While most VCs race to fund AI startups, top firms like Thrive Capital are taking a different route: buying “boring” legacy businesses.
Back when OpenAI generated just $50M in API revenue, Thrive, an early backer of OpenAI, Stripe and GitHub, pitched private equity firms a new thesis: what if we used AI to transform traditional portfolio companies? PE passed on the idea. So Thrive decided to do it themselves.
Startups usually build from scratch to disrupt incumbents from the outside. But Thrive saw that legacy companies already hold two unfair advantages: proprietary data and domain experts. Instead of building a startup, Thrive buys established platforms and uses AI to modernize them from within. The result: 70+ acquisitions across accounting, IT services and logistics.
AI models are a commodity. The real moat is execution:
- Operational executionCan you actually transform the workflow?
- Industry know-howDo you know what should and shouldn’t be automated?
- Proprietary knowledgeCan you turn the knowledge inside employees’ heads into a system?
- Distribution and trustCan you retain your customers and people?
Thrive isn’t the only player in the “AI roll-up” trend. Long Lake has acquired 70+ property-management companies, reaching $100M EBITDA and 2× margins in two years. Crescendo handles 90% of frontline contact-center tickets with AI, at 4× the margins of a traditional operator. Titan MSP has automated 30%+ of IT-services workflows, targeting 3× net margin. Dwelly cut real-estate problem resolution from 50 days to 20, at 2× margin. And General Catalyst has put $1.5B behind essentially the same thesis.
Sources: Crosslantic Capital; gregisenberg.com
