A new a16z report looks at which AI companies startups are actually paying for (techcrunch.com)

Andreessen Horowitz, with fintech partner Mercury, published an AI Spending Report using Mercury transaction data to rank the top 50 AI-native application-layer companies that startups actually pay for. The dataset reveals real purchasing behavior (not just downloads), with OpenAI and Anthropic topping the list and Replit ranking third; other notable entrants include Cursor (6), Lovable (18), Cognition (34) and Emerent (48). The list is dominated by horizontal apps (~60%) but includes a meaningful share of verticals (~40%) concentrated in sales, recruiting and customer service. A wide array of note-takers and developer tools appear, indicating no single incumbent has yet captured category dominance. The report’s key takeaway is market fragmentation and transition: startups are primarily buying “human augmentors” or copilots to boost productivity rather than deploying full end-to-end agents, suggesting AI is augmenting workflows for now. Consumer-grade tools (CapCut, Midjourney, Canva) are migrating into enterprise use, accelerating product professionalization and enterprise feature development. For dev tools, spend patterns diverge from consumer traffic (e.g., Replit sees more startup spend than Lovable), underscoring the importance of enterprise features and monetization. a16z expects rapid churn as legacy players add AI features and new entrants iterate—implying opportunity for multiple winners and fast-evolving product-market fit across AI tooling.