AI is reshaping the economics of IT services at two different speeds. In the near term, it is making traditional service delivery faster and less labor-intensive, putting pressure on established revenue models. Over time, AI is also expected to create demand for new technology services, expanding the market. We do not foresee a “service apocalypse”. For IT service providers, the challenge is that efficiency gains are materializing faster than new revenue opportunities.
The first effect is already visible. Projects can be delivered by smaller teams and in less time, creating immediate pressure on service providers that still earn revenue mainly by selling people and billable days. The second effect will take longer. Many new projects must be identified, funded, integrated with existing systems, and moved into production before they generate meaningful services revenue.
This gap in timing explains much of the current uncertainty around the sector. Providers can face slower growth and pricing pressure today, even as AI creates a larger market over the medium term.
Our central case estimates that the IT-services market will grow from around $1.2 trillion in 2025 to approximately $1.6 trillion by 2030, at about 5.5% per year. The analysis draws on more than 30 interviews with executives from IT services firms, hyperscalers, enterprise software companies, and frontier-model providers, as well as an Oliver Wyman market model.
AI could drive IT services growth after an initial reset
Four forces determine how AI affects the market: effort per project, project volume, value per day, and the share of work bought externally.
First, AI shortens many parts of delivery, especially repetitive tasks, and executives say they expect average reductions in effort per project of between 20% and 25% over the next three to five years. For providers paid by the day or by full-time equivalent, that productivity translates into a smaller invoice unless the contract changes.
Next, the same productivity shock also expands demand. Cheaper, faster delivery can revive projects that failed the return-on-investment test. It can also create additional work around the foundations needed to operate AI at scale.
The compression of business from the first impact of AI is immediate, but the new demand from the second impact takes longer to fund and convert into recurring revenue. The result is a J-shaped growth path, with pressure first and stronger growth as the new work matures.
The other two AI impacts — value per day and the share of work bought externally — determine which providers capture the growth. Revenue per day can rise as clients buy more scarce expertise and ask providers to take greater responsibility for delivery. Outsourcing shifts too, with clients relying less on generic staffing, but still needing partners to integrate systems, manage risk, and run complex environments. Eight models can win, and all rest on the same five enablers: a workforce redefined role by role, end-to-end workflow offers, productized IP, an AI-native operating model, and pricing that keeps the productivity gain.
AI will drive IT services growth through new project volume
The increase in project volumes is driven by four distinct market drivers — each with its own trigger and revenue characteristics. One we have already discussed: When projects can be achieved at lower costs and faster, some that were shelved for being too expensive are brought back to life.
AI also creates new demand by allowing business functions to redesign specialized workflows or automate tasks that were previously considered too complex, too small, or too expensive for a conventional technology project. This pool is driven by new capability rather than by lower cost alone.
AI creates new demand by making legacy modernization and stronger data foundations urgent. For instance, scaling AI use cases cannot be achieved with poor-quality data or disconnected legacy systems. Once a company decides to move an AI initiative into production, it must also invest in the data, integration, and system upgrades needed to support it. Modernization that was previously postponed, therefore, becomes a necessary part of the AI program.
The real IT services opportunity goes beyond AI pilots
Finally, production AI creates recurring orchestration and governance needs, such as the operating challenge of moving from a pilot to live agents. Enterprises need rules, monitoring, and clear human accountability, creating recurring work after deployment.
These four pools of new demand reinforce one another and demonstrate how less effort per project can still produce a larger market. They also explain why sector maturity is not a simple measure of market attractiveness. More advanced sectors are likely to spend more on AI-native use cases and the orchestration of systems already in production. Less mature sectors may generate equally significant demand through deferred modernization and foundational data work. Maturity, therefore, changes the mix and timing of services demand, rather than determining how large the opportunity will be.
IT services pricing models are changing in the age of AI
While we expect several established service lines to contract, the market should grow overall. Growth favors work requiring architecture, trusted data, integration and accountability; activities whose value still depends mainly on repeatable human effort face more pressure. Many providers have revenue on both sides of that divide.
Agentic services illustrate the shift. It has been estimated that they could make up close to one-quarter of the 2030 market (about $400 billion), representing the share of existing data, application, security, and operations work designed and bought around agents rather than a new segment on top of the old one.
Meanwhile, the commercial model is evolving alongside the service portfolio, with the contract determining which party benefits from the resulting productivity. Under time-and-materials, a 25% reduction in effort can result in a 25% smaller invoice, provided fees remain tied to the number of people and days used. Fixed-price packages and managed services weaken that direct link because the client buys an agreed scope or service level. Reusable assets can add a separate fee, allowing providers to capture value beyond billable hours.
This is why staff augmentation is expected to lose share, from around 28% in 2025 to roughly 20% in 2030, shrinking in absolute terms. Specialist talent will remain valuable, but clients will be less willing to fund large, open-ended teams when AI reduces the effort required. They will increasingly buy defined squads, work packages and managed services in which the provider is accountable for deliverables, rather than paying mainly for headcount and hours.
Who will win the next battle for enterprise AI services
Newcomers such as hyperscalers, software companies, data platforms, and frontier-model providers are moving closer to enterprise deployment; frontier labs put about $5.5 billion into deployment ventures in May 2026 alone. They have strong incentives to influence architecture, drive adoption, and secure the most valuable early use cases.
Yet their reach has limits. Enterprise deployment still involves legacy environments, local regulation, process change, and long-term operational responsibility. Enterprises are also keeping control of their AI architectures and running several models side by side. Arbitrating, integrating, and governing that mix is a role vendors cannot credibly fill for their competitors' products, leaving a substantial role for service providers.
The relationship will therefore combine competition and partnership. Providers will need deep capability in a small number of important ecosystems, supported by certified talent, production references, and co-sell activity. A long list of partnership badges will carry less weight than evidence that the relationship generates client access and repeatable delivery work.
Domain expertise makes that position defensible. Reliable AI depends on understanding the client's workflow and the information required at each step, and the decisions that still need human judgment. Providers who know the process can deploy faster and create more lasting value.
IT services firms need a new strategy to win the AI race
Enterprise adoption may take years, but future market shares are being fought over now. Already, technology companies are selecting preferred partners, large clients are consolidating supplier panels, and providers are looking to build deployable talent and references. Once these positions are set, late entrants can face a multiyear disadvantage as they try to catch up. IT services firms cannot wait for new demand to arrive.
For management teams, four priorities follow:
- Map the revenue exposed to delivery compression, contract by contract.
- Choose a limited number of replacement markets where the firm already has a credible right to win.
- Redesign pricing and delivery models so productivity supports margin and reinvestment.
- Build the talent, reusable assets, and ecosystem needed before demand accelerates.
The full report examines eight winning provider archetypes, detailed growth forecasts by service line, and five measurable signals to track by mid-2027. It also explains why the next 18 to 24 months will determine which IT services providers benefit from AI and which fall victim.
AI will not create a simple rising tide that will lift all IT service providers. Firms that move early can turn productivity into capacity, stronger client positions, and new revenue. Those who protect the old model for too long may find that the most valuable plays in the next cycle have already been made.