Through 2026, share prices across the major IT services firms have fallen sharply, including at firms reporting revenue growth and raising guidance. The pattern reads as a repricing of revenue built on billable people, applied across the sector regardless of how well an individual firm is executing.
This article sets out what the reported numbers show as of August 2026, what the market is repricing, why integration is the service line where the shift is sharpest, and what the practical alternatives look like for firms that sell integration work and for the enterprises that buy it.
Key takeaways
- Firms delivering growth have been marked down alongside those that are not, which points at the revenue model itself.
- ISG recorded the combined tech services market growing 43% year over year in Q2 2026, with cloud and infrastructure up 65% and managed services, the people-based category, up 2.7%.
- Accenture's Managed Services revenue passed its Consulting revenue in the quarter ended 31 May 2026, growing at five times the local-currency rate.
- No major system integrator reports integration as a revenue line. It is treated as a task inside projects, and its value keeps accruing to whoever operates it afterwards.
- When an integration fails, the cost never stops at IT hours: the business process the integration carries stops with it.
For the operating model behind managed integration services, the Integration Ops book covers lifecycle phases, ownership patterns, and playbook examples in depth. Download it free.
The repricing of IT services is the market shift now valuing revenue from operated, outcome-based services above revenue from billable hours. Firms are being marked down not for weak delivery but for how their revenue is produced: headcount-linked income carries a lower multiple than recurring, service-based income, even when the underlying business is growing.
What the 2026 numbers show
Share price movements across the sector over the twelve months to mid-August 2026, alongside the most recently reported growth figures.
Two entries in that table do most of the work. Capgemini grew 11.3% at constant currency in the first half of 2026 and raised its full-year outlook, and its shares are down roughly a third over the year. Accenture reported a beat on revenue with operating margin up 20 basis points in its third quarter, and the shares fell 20% on 18 June 2026, the largest single-day decline in the company's history.
A market punishing weak performance does not produce that pattern. Something else is being priced.
Figures are dated because they move. Verify current prices before citing them, and read this section as an analysis of a business model, not as a view on any security.
What is being repriced is the unit of sale
For three decades, revenue in IT services has been a function of billable people: more consultants, higher utilisation, lower-cost delivery locations. Every growth lever ran through headcount.
Where new money went in 2026 shows how that is being valued now. ISG's index for Q2 2026 recorded combined tech services annual contract value up 43% year over year to $42.4 billion. Cloud and infrastructure services grew 65%, with a substantial share of the increase attributed to AI demand. Managed services grew 2.7%.
Gartner's July 2026 forecast reaches the same conclusion from the buyer's side of the ledger, with worldwide IT spending forecast to grow 14.2% in 2026 while IT services grows 5.3% and data centre systems grows 62.5%. The market grew at record pace, and very little of that growth landed in categories where the unit of sale is a person.
Accenture has already crossed over
The clearest evidence that the industry is adapting comes from its largest firm. In Accenture's third quarter of fiscal 2026, the quarter ended 31 May 2026, Managed Services revenue reached $9.39 billion against Consulting at $9.33 billion. Managed services grew 5% in local currency, consulting 1%. Company guidance indicated consulting in low single digits and managed services continuing at mid single digits, with a stated direction toward commercial models not tied to full-time-equivalent headcount.
For the first time, the largest system integrator in the world earns more from operating things than from advising on them, and the operating side is growing five times faster. Consulting remains roughly half the business; the centre of gravity has shifted.
The shift from project delivery to operated service is visible in the incumbent's own reported segment mix.
The AI discount is priced in before it is delivered
A second pressure runs underneath the first, and it removes the option of waiting for the first one to pass. HFS Research reported in July 2026 that 40% of organisations expect AI-led services to cost 10% to 30% less, while only 19% have redesigned how they buy AI-delivered services. Buyers have priced the productivity gain into their expectations, and the contracts they sign with their service firms are still written in hours.
Inside a time-and-materials contract this produces a closed loop. Delivering the work more efficiently reduces billable hours, which reduces revenue, and the saving passes to the client because that is what the contract measures. Improving productivity becomes a revenue reduction exercise. ISG's commentary on the same period names pricing deflation directly and describes labour-intensive work being displaced by large language models.
The only exit from that loop is changing what is sold.
Why service integration is the sharpest case
Integration is the service line where the mismatch between the project model and where value accrues is most extreme. Three properties combine to produce that.
