmation, and an expensive one, that the lawyer who builds his own tools is working from the stronger position. The proof sits in the structure of the spend itself, which carries a cost the giant pays and the lawyer who builds does not.
Start with the diagnosis from the academy. A University of Houston law professor, Seth Chandler, put the problem plainly: once every associate at every top firm holds the same chatbots and the same research tools, the tools stop being a point of difference. What is left to compete on is the firm’s own internal knowledge. The model is rented, and everyone rents the same frontier. The advantage has to come from inside.
That cuts against the usual instinct to credit whoever moved first. The advantage was never the start date, and it was never the software. It is the depth of knowledge a firm pours into the tools, and that depth does not arrive on a launch day. It accumulates over years of practice. Whoever has been building that depth the longest, and can route it into the software with the least loss along the way, holds the one asset the tool itself cannot supply. A law professor just said as much, on the record.
Now look at the architecture of the spend, because the architecture is the argument. One hundred and eighty technologists. Two hundred and fifty lawyers explaining how they did their jobs. Two separate groups of people, joined by a project. A large share of that half-billion is the cost of translation: the work of pulling a lawyer’s knowledge out of the lawyer’s head and getting it, intact, into the hands of the people who write the code.
Every hand-off in that chain loses something. The lawyer half-describes a judgment that took decades to form, the engineer half-understands the description, and the software ends up with a secondhand version of both. The most valuable thing in the building is a seasoned lawyer’s sense of what actually matters in a deal or a filing, and at Kirkland that sense has to survive a trip across the line between two professions before it ever reaches the code. That trip is the largest cost nobody writes on the invoice. A good part of five hundred million dollars is what it takes to shorten it.
There is a way to avoid the trip altogether: have the same person hold the judgment and write the tool. Not every lawyer who learns to code qualifies, and neither does every lawyer who buys software. The person I mean is narrower: a lawyer with decades of judgment in a field, the working technical skill to build and maintain real platforms rather than merely commission them, and the patience to iterate until the result holds up against what a large firm can field. For clarity I will give that person a defined term and use it consistently: the Lawyer Developer.
When the practitioner and the builder are one person, there is no extraction, no translation, and no bridge between two hundred and fifty lawyers and one hundred and eighty engineers to fund and manage. The judgment a transactional attorney accumulates over thirty-plus years never leaves the head that earned it. It goes straight into the build, supervised by the same mind that holds it. Kirkland’s organizational chart is the best argument I have seen for why a Lawyer Developer does not need the organizational chart.
The dollar figures do not compare, and chasing the comparison is a trap. Five hundred million is one percent of ten billion, spent to solve a problem that exists only at scale: thousands of associates who would otherwise hold identical tools and so compete on nothing. That is a real problem when you employ thousands of associates. It is not a problem I have. My advantage is not something I purchase. It follows from the arrangement itself. A single Lawyer Developer in a practice area is distinct by construction, because the asset is that lawyer’s own accumulated judgment, written directly into the work. There is nothing to buy back, because nothing was lost along the way.
The arrangement is easy to state: the software does the work, and the attorney reviews it, shapes it, and signs it. The lawyer sits above the technology and directs it, rather than standing beside it as one of two groups negotiating across a gap. That produces the very thing the half-billion is straining to manufacture, a lawyer’s expertise expressed in the software without loss, and here it arrives as a byproduct of who is doing the building.
The most instructive detail sits at the bottom of the coverage. In the same stretch of weeks, Kirkland advised Blackstone on the launch of an enterprise services firm built to bring Anthropic’s Claude into companies’ core operations. The largest firm in the country is being paid to guide others toward a destination, frontier AI woven into the daily operation of the business, that a single lawyer can already occupy today, at little marginal cost, inside a compliance perimeter that took weeks rather than years to stand up.
The firm advising the migration and the lawyer who has already arrived are not the same kind of actor. There is nothing inconsistent in Kirkland doing both, since advising the migration is good business, but a lawyer reading the news should notice which side of that transaction is the easier one to occupy alone.
A University of Texas law fellow, Kevin Frazier, offered the cleanest line in the coverage: today’s rate of AI adoption across the profession is the lowest it will ever be. Adoption only climbs from here, and it climbs faster than the straight line most of the profession is penciling into its planning. The change that looks a decade away tends to arrive sooner, and all at once. Much of the profession will look up at the end of this year and ask what happened, and the answer will be that the curve did what curves do while they read it as a line.
Be straight about the advantage scale really delivers: breadth and parallel effort. Kirkland builds across every practice at once, with a redundancy and reach a single lawyer cannot match. I build one area at a time, across six practice areas, on a recurring ninety-day cycle. That asymmetry is real, and pretending otherwise would be foolish.
The right response is not to imitate it. It is to decline the comparison and compound on the one axis where being a single person is the advantage: depth in each area, with no translation cost, revised faster than an annual planning cycle can react. You do not need one hundred and eighty engineers. You need to never lose the connection between knowing the law and building the tool. That connection is the one thing a Lawyer Developer cannot lose, and it is the very thing half a billion dollars is being spent to recover.
So the headline is right that Kirkland has made one of the most ambitious technology bets a law firm has ever placed. It is wrong about what the bet proves. It proves the model is a commodity, the knowledge is the advantage, and the most expensive problem in legal AI is getting a lawyer’s expertise into the software without losing it in translation. A firm with ten billion in revenue solves that problem by spending five hundred million to shorten the distance between two groups of people. A lawyer who builds solves it by closing the distance to zero.
Code Counsel is the firm’s publication, written by a Texas attorney with 30+ years in practice and produced on the AI-augmented infrastructure the firm runs in-house. It is a joint Lozano–Claude publication: the AI executes, and the attorney reviews, shapes, and signs. When you call the firm, the licensed Texas attorney is the person who answers. (214) 531-3014.