AI Infra Companies Are Coming for Verticals

How long will Vertical AI and Infra companies stay frenemies?

It’s a running joke in enterprise software that AWS will eventually launch a product that competes with you—just give it time.

But in AI, you don’t need time. Infra-layer startups can go from helpful API or GitHub project to full-blown SaaS competitor in about the same amount of time it takes you to land an enterprise deal.

This phenomenon— infra companies moving up the stack to ship products—isn’t new. Llamaindex is suddenly a document “agent” company serving mainly finance industry, a long way from its SDK days. Heck, OpenAI just launched a meeting recorder for ChatGPT desktop.

But AI and coding agents accelerates everything, and infra companies will find it overwhelmingly tempting to dip into vertical products, stepping into vertical AI’s turf (or lack therof). I will call this trend infra verticalization.

As I wrote a year ago, pure play AI infra is tough. Thus, it’s only reasonable that they dip into vertical use cases to capture larger margins. And what do infra companies have that vertical companies don’t? Telemetry. They have a more birds eye view of what works (and what doesn’t), which can be used to compete with its customers.

This makes the relationship between vertical AI and infra companies… complicated. Let’s walk through a recent example through the lens of the voice AI infra market.