May's venture funding landscape revealed a decisive reallocation of capital within the AI sector. Of the 29 companies that achieved unicorn status last month, the majority clustered around AI services and robotics rather than large language model development. This represents a meaningful departure from 2024's heavy concentration on foundation model companies. While specific company names from May's cohort remain under embargo in most cases, the pattern aligns with broader trends: investors are increasingly confident that the core model layer is consolidating around a handful of players—OpenAI, Anthropic, and others—and that the real growth opportunity lies in vertical applications and implementation infrastructure. This mirrors historical software cycles where infrastructure booms eventually give way to application booms once the underlying platform stabilizes.
The shift reflects changing market conditions and customer demand dynamics. Enterprises have moved beyond evaluating whether AI works and are now focused on deployment, integration, and measurable ROI. Vertical AI startups pursuing larger annual contract values (ACVs) are discovering that traditional SaaS go-to-market playbooks no longer suffice. Leading vertical AI companies are increasingly leveraging private equity networks, industry conferences, and direct enterprise relationships rather than relying on self-serve or sales-assisted models. The higher deal sizes—often exceeding $500,000 annually—demand executive alignment and proof-of-concept rigor that only direct sales infrastructure can provide. This infrastructure investment requirement itself creates a funding moat: only well-capitalized startups can afford the sales teams needed to close these deals, driving consolidation within the vertical AI market.
The funding reallocation carries significant implications for venture allocators and the broader AI ecosystem. VCs betting on foundation models face intensifying pressure as capital requirements balloon and returns concentrate among the few winners. Conversely, the enterprise services layer now represents the frontier for diversified returns. Megadeals in the $100 million-plus range increasingly favor AI-powered enterprise software rather than model research. This environment will likely accelerate M&A activity as larger software companies acquire vertical AI specialists to build out their AI capabilities faster than internal development allows. For founders, the message is clear: building the next ChatGPT faces diminishing returns, while solving specific enterprise problems with existing models carries substantially higher probability of significant outcomes.