From Sanskriti Khandelwal | Product & Market Analysis

Service AI Spend Rose 38% on a 2% Budget. Here Is What It Displaced

On this page

Customer service leaders raised AI spending 38% while their total service and support budgets grew 2%, according to a Gartner survey of 199 leaders published on 26 August 2026. A budget that barely moves cannot absorb a line growing 19 times faster without something else shrinking. Gartner customer service AI spending is now a reallocation story, and the public filings of the largest contact center outsourcers show where the money is leaving.

Key takeaways

  • A 38% AI increase inside a 2% budget forces cuts elsewhere. On Gartner's reported figures, the non-AI part of an average service budget falls by roughly 1.8% even before any other technology line grows.
  • Labor and overhead are the named source of the money. Gartner director analyst Kim Hedlin said leaders are redirecting spending away from labor and overhead and toward technology to fund AI initiatives.
  • Direct layoffs are a minority move so far. A separate Gartner survey of 275 leaders found 21% had cut human agent headcount because of AI, with an average cut of 16.7% over 18 months among those that did.
  • Outsourced contact center volume is shrinking first. TTEC's Engage revenue fell 12.1% in Q2 2026, Concentrix guided Q4 revenue down 3% to 5% in constant currency, and TP blamed automation for lower Trust and Safety volumes.
38% vs 2%Growth in service AI spending against growth in the total service and support budget. Source: Gartner, August 2026.
21%Share of service leaders who say they have reduced human agent headcount because of AI. Source: Gartner, November 2026.
-12.1%Year-on-year revenue change in TTEC Engage, its outsourced customer experience segment, Q2 2026. Source: TTEC 8-K, August 2026.

What Gartner's service budget survey actually found

The headline comes from a Gartner survey of 199 customer service and support leaders, run in April and May 2026 and released on 26 August. Leaders reported that AI spending had increased by 38%. Over the same period, the overall service and support function budget grew by 2%.

That spread is the whole story. A line growing at 38% inside a container growing at 2% is not new money. It is old money moving between columns, and the question worth asking is which column is losing.

Gartner answered part of that question itself. Kim Hedlin, a director analyst in Gartner's customer service and support practice, put it directly. "To fund ambitious AI initiatives, leaders are increasingly redirecting spending away from labor and overhead and instead toward technology," she said. She added that the challenge is ensuring those investments produce measurable business value.

The 15% technology line sits between the two

Coverage of the same survey by Customer Experience Dive adds two figures. Technology spending within the service budget rose 15%. On average, organisations now spend 13% of their total service and support budget on AI initiatives.

So there are three growth rates, nested inside each other. AI grew fastest at 38%. Total technology, which includes AI, grew 15%. The whole budget, which includes technology, labor and overhead, grew 2%. Each step down is slower, and the arithmetic of that nesting is what tells you the non-technology lines shrank.

What the release does not say

Gartner did not publish a line-item breakdown of what was cut. There is no disclosed split between in-house agent wages, outsourced contact center contracts, workforce management software, training or facilities. The labor and overhead attribution is an analyst's reading of the survey, not a measured ledger.

That gap is why the rest of this post leans on company filings. Survey data tells you what leaders say they did with their budgets. Outsourcer revenue tells you what a large slice of those budgets actually paid for, quarter by quarter, under audit.

The reallocation math inside a 2% service budget

The fastest way to see the squeeze is to rebuild an average budget from Gartner's reported figures. Treat last year's service budget as 100. This year it is 102. If AI is 13% of this year's budget, it is about 13.3 units.

A 38% increase means last year's AI line was about 9.6 units. AI therefore took about 3.7 new units, while the whole budget gained only 2. Everything that is not AI went from roughly 90.4 units to 88.7. That is a fall of about 1.8% in the non-AI part of the budget, before any other technology spending is counted.

AI grows 38%, the budget grows 2%, everything else shrinks Illustrative average service budget, last year indexed to 100. Built from Gartner's 38%, 2% and 13% figures. Last year Non-AI 90.4 AI 9.6 This year Non-AI 88.7 AI 13.3 Total: 100 to 102 (+2%) AI line: +3.7 units (+38%). Non-AI lines: -1.7 units (about -1.8%). The non-AI fall happens even if no other technology spending grows at all. Source: Zan Digital calculation from Gartner, August 2026, and Customer Experience Dive coverage of the 13% share.
Illustrative, not measured. The point is the direction: on Gartner's own averages, the money for AI cannot come from budget growth, so it comes out of the other lines.

