VCs Sour on 'AI Wrappers'

While total VC investment in AI soared in 2025, reaching as high as $211 billion, the number of individual deals hit a decade-low. This concentration of capital reveals a "winner-take-all" dynamic, with investors placing bigger bets on fewer, more mature companies that can demonstrate deep intellectual property and proprietary data pipelines. The classic SaaS model of charging per-user, or "seat-based pricing," is rapidly falling out of favor. As AI increases individual worker productivity, companies need fewer licenses to achieve the same output, leading to higher churn for SaaS vendors clinging to the old model. Consequently, VCs are pushing for consumption or outcome-based pricing. Defensibility has become the critical hurdle for new funding. Investors are grilling founders on moats that can withstand competition from tech giants. The most valued forms of defensibility now include proprietary data that creates a feedback loop to improve the AI, deep integration into critical customer workflows, and vertical-specific expertise in regulated or complex industries. AI wrappers that simply provide a thin interface over third-party models like GPT-4 face brutal unit economics. Unlike traditional SaaS with gross margins of 70-80%, these wrappers incur significant API costs for every user interaction, crushing margins down to a "wafer-thin" 25-60%. The rise of agentic AI, which can reason and act autonomously, is seen as an existential threat to SaaS tools that rely on manual human input. These agents can automate complex workflows, making many current software interfaces redundant. VCs are now asking startups how they will compete when autonomous agents can perform the same tasks for free.