How many jobs is KPMG cutting?

About 200 roles in its UK advisory business, around 4% of the permanent advisory workforce, in the data and technology division. Affected staff are expected to leave in October, subject to consultation.

Why is KPMG cutting AI roles?

KPMG cited market dynamics combined with low levels of attrition, and told staff it needed to right-size capacity. Advisory revenue fell 3% in the year to September 2025. The cuts reflect lower demand for advisory hours, not a view that AI skills no longer matter.

Is this the first round of KPMG cuts in 2026?

No. In March, more than 500 UK roles were put at risk, mostly in audit, and in July around 200 roles were cut in group corporate services, including HR, marketing, technology and procurement.

Are AI jobs safe in consulting?

AI skills help, but they do not protect a role on their own. When clients buy fewer advisory hours, specialist and generalist capacity are both exposed. Work tied to products, fixed-price outcomes or systems clients keep using is more durable.

← All insights

September 23, 2026· 4 min readProfessional ServicesKPMGConsultingAI Strategy

KPMG Layoffs: Why the Firm Is Cutting Its Own AI Specialists

KPMG is cutting around 200 advisory roles, including teams working on AI and cyber security, and cites low attrition. Being the AI expert does not protect work that is still sold by the hour.

Quick answer

KPMG confirmed on 15 September 2026 that it is cutting about 200 roles, roughly 4% of its UK advisory workforce, including teams working on AI and cyber security. It cited market dynamics and low attrition. The cut targets billable capacity, not skills, so AI expertise sold by the hour faces the same squeeze as any other hourly work.

KPMG is cutting around 200 roles in its UK advisory business, and teams working on artificial intelligence and cyber security are among them. The firm gave two reasons: market dynamics and low levels of attrition. For anyone who has been told that learning AI is the way to stay safe in professional services, this is an uncomfortable data point.

What KPMG announced

The cuts were reported on 15 September 2026 by City AM and The Irish Times, and confirmed by the firm.

Size: about 200 roles, roughly 4% of the permanent advisory workforce.
Where: the data and technology division, including AI and cyber security teams.
Timing: affected staff expected to leave in October, subject to consultation.
Stated reason: "market dynamics combined with low levels of attrition", and a need to "right-size" capacity.

KPMG's statement read: "As our market evolves, we are adapting where we are focusing and how we are set up to make sure we have the right skills in place to best serve our clients."

The third round this year

KPMG UK workforce reductions in 2026
When	Roles	Area
March	500+ at risk	Mostly audit (about 440), plus about 120 in advisory
July	About 200	Group corporate services: HR, corporate affairs, marketing, tech, procurement
September	About 200	Advisory data and technology, including AI and cyber

The backdrop is softer demand. Advisory revenue fell 3% in the year to September 2025, according to The Irish Times, and the whole consulting sector has been adjusting after its pandemic hiring boom.

Why would a firm cut the people who work on AI?

Because what is being cut is not a skill. It is billable capacity.

A consulting firm earns money when its people are working on client engagements. When clients buy fewer advisory hours, the firm carries people it cannot bill. Low attrition makes that worse: when fewer people leave on their own, the firm has to decide who goes.

At that point, what the hours were about matters less than whether anyone is buying them. An AI specialist without a client engagement costs the firm the same as a generalist without one.

What this means for agencies and consultancies

Many smaller firms are repositioning as AI experts right now: new service pages, new job titles, new proposals. Often the model underneath has not changed. It still sells time, bills hours and grows by adding people.

That leaves AI expertise exposed to exactly the same pressure as everything else sold by the hour. The label changed. The unit did not.

We have seen the same dynamic in other data this month. The New York Fed survey found businesses are reshaping work more than cutting it, and the roles that disappear quietly are the ones that are never backfilled.

How to make AI expertise durable

Sell an outcome, not the hours behind it. A fixed-price result keeps its value when the work gets faster.
Turn repeat advice into a product. If you explain the same thing to every client, package it as a tool, template or system.
Leave something running. Automations and workflows a client keeps using after the engagement ends are harder to cut than a block of advisory days.

The firms that do well from here will not be the ones with the most AI people. They will be the ones whose AI knowledge shows up in something a client can keep, measure and pay for without counting hours.

Frequently asked questions

How many jobs is KPMG cutting?: About 200 roles in its UK advisory business, around 4% of the permanent advisory workforce, in the data and technology division. Affected staff are expected to leave in October, subject to consultation.
Why is KPMG cutting AI roles?: KPMG cited market dynamics combined with low levels of attrition, and told staff it needed to right-size capacity. Advisory revenue fell 3% in the year to September 2025. The cuts reflect lower demand for advisory hours, not a view that AI skills no longer matter.
Is this the first round of KPMG cuts in 2026?: No. In March, more than 500 UK roles were put at risk, mostly in audit, and in July around 200 roles were cut in group corporate services, including HR, marketing, technology and procurement.
Are AI jobs safe in consulting?: AI skills help, but they do not protect a role on their own. When clients buy fewer advisory hours, specialist and generalist capacity are both exposed. Work tied to products, fixed-price outcomes or systems clients keep using is more durable.