The software selloff is overstated

The recent market panic about AI adoption rendering the software sector obsolete came with predictable results: selloffs, lowered investor confidence and general doom-mongering. While we regard these concerns as legitimate, the notion that AI will make a clean sweep through the software economy is misguided.

Our own analysis of AI risk has identified three categories for software businesses: those facing genuine near-term disruption, those with structural resilience, and those which are actively benefiting from the AI revolution. Getting these distinctions right is crucial when selecting the right investments as software is not a single category at risk of uniform devastation.

Application software: Facing real disruption

Application software is how we describe the tech that enables daily workflows, customer service and operational processes. The threat to this segment is very real and is already happening. Wherever AI can replace the main function of a software application against the backdrop of limited switching costs for the customer, we are seeing the change being made.

For example, customer contact centres are replacing humans with AI agents with compelling economics which will only accelerate. Companies such as Omilia, a conversational AI business, are already deploying AI agents that directly replace human operators at scale. Routine data processing businesses and point solutions in document drafting also face similar pressures.

Despite this, not all businesses in this cohort are necessarily going to vanish. Risk is most acute for horizontal point solutions serving consumers or small businesses. In contrast, enterprise application software, with high cost and implementation barriers to switching due to the embedded nature of this technology, is in a much stronger position to weather the storm.

Infrastructure and developer tools: Active beneficiaries of AI

In contrast to the above, data infrastructure and developer tools serving the new tech stack are becoming even more crucial to their customers thanks to AI innovation. Every AI-native product requires software infrastructure to run properly. Every AI model requires data to be stored securely, queried and properly governed.

The growth in AI workloads is a direct tailwind for businesses in this segment with demand rapidly increasing across security, developer platform and observability categories. One notable example is graph databases as they address two of the most pressing challenges with AI use, with accuracy (models often hallucinate) and cost (inference at scale is expensive). As AI deployments get scaled up, this kind of software architecture becomes a core requirement for customers.

The competitive moat: Proprietary and regulated data

The role of proprietary data has been completely overlooked. In reality, software companies build on data which cannot be replicated. For example, years of patient outcomes in healthcare or financial transaction data have a form of protection that a generalist AI models cannot overcome. In these cases, the data itself is the foundation of the competitive advantage, training the AI models, and it only compounds over time, reinforced by regulation. Regulatory obligations make it very unattractive for customers to switch to different solution providers, so incumbents prove far more durable than expected due to meaningful switching barriers.

Where can investors go from here?

When considering all of the above, it is helpful to remind ourselves that the software sector has navigated major technology transitions before. In previous sector shifts, such as the Cloud rollout, vendors agile enough to adapt came out structurally stronger, with recurring revenue, lower distribution costs, and significantly higher margins than the legacy software it disrupted. Transitions can be jarring but businesses which adapt well should come out with stronger unit economics, not weaker ones.

Within the Patria Private Equity Trust (PPET) portfolio, technology is a core sector we invest in, at 23% overall, made up of private businesses across Europe. Omilia is one recent investment for us, and the European P/E landscape is teeming with companies which possess the characteristics we believe give them a competitive advantage in the new AI economy, such as large, complex customers with high switching barriers, and proprietary or regulated data which is hard to replicate.

Another great example from PPET’s portfolio is Visma, a Norwegian company and one of our top holdings for several years offering mission critical, vertically integrated SaaS solutions to customers across Northern Europe, Benelux and South America, with strong prospects for further growth.

We look for businesses operating in sectors where expertise and trust are paramount, as well as those actively embedding AI into their proposition, rather than just waiting to be disrupted by it. The window to gain exposure to one of the most compelling investment opportunities in decades is narrowing, therefore investing via software specialist P/E partners who understand the nuances of this sector is critical; a generalist approach will no longer deliver the historical results many investors may have become accustomed to.