Highlights: Fear and herd mentality are back to haunt the markets with the onset of the ‘AI fear trade’. It started with the recent release of Anthropic’s legal AI plug-in which turbocharged AI disruption fears - triggering a sell-off even in the most well established, globally entrenched software companies. This so called ‘SaaS-pocalypse’ is now spreading to other sectors, from tax advisory to real estate services. To us, the market appears to be selling first and asking questions later.

What are the market’s fears for Business Software? Pessimists fear that agentic AI will disrupt the traditional Software-as-a-Service (SaaS) model, and that the traditional ‘System of record + Human workflow + User Interface’ model could simply be replaced by vertically integrated AI agents that deliver both the intelligence and automated multi-step workflow execution, end-to-end, without any human intervention. 

Why is this fear is misplaced? Because it’s too simplistic and ignores crucial intricacies of business software, just like the DeepSeek moment of January 2025. We think it’s not that simple or easy to replace well entrenched SaaS models with Agentic AI, overnight. Here’s why:

• Big SaaS companies have clear moats: such as multi-year corporate contracts, proprietary IP, and pre-existing partnerships with leading companies. All these are extremely difficult and time-consuming factors to build and replace
• Proprietary data and domain IP: Large, complex enterprise class system codes, data repositories and IP are proprietary. They’re fiercely guarded by these companies and therefore are not available for training the foundational models. Therefore, big SaaS companies are unlikely to get ‘vibe-coded’ out of existence by generic LLM models that are trained with mostly publicly available data on the internet
• The SaaS model is best placed to monetise Agentic AI: We think it’s the enterprise software companies that are best placed to embed AI’s different use cases in their platforms and drive monetisation by expanding their and their clients’ TAM (Total Addressable Market). Software companies should also be one of the beneficiaries of the USD650 billion AI capex investment announced by big tech
• AI and SaaS are complementary, not mutually exclusive: We think AI’s usefulness in an enterprise setting is likely to remain limited without a wholesome software ecosystem. For optimal productivity, software platforms are needed to orchestrate the interactions between the AI and non-AI enterprise components. History shows that the cheapest solutions don’t always win. Just because AI can write code, doesn’t mean it will be trusted. Longstanding credibility and seamlessness of established US enterprise vendors is crucial to minimise business continuity and reputational risks

What do long term investors need to mindful of? 

• Agentic AI is unlikely to kill SaaS. It will make it more productive and alter its revenue model from effort based to outcome based.
• Fear driven pockets of selloff often create attractive entry points. They reiterate our conviction in the broadening of the market and looking across and beyond AI for equity returns. From a purely fundamentals’ perspective, software’s earnings strength is intact. We like software infrastructure (cloud providers) & cybersecurity software providers because of rising demand and their impenetrable moats. Overall, we remain well diversified & focus on bottom-up stock selection.