Audit Under Crossfire: Partner Cuts at KPMG Find BIS Warning on AI Capex

The calendar brought together two news pieces that, in isolation, would sound insignificant. The Bank for International Settlements published its 2026 Annual Economic Report on Sunday, June 28, warning that the top five hyperscalers surpassed $1 trillion in accumulated AI capex between 2025 and 2026. In parallel, KPMG continues to execute the program announced on April 24, which removes about 100 partners, 10% of its auditing practice in the United States, and cut approximately 400 additional positions in its American advisory division. What connects the two events is the fee pressure mechanism that defines the financial health of auditing.

The simplistic thesis argues that AI is dismantling the cost structure of the Big 4 and that KPMG's cuts signal capitulation. This interpretation deserves caution. KPMG communicated in an internal memorandum that the cuts are part of a multi-year "rightsizing" and that the business remains strong, with client gains in auditing. This counter-argument carries weight: the number of mandates has not decreased; what has declined is the profit per partner hour in segments where productivity gains through automation translate into price pressure.

The Counterpoint Argument Worth Mentioning

KPMG itself supports the opposing interpretation. The internal announcement sent on April 23, cited in a report by Accounting Today, described the move as "multi-year rightsizing" and stated that the goal is to "adjust the ratio of partners," rather than respond to a drop in demand. The firm reiterated that business remains robust, with client gains in the audit line, and that the departures are not tied to individual performance. From the firm's perspective, attributing the cuts to AI could be seen as storytelling on a post-pandemic capacity realignment that was already on the calendar regardless.

The counterpoint holds empirical merit. The Big 4 hired around 30% above historical trends between 2021 and 2023, and five years later, demand for regulatory and compliance hours is in relative decline, especially in segments like SOX testing and KYC, where specialized tooling captures the gains. However, the thesis has a flaw: if the problem were merely overstaffing, the cuts would occur at the lower-end of the pyramid. The choice to remove partners—positions of higher unit cost and more difficult to replace—suggests that the firm is calibrating for a future regime of compressed margins at the top.

The Data That Weakens the Optimistic Reading on AI

An inconvenient piece of evidence for those fully invested in the productivity thesis comes from Accenture itself. The consultancy reported in mid-June guidance cuts for revenue to 3-4% for the year, down from a previous estimate of 3-5%, and the stock fell 18% on the day, marking the largest one-day decline in its history. Cognizant lost 11%, Capgemini dropped 9%, and Infosys ADRs declined nearly 10%. The relevant detail is that Accenture is the AI-focused Big Tech consulting firm within the sector, and this stumble suggests that translating technical capability into revenue is not happening at the expected pace.

This changes the landscape. If the biggest natural beneficiary of the AI boom in consulting is not managing to grow at the promised rate, the pressure on traditional auditing and advisory practices is more acute, not less. There is no Plan B for revenue compensating for what is exiting the core.

The Distinction That the Shallow Debate Ignores

A profitable hyperscaler financing capex with operating cash flow is not an AI lab burning through a venture round. The BIS makes this distinction in Bulletin 120, signed by Iñaki Aldasoro, Sebastian Doerr, and Daniel Rees. This difference also applies to auditing: the Big 4 that offer AI advisory develop a productivity thesis that they themselves deliver, with recurring revenue, while the Big 4 that will only audit companies exposed to opaque multi-year compute contracts bear reputational risk without capturing comparable upside.

The BIS refers to this as circular financing: hyperscalers take equity stakes in AI labs that commit to buying chips and capacity from their own investors. "The terms are often less disclosed," states the annual report, "with the risk of the same asset being pledged multiple times." Auditing companies within this chain requires a methodology that the Big 4 have yet to publicly consolidate, and the litigation risk of a misstatement in a multi-year compute contract with an equity component is difficult to price ex-ante.

The Geographical Perspective

In the United States, capacity adjustments are occurring first in public auditing, precisely where the demand cycle is most visible. KPMG has also announced the multi-year exit of around 450 federal audit professionals. In India, where Deloitte, EY, PwC, and KPMG operate centers of excellence serving global audit delivery, adjustments tend to appear with a lag of six to twelve months, typically through hiring freezes rather than direct layoffs, a pattern already observed in Bangalore and Gurgaon.

In the UK, KPMG has already communicated cuts in the local division in May, explicitly justifying it by the increasing use of automation in substantive testing. In Brazil, where the Big 4 serve 100% of listed companies in the Novo Mercado, the second-order effect is less visible but equally real: client fee pressure from global multinationals spills over to Brazilian offices, which historically sustain local audit margins through carve-out work on consolidated financial statements.

The right question is not whether AI will cut positions in auditing; that process is already a reality. The question is whether the Big 4 can shift the revenue base from partner hours to AI-assisted assurance products before the margins of legacy contracts drop below the operational breakeven point. Whoever answers first will define the industry landscape for the next decade, and none of the four firms has yet communicated a convincing public strategy in that regard.