Laura Cooper
It is no longer just about owning the winners. It is increasingly about avoiding the losers – those with business models most at risk of disruption. The transition started with digital businesses, yet the software selloff is spilling across asset classes – from heavy software exposure in private credit to pockets of public markets as AI disruption risks rise.
Last week, I had the opportunity to speak with several tech and software investors at Nuveen. One theme kept surfacing: investors are no longer asking ‘who benefits from AI?’ but ‘who gets displaced’. And the selloff is changing that opportunity set in real time.
Markets shooting (software) first, asking questions later
Software has been at the centre of the recent repricing. The rapid advance in large language models is changing the cost of building and delivering software, lowering barriers to entry and intensifying competition. This is challenging existing business models, driving a structural shift from subscription (Software as a Service) to consumption-based pricing, and raising questions about the future of the SaaS model itself.
At the same time, markets may be overestimating the speed of that disruption. Enterprises move slowly, workflows are deeply embedded, and switching costs remain high. Forward revenue multiples have compressed from roughly 10x to 4-6x for many software names, and while a re-rating was warranted, the selloff has been indiscriminate.
This is not a ‘sell all software’ moment – but presents an opportunity for security selection. The sector is experiencing necessary price discovery as the market distinguishes between companies with durable models with pricing power and those facing disintermediation.
Finding the winners among the losers
Our investors remain focused on frameworks to identify companies with demonstrable AI integration, strong network effects, and the flexibility to transition pricing models:
Winners: “Companies operating in categories with high determinism and customisation requirements are best positioned – think design software, vertical-specific solutions, and enterprise resource planning systems. At the vendor level, success favors those with strong data and workflow moats, and usage or outcome-based pricing models.”
Losers: “Service-oriented applications, and creative apps with low customisation needs are particularly vulnerable. Those with weak moats, seat-based pricing models, limited AI progress, and closed ecosystems face heightened disruption risk.”
Security selection, not just a software story
Equity markets tend to reprice first while other asset classes lag, creating a window where risk is reassessed. Yet spreads in parts of the broadly syndicated loan market have already widened by 100-150bps, particularly for issuers with heavier software exposure and more aggressive capital structures. Meanwhile, private credit has increasingly been a source of refinancing for capital structures that were of lower quality, creating pockets of risk in the asset class.
As valuations re-rate and equity cushions shrink, loan-to-value ratios rise, raising refinancing risk and creating less margin for error. While vulnerable issuers at the lower end of the credit spectrum might face refinancing challenges if private credit becomes constrained, the broader, higher-quality loan market should remain relatively insulated from direct spillover effects. And it remains to be seen whether private credit appetite for software will change, or they will continue to provide capital but at wider spreads.
Either way, dispersion is likely to increase not only across borrowers, but across managers focused on the durability of business models.
Big picture: where are the opportunities?
AI is not a bubble technology, but that doesn’t mean every AI bet will pay off. There are companies spending significantly on AI that likely won’t see a return. And the transition from focusing on eyewatering capex to the return on investment has already begun – large, scaled incumbents are investing aggressively not only to innovate, but to protect the moats they have already built with the winners still unknown.
From an opportunity standpoint, the infrastructure and cybersecurity subsectors carry the highest ‘perceived AI safety’. The former is underpinned by strong demand driven by data movement, storage, and analytics. And as AI expands the ‘attack surface’, incremental spending in cybersecurity is likely to remain resilient in this relatively unique part of the market.
In contrast, the application layer is most exposed. This is where the user interface sits, where disintermediation risks are highest, and where the shift from SaaS will be felt most acutely.
Still early stage of AI supercycle
For investors, the passive exposure to software is no longer the same bet in was three years ago. The index-level trade has become a stock-picking story. In credit, the same logic applies where selection and credit differentiation matter more now than in recent years.
The AI investment thesis hasn’t disappeared – it’s just evolving. The recent selloff reflects a repricing within a transformative AI cycle, and periods like this tend to create fatter tails: more winners, but also more losers. The opportunity lies in identifying this flip: who will win versus who will lose.
