Infrastructure Outsourcing Services

In 2009, Forrester described the smarter buyers in this market as those looking at infrastructure outsourcing as a way to deliver near-term cost savings while positioning IT as a business enabler rather than a keep-the-lights-on tax on the firm.

In 2026, Forrester's description of what separates Leaders in this market includes positioning infrastructure outsourcing as an enabler of business transformation rather than just technology support.

Seventeen years apart, and the same sentence. The industry has been promising to move beyond cost arbitrage since before the iPad existed, and an analyst firm is still using that transition as the marker of leadership.

That persistence is the most honest thing to know about this category. The promise is real, some providers deliver it, and the gravitational pull toward cost has never weakened.

What is actually being bought

Infrastructure outsourcing hands responsibility for running technology infrastructure to a third party. Compute, storage, networking, data centre operations, end-user computing, service desk, and the operational disciplines around them: monitoring, patching, incident response, capacity management, and disaster recovery.

Forrester's current Landscape framing describes the value as consistent, reliable, scalable infrastructure management, the ability to address regional and local requirements, and support for experimenting with and adopting new technologies.

The middle item deserves attention because it is the one organisations underweight. A multinational running operations across a dozen countries faces different regulatory requirements, different data residency rules, different labour arrangements, and different local suppliers in each. A provider that already operates in all of them is selling coverage that would take years to build internally, and that is frequently worth more than the hourly rate difference.

Seventeen years of renaming

Forrester has scored this market repeatedly and changed its name four times, which tracks how the industry wanted to be seen at each point.

The Q1 2009 evaluation was Global IT Infrastructure Outsourcing, authored by Paul Roehrig, covering fifteen firms against thirty one criteria plus a twenty five item reference client survey. Cognizant entered the Leaders for the first time.

The Q1 2011 edition reviewed nineteen providers against thirty six criteria, with Tata Consultancy Services placing as a Leader and taking the highest customer reference scores of any vendor evaluated.

The Q1 2015 edition dropped IT from the name, becoming Global Infrastructure Outsourcing. Capgemini placed as a Leader, with its own commentary emphasising cloud brokering, service integration, orchestration, and aggregation.

The Q4 2017 edition was titled Next-Generation Infrastructure Outsourcing, scoring fifteen providers against twenty four criteria: Accenture, Atos, Capgemini, CGI, Cognizant, DXC Technology, Fujitsu, HCLTech, IBM, Infosys, NTT DATA, TCS, Tech Mahindra, Unisys, and Wipro.

Next-generation is a revealing choice of adjective. It signals a market whose established version had become unattractive enough that the research needed to distinguish the modern kind.

The Q4 2024 edition settled on Infrastructure Outsourcing Services, covering thirteen providers, and the Q3 2026 edition published on 28 July 2026 under the same name, following a Landscape in Q1 2026.

Nineteen providers in 2011, fifteen in 2017, thirteen in 2024. Steady contraction among firms that were already very large.

What distinguishes Leaders in 2026

Forrester's characterisation of the current Leaders describes superior current offering scores across compute, storage, networking, data centre, resiliency, and advanced operations, with strong strategy scores covering vision, sustained innovation, partner ecosystems, pricing flexibility, talent strategy, and global delivery.

Three themes in the current evaluation are worth separating out.

AI-first operations, meaning zero-touch operations and predictive automation applied to infrastructure management rather than sold as a client-facing capability. This is the provider automating its own delivery.

Digital twins augmenting observability, allowing scenario modelling and anticipation of operational outcomes rather than reacting to them. Simulating a change before applying it to a production estate is a genuinely useful capability and a difficult one.

And sustainability, with data centre services assessed on energy optimisation. That has moved from a reporting requirement to a scored capability, which reflects both cost pressure and regulatory disclosure obligations.

The pattern across all three is the provider industrialising its own operations. Which brings the pricing question directly into view.

