Harvey, the legal-focused AI startup, raised another $550 million this week at a $15.5 billion valuation, co-led by Diffusion and Lightspeed. In the same news cycle, Clay, an AI sales-and-go-to-market tool, raised $115 million at $7.1 billion, more than doubling its valuation in a year. Notice the pattern: the biggest cheques are increasingly going not to the general-purpose model labs, but to companies wrapping those models in one specific profession.
This is the “vertical AI” thesis playing out in real time, and it is arguably the more investable trend than betting on which frontier model wins. The logic: a raw model is a brilliant generalist that knows nothing about your industry’s workflows, compliance rules, or document formats. A vertical company like Harvey buys or licenses the model, then layers on the legal-specific tooling, security, and integrations that a law firm will actually pay for. The model is the engine; the vertical is the car people can drive.
Why investors love it (and the catch)
The bull case for vertical AI is that it captures durable value the model labs struggle to. Switching costs are high once a firm builds its workflow around you, the customers are wealthy and sticky (law, finance, healthcare), and you are insulated from the brutal price war between the model providers, because you can swap the engine underneath and keep the customer. That is why Harvey can command a $15.5bn tag while being, at some level, a very sophisticated wrapper.
The catch is exactly that dependence. Vertical AI companies live one product decision away from their model supplier deciding to compete directly, or from an open model getting good enough to erase their edge. And $15.5bn is a rich price for a company whose core intelligence it does not own. The bet you are making, if you buy the vertical thesis, is that domain expertise and distribution matter more than the model itself. This year’s funding rounds suggest a lot of serious money agrees. (Not investment advice.)