It is not reported anywhere. No major system integrator breaks out integration as a revenue line. It sits inside technology services, applications, or consulting. That accounting reflects how it is treated internally, as a task performed within projects, with no lifecycle and no named owner once the project closes.
Its value accrues after delivery. An integration produces almost nothing on the day it passes acceptance testing. The value comes from the years it keeps working through platform upgrades, API version changes, partner migrations and acquisitions. A model that recognises revenue at go-live gets paid before that value exists, then hands the remaining obligation to the customer.
Its build work is pattern-regular. Field mappings, transformations, error handling and connector code are among the most automatable categories of billable engineering. Where AI compresses billable hours, it compresses them here first.
There is one more thing the accounting never shows. When an integration fails, the failure does not stay in IT. The business process the integration carries stops: orders stop flowing, tickets stop syncing, customers and employees wait, and SLA penalties start accruing while an IT team firefights. The cost shows up twice, in the hours IT spends restoring the connection and in what the organisation cannot do while it is down.
Put together, integration is a service line where the traditional model bills for the least valuable part, hands over the most valuable part, and is losing the billable hours attached to the part it does bill for. Our analysis of when the project model breaks for integration work covers the buyer's side of that argument, and the true cost of enterprise integrations covers the economics across delivery models.
What selling an operated integration outcome requires
The standard recommendation is that firms should sell outcomes instead of effort. The recommendation is right. Everything difficult is in the mechanics.
A firm offering an operated integration outcome has to settle four contractual questions, none of which are strategic, and all of which have to be answered before a single client conversation.
Two routes answer them. Build the capability, which means standing up 24/7 operations, monitoring, tooling and a service catalogue for a category currently treated as a task inside projects. Or partner with an operator who already runs one, and keep the client relationship.
The partner route
For firms that already own strong client relationships, the partner route is the faster of the two, and it is already in use. ONEiO operates inside partner ecosystems on exactly this basis, including its partnership with CGI for ServiceNow integration delivery, where the integration operations sit with the specialist and the client relationship stays with the integrator.
The commercial shape that works has three properties. The service level sits with the party that operates the connections, because accountability without operational control is unenforceable. The client relationship, advisory work and endpoint expertise stay with the integrator, since that is where its advantage genuinely is. And the revenue is recurring for both parties, which is the property the market is currently rewarding. That last point is why these arrangements hold: each side keeps a durable, recurring position it would be expensive to rebuild alone.
Our overview of integration methods, platforms and services for system integrator firms covers the delivery options in more detail.
If you are the enterprise buying this
The same analysis has a shorter reading from the customer side. Your integrator is under margin pressure that has nothing to do with your account, and the pressure runs specifically against long-tail maintenance work on integrations delivered years ago. That work is low-margin, hard to staff, and increasingly hard to justify inside a firm being valued on recurring revenue.
For you, the exposure is concrete. When one of those ageing integrations fails, the process it carries stops with it. Orders, tickets or invoice data stop moving, customers and employees wait, and the SLA clock runs while someone who still understands the connection is found.
Many integrators are excellent, and most enterprises should keep using them for transformation work, which is what they are genuinely best at. The question worth asking is who operates each of your integrations three years after the project closed, and whether that arrangement has a service level attached to it. Our comparison of integration service providers sets out the models.
When the project model is still the right answer
Project-shaped delivery remains the correct choice for genuinely finite work: a one-time data migration, a decommissioning, a discrete transformation programme with a defined end, or deep bespoke work on a system nobody else understands. System integrators built the enterprise integration landscape and remain the right partner for that class of engagement. The mismatch appears where the work is continuous, where the estate keeps changing, and where value depends on something still functioning years after the invoice cleared.
Bottom line
The 2026 share price decline across IT services has been applied to firms that are growing and firms that are not, which points at the revenue model itself. Revenue built on billable people is being valued below revenue built on operated outcomes, and the largest firm in the sector now earns more from managed services than from consulting.
Integration sits at the sharp end of that shift. It is unreported as a revenue line, its value accrues after delivery, and its build work is the most automatable thing on the invoice.
The repricing has already happened. The response that fits the evidence is changing the unit of sale in the service lines where value accrues after go-live: build an operations capability, or partner with one and keep the client relationship. Cost discipline and an AI practice attached to an unchanged contract address neither pressure.
Not sure where your integration layer stands? The Integration Ops maturity assessment gives you a structured read on ownership, monitoring, and change readiness across your estate.