Why cost keeps winning

The transformation promise has been available since 2009 and cost remains the dominant purchase driver. The reason is structural rather than a failure of imagination.

Cost savings are contractible. You can write a number in an agreement, measure it, and hold someone to it. Transformation is not contractible in the same way. It depends on the client's willingness to change how it works, on decisions made outside the provider's control, and on outcomes attributable to many causes.

Procurement functions are measured on the first. Nobody gets promoted for signing a contract whose benefits are described as strategic and realised over five years.

There is a second reason that providers do not raise. A transformation that genuinely modernises a client's estate reduces the operational effort required to run it, which reduces the revenue from running it. A provider paid per unit of effort has no commercial reason to eliminate effort quickly.

That conflict is the same one visible in Forrester's coverage of application modernisation and multicloud managed services, where customers worried explicitly that an end-to-end provider would optimise on the run side what it should have optimised on the build side. It applies here in the same form.

The organisations that resolve it write the modernisation outcome into the commercial model, with committed cost reductions tied to the estate actually simplifying. The ones that do not discover that year three of a seven-year contract looks a great deal like year one.

What cloud did, and did not, do

A reasonable question is why this category survived the cloud transition at all. If infrastructure moves to hyperscale providers, the argument goes, there is nothing left to outsource.

That prediction was wrong in a specific and instructive way.

Cloud changed what infrastructure is without reducing how much of it there is. An enterprise running three cloud platforms, a private cloud, remaining data centres, edge locations, and a set of systems nobody will migrate has a more complex operational estate than one running two data centres, not a simpler one.

It also changed the skill requirement rather than removing it. Managing cloud infrastructure well requires cost management, security posture, identity, network architecture, and automation expertise that is scarce and expensive, which is precisely the condition under which outsourcing makes sense.

What cloud did remove is the asset transfer that used to anchor these deals. Historically an outsourcing contract involved the provider buying the client's data centres and hiring its staff, which created switching costs so high that the relationship was effectively permanent. Without that anchor, contracts are shorter, more contestable, and more frequently multi-sourced.

That is better for buyers and it explains the contraction in provider count, since a market with lower switching costs rewards scale and punishes the mid-sized.

The arbitrage question returns

The labour model underneath this industry is facing the same pressure visible across every services category in this series.

Infrastructure outsourcing was built on wage arbitrage plus process discipline: do the work somewhere cheaper, standardise it, and apply economies of scale across many clients. That model created several very large companies.

Automation has been eroding it for a decade and agentic operations accelerate that. If monitoring, first-line triage, patching, and routine remediation are increasingly performed by software, the headcount that the pricing model rests on shrinks.

The Leaders being credited for zero-touch operations and predictive automation are doing exactly this to themselves, which is the correct strategic response and produces an obvious commercial question for buyers: if the provider needs fewer people to deliver the service, does the price reflect that, or does the margin?

That question is negotiable in a way it was not five years ago, because the provider has publicly claimed the productivity gain in an analyst evaluation. A vendor citing its automation maturity has supplied the evidence for the conversation.

What the reference process reveals

One detail worth noting about how these evaluations work, because it affects how much weight to place on them.

Forrester's methodology draws on customer references supplied by the vendors themselves, and participation varies. In the current evaluation, at least one provider's assessment notes an unusually thin reference response.

That is worth reading as information rather than as an accusation. A large provider unable to surface engaged reference customers for an analyst evaluation is telling you something about the state of its client relationships, and it is the kind of signal that does not appear in a capability score.

The broader point for any services evaluation is that references are the most useful and least standardised input. Ask providers for references matching your profile in scale, geography, and estate complexity, and treat a provider's reluctance as data.

Alongside that, the common thread in customer feedback across this category's history is governance. Reference customers repeatedly emphasise the need to actively manage the relationship, set expectations, and define governance models early. That is the least glamorous advice available and it remains the most consistent predictor of whether these arrangements work.

Forrester does not endorse any provider named here, and tier placement should not be read as a recommendation to buy.