The IT function has a new job.

For two decades, the IT function was measured by how well it kept technology running. Stability, cost discipline, predictable service levels. Those criteria still matter. They are no longer sufficient.

The primary job of the IT function is no longer to run technology. It is to increase the rate at which the business absorbs AI into recurring value.

The productivity gains created by AI only become repeatable business capabilities once the IT function makes them secure, integrated, and scalable. Those capabilities are what the business can eventually convert into recurring EBITDA. Without that translation layer, every AI gain stays trapped at the individual level. With it, the business gains a capability it can compound.

The outsourcing model did not become obsolete because vendors failed. It became obsolete because the economics of the IT function changed.

Why outsourcing economics have changed.

For twenty years, outsourcing IT was often the right decision for mid-cap portfolio companies. Three conditions made that decision correct: scarce IT talent, peripheral IT skills, and a cost the board could read in one P&L line.

The first condition has shifted. AI-assisted operations have compressed the productivity gap between a generalist IT team and a specialist outsourcer.

The second condition has shifted further. IT skills are no longer peripheral because the IT function is now responsible for turning isolated AI gains into repeatable business capabilities.

The third condition has shifted too. What the IT function spends is no longer the relevant number because buyers increasingly price what that function enables the rest of the organization to capture.

The board is no longer solving the same problem.

What the IT function actually does now.

The new job has three layers.

At the level of IT operations, the function manages the cadence at which the system is recomposed. Operational runbooks, automation workflows, monitoring rules. What used to be revised every two years is now being revised every few weeks because the vendors underneath ship AI capabilities continuously.

At the level of technology capabilities, the function owns the integrations between the company’s systems and the AI services those systems consume. APIs, integrations, and data pipelines. Those integrations determine whether a local AI use case can become an enterprise capability.

At the level of vendor governance, the function renegotiates licensing models that were signed before AI consumption existed. Per-token and per-agent pricing were not part of contracts written on per-user or per-server logic. Contracts that have not been revisited often leave AI value uncaptured.

A function that performs these three jobs consistently gives the business the capacity to scale AI. A function that does not slows that capacity down.

What the board should ask the outsourcer.

A board reviewing an IT outsourcing contract today should be able to answer three questions.

How frequently are operational runbooks and automation workflows being updated?

To what extent has the outsourcer embedded AI into its own operating model?

How have commercial terms evolved to reflect AI consumption, AI security, and AI governance?

The answers reveal whether the outsourcer is still optimizing technology operations or helping the company increase its AI absorption capacity.

What the buyer will see at exit.

Strategic acquirers are expanding, not replacing, the way they diligence IT. Stability and cost still matter. Increasingly, they also want to understand whether the IT function can continuously transform AI investments into operating improvements that the business can scale.

When the IT function cannot do that, buyers see more than a technology issue. They see execution risk, delayed value creation, and additional post-deal investment. Those risks are reflected in valuation.

The shift in the board’s question.

For twenty years, the right board question was: are we paying a fair price for IT stability and predictability?

Today the better question is: does our IT operating model maximize the organization’s ability to convert AI into recurring enterprise value?

These are different questions producing different sourcing decisions.

Every decision about IT sourcing is now a decision about the organization’s capacity to convert AI into recurring enterprise value.