AI and IT Services: The New Rules of Financeability
| A market perspective from the capital advisory desk
Over the past months, one question has regularly arisen in conversations with banks, private debt funds and private equity investors: How is artificial intelligence changing the financeability of IT services businesses? This question no longer comes only from the credit side. Sponsors increasingly ask how lenders might think about the sector, often before they commit to a platform. They want to know how durable the financing of a given business model really is and whether the margin and cash flow assumptions of the last decade still hold. Somewhat surprisingly, the market view is not that IT services has become non-financeable. The view among financiers has simply become much more differentiated. That differentiation is the real story of the current market. |
Summary
- Lincoln International’s experts answer: How is artificial intelligence changing the financeability of IT services businesses?
- Sign up to receive Lincoln's perspectives
Why IT services became a sponsor and lender favorite
It is worth recalling why IT services, and software and managed services in particular, became one of the most attractive sectors in middle market private equity.
IT services is a broad sector, covering businesses that provide the technology infrastructure, software and support that other companies rely on to operate. In practice, the sector spans several business models: managed service providers (MSPs), which run a client’s IT operations on an ongoing contracted basis; software and systems integrators, which implement and customize platforms such as ERP or CRM systems; and specialist providers focused on areas such as cybersecurity, cloud hosting or data management. These businesses offered recurring or repeatable revenue, low capital intensity, high cash conversion and limited working capital. The market was also highly fragmented, with many founder-owned and succession-driven targets. That made it close to an ideal buy-and-build market.
This was especially true in the German Mittelstand. Many smaller IT and software services businesses are simply too small for larger strategic buyers. Sponsors could build platforms, add scale through acquisitions and benefit from the valuation arbitrage between small single assets and larger platform-grade businesses.
During the post-COVID digitalization boom, valuations rose significantly. They normalized again in 2022, when rising rates and tighter acquisition financing cooled the market. Throughout that cycle, scaled platform assets traded at a clear premium, while smaller single assets traded at a discount. This spread supported the buy-and-build model and attracted significant sponsor capital.
For lenders, these were attractive credits: predictable revenue, diversified customers, strong cash conversion and a reliable exit path through secondary and strategic demand. This is why so many sponsors and debt funds built dedicated investment theses for the sector.
The financing logic behind the thesis
The earlier credit case relied on a few metrics. Taken together, they justified more leverage than a purely project-based services business would normally support.
Lenders looked at the share and stickiness of recurring revenue, net revenue retention, gross margin and cash conversion. They were comfortable where revenue was visible, customers stayed, margins were good and EBITDA converted well into cash.
A managed-services platform with a high recurring-revenue share, net revenue retention above 100%, low churn and disciplined working capital could support attractive leverage in the unitranche market, often with light covenants and meaningful headroom. For the strongest credits, even covenant-lite was achievable.
Lenders and sponsors made two key assumptions: First, contracted recurring revenue made historical EBITDA a reasonable proxy for future cash flows. In other words, last year’s EBITDA was a credible starting point for next year’s debt service.
Second, investors assumed that the human work behind the service was a cost item, not the product itself. The business was expected to scale like a platform: more revenue, better margins and a Rule-of-40 profile that will persist.
Importantly, if a business requires more people every time its revenue grows, it looks like a traditional services company. If it grows faster than its headcount and improves margins at scale, it looks like a platform. AI now puts pressure on exactly this point. That is why the financing discussion has shifted.
Where the current uncertainty comes from
The concern is not that AI makes IT services obsolete. The concern is who captures the productivity gains and therefore how much historical EBITDA is truly defensible.
A company with a billable-hours model, for example, may experience a real efficiency gain by AI in code generation, testing or first-level support, but that does not automatically benefit the provider. However, if the client expects now fewer billed hours or lower budgets, the gain is passed on through price pressure. That is a direct challenge to the EBITDA base of the company, which historically looked defensible.