Why the labor and overhead line has to shrink

The 15% technology figure tightens this further. Technology grew 15% overall. If technology was a fifth of last year's budget, it rose from 20 to 23 units. The non-technology lines, labor and overhead, then fell from 80 to 79.

The bigger technology's share, the harder the squeeze on people. Any technology share above about 13.3% of the budget forces labor and overhead to shrink in absolute terms, because 15% growth on that base already exceeds the 2-unit total increase.

What happens to labor and overhead under three technology shares
Technology share of last year's budgetLabor and overhead, last year to this yearChange in labor and overheadNon-AI technology change
20%80.0 to 79.0-1.3%-6.3%
30%70.0 to 67.5-3.6%+4.2%
40%60.0 to 56.0-6.7%+7.7%

Zan Digital scenarios. Inputs: total budget +2%, technology +15%, AI +38%, AI at 13% of this year's budget (Gartner, August 2026, and Customer Experience Dive). The technology share is assumed, because Gartner did not disclose it. Averages from a survey do not combine exactly; read the direction, not the decimal.

The 20% row carries a second finding. At a low technology share, AI growth is so large that it squeezes other software too. Workforce management, quality monitoring and legacy telephony tools would have to shrink alongside labor. That fits a pattern covered in the analysis of NICE and Five9 seat disclosures, where seat-based contact center software is being repriced as agent counts flatten.

Where the money comes from: labor, but slower than the hype

If labor is the source, the obvious check is headcount. Gartner ran that check in a separate survey of 275 service and support leaders in July and August 2026, published on 5 November. Only 21% said they had reduced human agent headcount because of AI.

Among the organisations that did cut, the average reduction was 16.7% of the agent workforce over the previous 18 months. That is a large cut where it happens. It is not happening at four in five organisations.

My reading is that the 38% AI increase is being funded faster than it is being earned. Leaders are committing technology money in anticipation of labor savings that most of them have not yet booked as layoffs.

Attrition is the quiet funding source

The cheaper route to a smaller labor line is not firing people. It is declining to replace them. Contact centers have historically run with high agent turnover, so a hiring pause shrinks payroll within months without a single announced layoff.

Gartner's November coverage notes that some leaders report AI has let them pause backfilling agent roles. That mechanism never shows up in a layoff count, which is why the 21% figure understates how much of the AI budget is really coming out of payroll. What a slimmer team then has to handle is examined in the piece on how support staffing changes after deflection.

Gartner itself expects part of this to reverse. In February 2026 it predicted that half of companies that cut customer service staff because of AI will rehire by 2027, often under different job titles. If that prediction holds, a slice of today's labor saving is a loan against next year's budget.

What contact center outsourcing filings add

A large share of service labor is not on the buyer's payroll at all. It sits inside outsourcing contracts with business process outsourcers such as Concentrix, TP and TTEC. When a service leader moves money from labor to AI, these companies feel it as lower volumes, shorter programs or slower new business.

Their filings are useful for a reason surveys are not. They report audited revenue every quarter, and they have every incentive to explain a decline in terms that protect their valuation.

Latest reported revenue at three contact center outsourcers
Company and periodRevenueChangeWhat the company said
Concentrix, fiscal Q3 2026 (to 31 Aug)$2,453.7M-1.2% reported, -0.5% constant currencyQ4 guide of $2.41B to $2.46B, a 3% to 5% constant currency decline
TP (Teleperformance), H1 2026EUR 4,883M-1.7% like-for-like, -4.5% reportedPlus 1.7% like-for-like excluding Trust and Safety; AI-powered solutions grew double digits
TTEC, Q2 2026$455.5M-11.3%Engage down 12.1% to $351.5M; blamed a small number of clients and delayed new business

Sources: Concentrix Form 8-K Exhibit 99.1, 29 September 2026; TP first half results release, 30 July 2026; TTEC Form 8-K Exhibit 99.1, 10 August 2026. Periods differ by company and are not directly comparable.

Concentrix: half the revenue is new since AI

Concentrix reported fiscal third-quarter revenue of $2,453.7 million, down 1.2% on the prior year. It also took a $1.05 billion non-cash goodwill impairment, which the company tied to its stock price and market capitalisation.

The more revealing line came from chief executive Chris Caldwell. He said 50% of revenue now comes from business won and deployed in the three years since the introduction of AI. He also described the company as "aggressively disrupting our own traditional business". That is an outsourcer telling you its legacy voice and chat work is shrinking, and that it is replacing it with AI-attached services.

Fourth-quarter guidance points the same way. Concentrix guided revenue to $2.41 billion to $2.46 billion, a constant currency decline of 3% to 5%. A company with half its revenue in growing new business still guiding down tells you the old book is falling faster.

