I had a meeting with a Top 5% European lower mid-market GP today, and one of the partners shared is thought about Software Investing 👇
« Software is not 'dead' because of AI, but you now have to be much more disciplined in what you back. The market reaction has been partly justified, but the idea that all software companies will be disrupted is intellectually lazy. Instead, there is a clear split between fragile and highly defensible software businesses.
The types of software we like have common characteristics
1. Backbone systems: software that sits at the core of a client’s operations and cannot be removed without breaking the business (e.g., accounting or ERP systems).
2. Ownership of proprietary data: the software generates and controls the underlying data, rather than just layering on top of existing datasets.
3. Strong distribution moat: customer bases built over many years through direct sales and deep integration, not easily replicated by a new AI-native entrant.
4. Regulatory barriers: software operating in regulated environments (e.g., certified platforms) is far harder to replace quickly.
5. Low tolerance for error: when mistakes have financial, legal, or operational consequences, clients are reluctant to switch to less proven AI-driven tools.
6. High technical depth: complex products with large codebases are more defensible than lightweight tools that can be rebuilt easily with vibecoding.
Bottom line: we remain positive on software, especially vertical, mission-critical, and regulated solutions. The current environment as potentially attractive for investing, as valuations may be over-correcting due to generalized fears around AI. »