Early stage venture capital is, right now, one of the most compelling places to deploy capital in private markets. A two-person team can build in twelve months what historically required twenty engineers, the best talent in technology is overwhelmingly choosing early stage companies over established software businesses, and AI is creating product categories with no incumbents and enormous latent demand. The entry point, the capital efficiency, and the structural dynamics all favour early stage investors who know what they are looking for.

The late stage software picture is more uncomfortable, and the market has been slow to fully confront it. A significant proportion of 2021-vintage positions were written at multiples that assumed durable SaaS economics and sticky recurring revenue. AI is testing all three assumptions simultaneously, and not all of them are holding. The disruption is not uniform across the category, and there are late stage software businesses that are genuine beneficiaries of AI adoption rather than victims of it. The analysis required to distinguish between the two is more demanding than applying a standard SaaS framework, and many investors are not yet doing it.

The dividing line is a simple question applied to every position: is this company building the disruption or absorbing it? That answer determines more about forward returns than almost any other variable in the current environment.