TP: automation named as a cause of lower volumes

TP, formerly Teleperformance, reported first-half revenue of EUR 4,883 million, down 1.7% like-for-like. Excluding its Trust and Safety content moderation business, the group grew 1.7%.

The release is unusually direct about why. In the Americas it attributes the sharp Trust and Safety decline to "accelerated automation leading to lower volumes". That is the closest any of these filings comes to stating plainly that a client budget moved from people to software. TP also reported more than 1,100 AI projects running on its platform at 30 June, which it described as 2.3 times the prior-year level.

TP is also cutting its own cost base. It guided to EUR 150 million to 170 million of annual run-rate savings from an expanded efficiency plan. The buyer's reallocation becomes the supplier's restructuring.

TTEC is the sharpest decline and the weakest evidence. Its Engage segment fell 12.1% in the second quarter, but management blamed underperformance at a small number of clients and a delay in closing new business. It did not cite AI. I include it because the scale of the drop matters, not because the filing proves the mechanism.

Contact center outsourcers: latest revenue change by line Year-on-year change, company filings. Periods differ: TTEC Q2 2026, TP H1 2026, Concentrix fiscal Q3 and Q4 2026 guide. 0% TTEC Engage, Q2 -12.1% TTEC total, Q2 -11.3% Concentrix Q4 guide, cc -3% to -5% TP group, H1 LFL -1.7% Concentrix Q3, reported -1.2% TP ex Trust and Safety, H1 +1.7% Sources: TTEC and Concentrix Form 8-K Exhibit 99.1; TP first half 2026 results release.
Notice the only green bar. Strip out the line TP says automation is shrinking and the group grows. The money is not leaving outsourcing evenly; it is leaving specific programs.

Why outsourced volume is the first budget line to go

If you run a service budget, outsourced volume is the easiest place to find money for AI. An in-house layoff needs HR process, severance and internal communication. An outsourcing reduction is a change to a forecast you send a vendor.

That asymmetry explains why the outsourcer filings move before the layoff surveys do. Gartner's 21% figure counts organisations that cut their own agents. It does not count the leader who trimmed a vendor's monthly volume commitment by a tenth and moved the saving into an AI agent licence.

I think this is the single most underreported effect of the Gartner spread. The 38% increase is being paid for disproportionately by headcount that never appeared on the buyer's org chart. It shows up instead as the Engage and Trust and Safety declines above.

The outsourcers know this, which is why every one of them is repositioning as an AI services provider. Concentrix's 50% new-business figure and TP's 1,100 AI projects are the same message: the supplier wants to capture the AI line rather than lose the labor line. Whether a buyer should let the same vendor run both is a fair question, and the economics behind it are laid out in the comparison of support ticket costs, human against agent.

Where this argument is weakest

The reallocation story is tidy, and tidy stories deserve suspicion. Three weaknesses matter here.

Survey averages do not add up cleanly

The 38%, 2%, 15% and 13% figures are averages across respondents, and two of them come only from secondary coverage of Gartner's release. Averages of growth rates across organisations of different sizes do not combine into one consistent budget. My indexed example is a directional illustration, not a reconstruction of any real organisation's books.

The sample is also 199 leaders reporting their own spending. Self-reported AI budgets are prone to generous classification. A contact center software renewal that now includes an AI add-on can be counted as AI spend in full, which would inflate the 38% without any money actually moving away from labor.

Filings show correlation, not cause

Outsourcer revenue falls for many reasons. Clients move work offshore to cheaper locations, which reduces revenue per hour without reducing volume. Programs end. Currency moves. TTEC, the steepest decline in the table, did not attribute its drop to AI at all.

Only TP's release names automation directly, and only for one business line in one region. The strongest version of this argument would need buyer-side data showing outsourcing contracts cut and AI contracts signed in the same budget cycle. Nobody has published that, and I would not claim it is proven.

There is also a credible counter-case. Gartner predicted in March 2026 that over half of customer service organisations will double their technology spend by 2028 without an equivalent cut in talent. If that is right, today's squeeze is a timing effect, and budgets will grow to carry both lines.

How to test your own contact center AI budget

The Gartner spread is a market average. What matters to you is whether your own AI line is replacing cost or stacking on top of it. Four checks settle that.

First, split your AI spend into money that replaced an existing contract and money that was added. A rebranded renewal is not new AI investment, and counting it as one will make your reallocation look bigger than it is.

Second, trace where the funding actually came from. If the answer is "the outsourcing forecast," write down the volume you cut and the resolution rate your AI agent must hit to absorb it. Vendors price this very differently, as covered in the resolution cost comparison across Fin, Zendesk and HubSpot.

