CONTRIBUTORS
Anant Kumar
Managing Director, Head of US Research
Benefit Street Partners
AI disruption in software has quickly become one of the biggest concerns in private credit. But the issue is not simply how much exposure a manager has; it is whether lenders genuinely under-stand how durable the businesses behind their loans are.
Not all software businesses are the same. Some will adapt and strengthen as AI adoption accelerates, while others may see their competitive advantages erode far faster than markets currently appreciate.
This is where AI pressures are emerging first, but the broader challenge is understanding how AI will reshape knowledge-work industries. For private credit managers, this is no longer just a technology debate, but a core underwriting issue.
Software is not one sector
Markets have recently treated software through a binary lens – AI winners versus AI losers – leading to frequent indiscriminate selloffs. But what private credit investors really need to under-stand is whether a company possesses a genuine moat that can withstand AI-driven disruption over the life of a loan – however long that may be.
Recent reactions in public markets illustrate the problem. Insurance brokers sold off sharply after Anthropic released new AI tools aimed at improving insurance workflows, before rebounding as investors reassessed the likely impact on the industry. The episode highlighted how quickly markets can move from assuming wholesale disruption to recognising that many workflows, customer relationships and compliance structures are harder to displace than initial volatility suggests.
In software, the businesses facing the greatest pressure are typically narrow, single-purpose software tools – think expense management, contract review, or data entry automation – products performing highly repeatable workflow tasks that increasingly sophisticated AI agents may soon replicate at a fraction of the cost.
By contrast, system-of-record software, which is deeply embedded in core enterprise processes – think ERP systems, CRMs, or financial ledgers – is in a fundamentally different position. These companies often benefit from switching costs, proprietary data, complex integrations and long implementation cycles. In sectors such as healthcare and financial services, regulation can create an additional layer of protection.
AI will still change these businesses. But changing a workflow is not the same as replacing a software company entirely.
The underwriting questions are changing
Historically, software lending focused heavily on recurring revenues, leverage levels and sponsor backing. Those factors still matter, but they are no longer sufficient on their own.
Today, investors increasingly need to ask different questions. How much of a borrower’s revenue is genuinely vulnerable to automation? Does the company possess proprietary data or benefit from regulatory protections? Is it deeply embedded within customer operations? How difficult would it realistically be for customers to switch away?
Coding has been disrupted first because outputs are measur-able and easy for AI systems to test and refine, but the same disruption is increasingly spreading across other knowledge-work sectors, from insurance broking and legal workflows to compliance-heavy administrative functions across healthcare and finance.
By contrast, sectors tied more directly to physical assets or essential infrastructure face a different disruption profile. A data centre, power network or industrial asset may benefit from AI-driven demand growth without facing the same immediate threat of technological displacement.
Disruption is not the same as default
Many businesses exposed to AI pressure will not disappear. Instead, they may evolve toward lower-margin or usage-based models as pricing power weakens and competitive dynamics shift.
The key challenge for lenders is distinguishing between busi-nesses facing temporary repricing pressure and those whose underlying economics are being permanently impaired by AI-driven changes to customer behaviour, pricing structures or competitive barriers.
In this new world, headline software exposure numbers are becoming less useful on their own. Two private credit managers may report identical software allocations while holding radically different underlying risks depending on the durability of the businesses they finance.
AI has changed the hold-period equation
In liquid credit markets, investors can often reduce exposure or exit positions as disruption unfolds. In private credit, managers may be underwriting businesses for four or five years in a technological environment evolving at unprecedented speed.
AI risk is no longer simply a question of whether a manager owns too much software. It is a question of whether the borrower’s business model can remain durable for the full life
of the loan. In liquid credit, investors can change their mind. In private credit, they must live with the consequences.
RISKS
Investment strategies involving Private Markets (such as Private Credit, Private Equity and Real Estate) are complex and speculative, entail significant risk and should not be considered a complete investment program. Such investments should be viewed as illiquid and may require a long-term commitment with no certainty of return. Depending on the product invested in, such investments and strategies may provide for only limited liquidity and are suitable only for persons who can afford to lose the entire amount of their investment. Private investments present certain challenges and involve incremental risks as opposed to investments in public companies, such as dealing with the lack of available information about these companies as well as their general lack of liquidity. There also can be no assurance that companies will list their securities on a securities exchange, as such, the lack of an established, liquid secondary market for some investments may have an adverse effect on the market value of those investments and on an investor's ability to dispose of them at a favorable time or price.
WF: 11533808