AI Eats the Bottom of Law Firms Before Impacting Partners: Why the BigLaw Pyramid is Sinking

K&L Gates confirmed in July the layoff of approximately 10% of its non-legal staff, roughly 130 to 170 people from a global payroll of over 3,000 professionals, following a months-long review of its support structure. The statement, made by spokesperson Steven Eliopoulos, was surgical: none of the lawyers were affected; all cuts fell on "allied professionals", from operational entry-level to senior managers. This was the latest cut in a series of decisions that began in 2025 and today forms a clear pattern at the top of the industry: AI first eats away at the bottom of the firm before it affects any partner sheet.

The name that is often missing from this conversation is that of those who had already begun to spend cash earlier. Kirkland & Ellis, the highest-grossing firm in the world, announced in May 2026, through chair Jon Ballis, an investment of $500 million in proprietary AI over four years, with $100 million allocated for 2026 alone. The firm involved about 250 lawyers in the platform design and more than 180 technology professionals in development. This is not the only significant investment. Baker McKenzie opened the cycle in February this year, when it announced the cut of 600 to 1,000 business services positions, the largest downsizing explicitly attributed to AI in legal history. Prior to this, in November 2025, Clifford Chance cut 10% of its business staff in the UK, and in September, Freshfields shuttered part of its paralegal hub in Manchester.

What the Numbers Hide

The easy reading is that AI is streamlining support. The correct reading is more precise: AI is first eliminating the tier that existed between the client and the senior attorney. A study by McKinsey cited in the 2026 report from 8am Legal Industry estimates that 69% of typical paralegal tasks can be automated with technology currently available, and hiring of paralegals in the 250 largest American firms has dropped by 26% since 2018. The first-pass document work, the so-called "first-year work" that sustained the BigLaw pyramid for three decades, is exactly the type of task in which models excel and improve with each release. General legal employment in the United States is still at a historical high, with 1.208 million positions in December 2025 according to the BLS. What changes is the shape of the pyramid, not its size.

There are voices urging caution. Fried Frank, questioned in July about alleged job cuts due to AI, publicly denied this. Recruiters interviewed by Law.com and the ABA Journal have repeated that AI alone does not explain all recent layoffs and that some firms are using the technology narrative to justify cost adjustments that were already inevitable due to declines in M&A and finance mandates in a high-interest rate cycle. This is the same valid criticism regarding layoffs in consulting and investment banking: AI accelerates decisions that the economy has already made possible. Accepting this consideration does not negate the industry pattern; it adjusts the weight of the cause.

How the Wave Reaches Two Geographies Beyond the U.S.

In the UK, the pattern has been faster. Clifford Chance and Freshfields made cuts before the American firms for similar reasons, with tougher accounts due to office costs in London and a weak capital markets cycle. Allen & Overy Shearman, now merged, has been pressuring the paralegal base in finance accounts, and the knock-on effect is seen in the Belfast hub, which absorbs a lot of commodity work. The move of former Harvey Partnerships VP Suril Patel to Kirkland in July, with prior tenure at Allen & Overy, highlights how the talent market in legal AI flows between Wall Street and the City.

In India, the effect is indirect and worth more money. The captive alternative delivery centers of American BigLaw in Gurgaon, Bangalore, and Hyderabad, which have emerged over the last ten years as a cheaper option than traditional outsourcing, are the points of greatest exposure. If AI absorbs documents, repetitive contracts, and basic diligence, the pipeline that sustains these hubs dries up from within. This is the same reading that has appeared in the Acceleration Centers of the Big 4 and in the Indian hubs of TCS and Infosys.

The structure of BigLaw in the next five years will depend less on technology and more on the courage of partners to dismantle their own billing model. The prevailing question is whether any major partner will accept a lower commission per billable hour when the hour costs significantly less to produce. Until this conversation happens in the executive committee, each new cut announcement will first fall on those who do not sign the billing sheet.