This is now one of the first questions in credit discussions with lenders: If AI makes delivery faster, does the IT services provider keep the margin, or does the customer take it through lower pricing?
Three broad groups are emerging as a pattern, and the framework will be tested against real transactions in the coming quarters.
Group 1:Under pressure. |
The most exposed business models are built around billable hours and high headcount intensity: classic body leasing and staff augmentation, offshore and nearshore developer pools, commodity coding without real domain or architecture depth, manual testing and quality assurance and first- and second-line support.
Large global IT services groups have meaningful exposure to these models, even though their overall businesses are far broader. The point is not that these companies are weak. They simply illustrate the delivery model lenders now scrutinize more closely. Then the obvious question is: if AI cuts the hours needed for a task, who keeps the benefit? If the provider does the same work with fewer people and holds pricing, margins improve. If the client expects lower budgets or fewer hours, the benefit flows to the customer. In that case, historical EBITDA is a weaker guide to future performance. For lenders, diligence will likely sharpen around pricing power, utilization, customer concentration, retention and cash conversion. I would expect more conservative leverage, more caution on add-backs tighter credit terms and a generally closer look at whether the business is more than a well-run capacity platform. |
Group 2:Mixed. |
Another part of the market sits in the middle: implementation and transformation partners around the major enterprise software platforms, like SAP or Sage; cloud migration and operations providers; and data engineering and business intelligence specialists.
These businesses are not automatically at risk. Their value often comes from process knowledge, customer intimacy, system integration, sector expertise and change management. A provider who understands a client’s ERP landscape, finance processes, supply chain, data environment and regulatory requirements is not, and will not, be easily replaced by an AI tool. But the equity story now needs to be far more precise. A growing IT services player in an attractive market is no longer enough on its own. For lenders, their due diligence focus will turn on the following questions:
If the answers are convincing, these businesses stay financeable on attractive terms. If they are not, the business may still be bankable, but at lower leverage and with more structure, such as cash sweeps, tighter covenants, larger equity checks and a heavier diligence burden. |
Group 3:Advantaged. |
The most interesting credits use AI not only as an internal efficiency tool but as the basis for new, repeatable and scalable offerings.
Examples include AI-enabled managed services, automated software modernization and data platforms, as well as more specialized offerings in areas such as cybersecurity and process automation. What these have in common, and what makes them interesting from a financing perspective, is that they are built on proprietary software tools or reusable IP, which is what turns a one-off project into a repeatable, scalable revenue stream Another area is machine learning operations (MLOps). It covers the tooling and services needed to run AI models reliably and compliantly over time: deploying them, monitoring whether they still perform and retraining them when the data shifts. From a financing perspective, this is interesting because it creates recurring managed-service revenue rather than one-off project income. In these cases, AI improves both sides of the P&L. It lowers the cost of projects and creates new revenue. In recent conversations with lenders, these businesses attract clearly more interest. They tend to show higher gross margins, more recurring revenue, stronger net revenue retention and clearer differentiation. But the AI story has to be backed by hard numbers and KPIs, not only positioning. Lenders want to see real gross margin improvement, the recurring revenue mix, low churn and evidence of ROI for the customer. |
What this means for financing
AI does not make IT services harder to finance across the board, but it widens the gap between strong and weak business models.
Financiers will underwrite less on the broad IT services label and more on the specific AI exposure of each business. Is the service substitutable? Does AI only make projects cheaper? Or does it build a more scalable, more differentiated and more resilient business?
The metrics that were always important still are: quality of recurring revenue, net revenue retention, gross margin and cash conversion. But they are now being stress-tested against a faster substitution risk.
For lenders and sponsors, diligence has to go one level deeper. It is no longer enough to ask whether a business is growing, profitable and in an attractive market. The harder questions are which parts of delivery AI can automate, who captures the gain, whether pricing is linked to hours or to outcomes, whether stickiness goes beyond people and capacity and whether margins improve from real scalability or only temporary efficiency.
These questions will become a much larger part of credit committee and investment committee discussions.