Third, measure the escalation load your remaining human agents now carry. Cutting volume at the front often concentrates the hardest contacts at the back. The failure points in that handoff are mapped in the analysis of AI escalation handoff failures.

Fourth, set a date to check outcomes against the baseline you recorded before the shift. If you cannot find a baseline, that absence is the finding. The broader evidence on which AI deployments return money is in the review of where measurable GenAI return has shown up.

Frequently asked questions

How much did customer service AI spending increase in 2026 according to Gartner?

Gartner surveyed 199 service and support leaders in April and May 2026 and reported, on 26 August, that AI spending had increased by 38%. Over the same period, overall service and support function budgets grew by just 2%. Secondary coverage adds that total technology spending rose 15% and that AI now takes about 13% of the average service budget.

Where is the money for customer service AI coming from?

Gartner director analyst Kim Hedlin said leaders are redirecting spending away from labor and overhead toward technology. On Gartner's own averages, the non-AI part of a typical service budget falls by roughly 1.8%. Outsourced contact center volume appears to be a major source, based on revenue declines at TTEC, Concentrix and parts of TP during 2026.

Are companies laying off customer service agents because of AI?

Some are, but fewer than headlines suggest. A Gartner survey of 275 leaders in July and August 2026 found 21% had reduced human agent headcount because of AI. Among those, the average cut was 16.7% over 18 months. Many others are pausing backfills when agents leave, which shrinks payroll without a recorded layoff.

Is AI hurting contact center outsourcing companies?

The filings point that way, with caveats. TP said accelerated automation lowered volumes in its Americas Trust and Safety business. Concentrix guided fourth-quarter revenue down 3% to 5% in constant currency while saying half its revenue comes from business won since AI arrived. TTEC's Engage segment fell 12.1%, though management blamed specific clients rather than AI.

What share of the customer service budget goes to AI?

Customer Experience Dive, reporting on Gartner's August 2026 survey, said organisations spend on average 13% of their total service and support budget on AI initiatives. That figure did not appear in the excerpts of Gartner's own release reviewed for this article, so treat it as reported secondary data. The share varies widely by organisation size and industry.

How should a service leader justify a contact center AI budget increase?

Show what the AI spend replaced, not just what it costs. Record the baseline volume, resolution rate and cost per contact before the change. Name the line the money came from, whether payroll, outsourcing or legacy software. Then set a fixed review date. Gartner's own analysts warn that hype alone does not make a technology the right answer to a business problem.

Where to start this quarter

Pull your last two budget cycles side by side and draw one arrow for every pound or dollar that moved into AI, showing where it came from. If more than half of the arrows start at an outsourcing contract, your vendor already knows your AI strategy better than your finance team does.

Then ask your outsourcing provider for one number: contact volume handled for you this quarter against the same quarter last year. Their revenue line is falling for a reason, and it is better that you read it in your own data than in their next filing.

Related analysis

For how the three main support agent platforms price the work that replaces outsourced volume, read the comparison of Agentforce, Fin and Breeze support agents.

References

  1. Gartner, Gartner survey finds AI spending by customer service leaders has surged by 38 percent despite overall service and support function budgets rising by just 2 percent, 26 August 2026. Used for the 38%, 2% and sample figures.
  2. Customer Experience Dive, CX workforce budgets take a hit as organizations increase AI spending, 10 September 2026. Used for the 15% technology and 13% budget share figures.
  3. DestinationCRM, Customer service AI spending surges amid budget stagnation, 25 September 2026. Used for the Kim Hedlin quotes.
  4. Gartner, Gartner survey finds 21% of service and support leaders have reduced human agent headcount because of AI, 5 November 2026. Used for headcount figures.
  5. Gartner, Gartner predicts half of companies that cut customer service staff due to AI will rehire by 2027, 3 February 2026.
  6. Concentrix, Fiscal third quarter 2026 results, Form 8-K Exhibit 99.1, 29 September 2026.
  7. TP, 2026 first half results, Business Wire release republished 30 July 2026.
  8. TTEC Holdings, Second quarter 2026 financial results, Form 8-K Exhibit 99.1, 10 August 2026.

Weakest point in the source base: the 15% technology and 13% AI share figures come from trade press coverage rather than Gartner's own text. None of the outsourcer filings quantify how much of their decline is caused by client AI adoption. Figures last verified 11 November 2026.

SK
Sanskriti Khandelwal
Founding Member, Zan Digital. Writes about AI product economics, B2B software markets and what the numbers behind vendor claims actually say.

Related reading