PwC's global chair has quietly scrapped a pledge his predecessor made in 2021: to add 100,000 people to the firm's headcount by the middle of this year. Instead, PwC cut 5,600 jobs in the twelve months to June 2026, taking global headcount below 365,000. That came even as revenue rose 2.7% to $57 billion, growth that still lagged Deloitte and EY.
PwC is not the outlier. It's the most visible data point in a two-year trend across all four firms, and the part of it that should worry GCC clients isn't the headline number. It's where the cuts are landing.
The entry point is closing first
Graduate intake across the Big Four has fallen sharply over the past two years. KPMG made the steepest cut, trimming its graduate intake from 1,399 to 942, a 29% drop. Deloitte cut 18%, EY 11%, and PwC 6%. Across UK accounting more broadly, graduate job listings are down 44% year-on-year.
The reason is not hidden. The work that used to go to first- and second-year staff (compliance checks, document summarisation, first-draft memos, journal entries) is now handled by generative AI, often built in-house. EY now runs roughly 150 AI agents serving 80,000 tax professionals. KPMG has committed $2 billion over five years, targeting $12 billion in AI-enabled revenue. Deloitte's internal tool, Zora, automates invoice processing. PwC's own GL.ai handles journal entries. As one of PwC's AI assurance leaders put it, AI is now "functioning at more of the staff and routine task level", which is precisely the level graduate hires used to occupy.
Why this matters for a ten-year programme, not a two-year headcount chart
GCC national transformation programmes (the multi-year, multi-hundred-person delivery efforts sitting under Vision 2030, D33, TASMU and their equivalents) have historically leaned on Big Four graduate-fed delivery pyramids for the bulk of their execution capacity: business analysts, PMO support, testers, and delivery leads who climb the pyramid over a decade into programme directors.
That pyramid doesn't just get thinner at the bottom when graduate intake falls by a third. It stops producing the senior programme leads a client will need in eight to ten years: precisely the point at which the current wave of GCC giga-programmes moves from build phase into the sustained operational maturity that needs experienced delivery leadership, not entry-level capacity. A pyramid with no base in 2026 has no middle in 2033 and no new apex in 2036.
This is not an argument that AI shouldn't be doing junior-level work. Most programmes we see are thin at the review and oversight layer, not overstaffed at the doing layer, and automating routine analysis is a legitimate efficiency gain. But it is a structural question GCC programme sponsors should be asking their delivery partners now, while there's still time to plan around the answer, not in 2032 when the gap shows up as an unfillable programme director vacancy on a critical national initiative.
What GCC programme sponsors should ask now
- Ask delivery partners how they're building their next cohort of programme directors, not just how many AI agents they've deployed. "We've automated the entry-level work" is not an answer to "who leads this programme in 2034?"
- Watch for the hollow middle on multi-year contracts. A bid team stacked with senior partners and AI tooling can look impressive at kickoff and still have no visible pipeline of people who will still be there, experienced, in year six.
- Ask what "experience" means when AI has absorbed the tasks that used to build it. Judgement on a complex programme is built by doing years of the routine work first. If that route is closing, delivery partners need a different, explicit answer for how their people develop it.
- Weigh specialist boutiques and in-house capability-building more seriously. Not because they're immune to AI (they aren't) but because they were never built on the same graduate-volume assumption, and the strain on that assumption is now visible in the numbers.
The bottom line
A 29% cut to a single year's graduate intake is a headcount decision. Repeated across four firms, two years running, while revenue growth slows and AI absorbs the work that used to train juniors into seniors, it becomes a capacity decision with a decade-long delay before the bill arrives. GCC clients signing multi-year delivery contracts today are, in effect, pre-ordering the leadership bench their programmes will draw on in the 2030s. It's worth confirming that bench is actually being built.