Apocalypse now?

The US software sector has had a brutal start to the year, dubbed the ‘software apocalypse’, amid fears that startups built using generative AI (genAI) from the outset – ‘genAI natives’ – could replace incumbents’ service offerings.

The software-focused IGV index was down 20% relative to the US benchmark S&P 500 at the time of writing. It’s not just software stocks that have suffered. Information services companies, such as London Stock Exchange Group and RELX, have also experienced collateral damage.

Until last year, these once high-flying sectors were regarded as home to some of the best businesses in the world. They were characterised by dependable recurring subscription revenue, growth rates above GDP, pricing power, and excellent cash flow generation. In short, they appeared to have everything you could ask for in a high-quality business.

Organic revenue growth (excluding growth from mergers and acquisitions) is typically over 10% in the software sector. Many companies hold net cash on their balance sheets (more cash than debt), and profit margins have been rising. If forecasts are correct, stocks in ‘global enterprise software’ – businesses that sell large-scale, integrated software systems to other companies – look attractively valued.

So why are these software companies getting cheaper, relative to their ability to make money? Putting it another way, why have we seen such a sell-off?

A plug-in threat

The most recent bout of selling was triggered by the release of plug-ins from Anthropic, a leading large language model (LLM) lab specialising in coding tools. A plug-in is software that increases the functionality of another software application. Anthropic introduced a new feature called Skills that, rather than simply enhancing coding and other applications, competes more directly with software service providers.

There’s been no immediate impact on incumbents’ revenues. But the decline in share prices reflects concerns. Some investors fear incumbents could eventually be replaced.

LLMs enable AI agents to perform tasks previously handled by humans by combining understanding, reasoning and action to automate complex workflows. This new technology has generated several concerns:

• Demand for enterprise software could shrink if automation reduces headcount and workflow complexity at companies using platforms supplied by businesses such as Microsoft, Salesforce, SAP and Oracle.
• GenAI-native startups may operate with lower cost structures and superior functionality compared with incumbents retrofitting AI into legacy products.
• Margins could be compressed if more value accrues to LLM providers and genAI chip makers, leaving software incumbents with a smaller share of the economic value created.

The innovator’s dilemma

A classic treatise on innovation is The Innovator’s Dilemma, written in 1997 by Harvard professor Clayton Christensen. He distinguishes between two types of innovation:

• Sustaining technologies, which improve the performance of established products that mainstream customers already value, such as faster processors or better screen resolution.
• Disruptive technologies, which create simpler, cheaper or more convenient ways of doing things, such as digital cameras replacing Kodak film, or Google overtaking the Yellow Pages.

What about incumbents adding genAI natural language interfaces to their applications? We think this is a sustaining rather than disruptive technological phenomenon. It doesn’t fundamentally undermine their economics. In fact, it could enhance revenue growth and margins. By making customers more productive, software providers may be able to charge more, effectively taking their share of the money freed through labour savings. However, the question remains: could incumbents ultimately be replaced?

Software moats to keep the startups out

To withstand this wave of AI-driven disruption, software companies need durable moats – competitive advantages that retain customers and deter entry. Companies such as Salesforce and SAP benefit from deep integration within clients’ ecosystems. These integrations create ‘stickiness’, making it costly and complex to switch providers. That should make them resilient in the face of AI-native challengers.

The real risk is complacency. Failure to embrace AI could allow superior rival products to overcome switching costs. But we expect incumbents to evolve, as they did during earlier transitions from PC to mobile and from on-premises servers to cloud computing.

While the market focuses on downside risks, AI agents also present a significant opportunity. Hundreds of billions of dollars in incremental revenue could be unlocked through automation of repetitive, data-driven back-office processes. From this perspective, AI could enhance incumbents’ pricing power and even lift margins as they reduce their own labour costs. For now, however, markets are unwilling to price in this upside scenario.

Software is stayin’ alive

We see durable long-term value in the software sector, particularly after the recent sell-off. But we can’t predict when sentiment might turn positive again. It may remain negative for some time, especially if further product launches from LLM providers reignite disruption fears.

For now, valuations – such as prices relative to earnings – are likely to remain depressed. We can’t call the exact floor. But at current levels, markets appear to have priced in significant and widespread market share losses across the sector.

One potential catalyst for a rerating would be a meaningful acceleration in revenue growth among incumbents. However, that seems unlikely within the next 12 months, as AI-enabled enhancements are still in the early stages of rollout.

In our view, incumbents are unlikely to be fully disrupted. Over time, markets should recognise this. Meanwhile, these businesses continue to generate strong free cash flow – cash remaining after operating expenses and capital investment – and robust profit growth. Their task is to keep delivering, reinvesting appropriately in research and development, and integrating genAI functionality while using surplus cash prudently. That includes share buybacks where appropriate.

What would change our view? Clear evidence of sustained market share loss to AI-native competitors. That could occur if moats prove shallower than expected or management teams are slow to embed genAI into their core offerings.

For now, we remain confident that high-quality, sustainable growth can still be found among many leading software companies. At lower valuations, they may even offer more compelling long-term opportunities for selective investors. That said, rapid technological change reinforces the importance of diversification, avoiding excessive exposure to A-Irelated risks within portfolios.