The software development vertical is the primary AI battleground due to tools offering both rapid value and long-term durability, with the potential for a single company to become the first-ever platform for the entire engineering vertical.
The future of software development is a decisive shift away from IDE-centric tools towards AI agents, which are demonstrating massive user adoption and engagement.
A small number of elite private AI companies, such as Anthropic, are operating on a different plane and have a plausible path to becoming trillion-dollar businesses.
Modern venture capital strategy requires 'laddering up' ownership over multiple rounds, and firms must be willing to learn from significant misses (like Rippling or ServiceTitan) to avoid being too rigid on valuation rules for generational companies.
AI tools for regulated industries like legal and accounting have a longer time-to-value but exhibit greater 'durability' and become transformational once integrated into core workflows.
▶The Anatomy of an AI Hyperscaler
Clements provides a detailed case study of Cursor's explosive growth, focusing on its ARR trajectory from an estimated $100M to a reported $2B. He emphasizes metrics like 90% daily active usage of its agent product and 15x agent growth as the key drivers, fueled by a consumption-based model rather than per-seat licenses.
For investors, this highlights that modern AI SaaS growth is less about seat licenses and more about deep, consumption-driven integration into user workflows, making engagement metrics paramount for predicting revenue.
▶Investing Through RegretMar 2026
A significant portion of Clements's analysis is framed by Excel's investment misses, which he openly discusses. He cites Rippling, ElevenLabs, ServiceTitan, and the initial wave of foundation models as crucial learning experiences that have reshaped the firm's strategy and risk calculus.
This transparency suggests a firm culture that values post-mortems and adapts its investment thesis, indicating that future decisions will be less constrained by historical valuation rules, especially for 'generational' companies.
▶The Engineering Vertical as the Platform PrizeMar 2026
Clements posits that the software development vertical is the ultimate battleground for AI because it uniquely combines rapid time-to-value with long-term, compounding durability. He believes no company has ever successfully become a true platform for the entire engineering stack, and that AI tools like Cursor are the first to have a credible shot at this massive prize.
This thesis signals a focus on 'platform' investments within specific, high-value verticals rather than broad, horizontal AI applications, suggesting a belief that vertical-specific AI will create the most defensible moats.
▶The New Private Market ParadigmMar 2026
Clements observes that top private tech companies no longer need to go public for liquidity or M&A currency, creating a class of permanent private giants. He predicts that a small number of these, like Anthropic, could reach trillion-dollar valuations, while others unable to IPO at peak valuations may become targets for large private equity firms.
This suggests a bifurcation in late-stage venture, where analysts must differentiate between potential trillion-dollar outliers and high-quality but overvalued companies that represent a different kind of M&A opportunity for PE funds.