EliseAI announced on 29 September that it had raised 350 million dollars at a 4 billion dollar valuation, co-led by Andreessen Horowitz and Bessemer Venture Partners, with Ontario Teachers’ Pension Plan joining alongside Sapphire Ventures and Navitas Capital. It is the fourth time a16z and Bessemer have backed the company since 2023, and roughly thirteen months since a 250 million dollar Series E valued it at 2.2 billion.
The underlying numbers are the reason. The company passed 200 million dollars of annual recurring revenue in June 2026, having doubled revenue year on year for five consecutive years. Its platform now covers around one in six American apartments, and more than 30 million Americans have interacted with it since the company was founded. The round was entirely primary capital, with no secondary sale for early investors or employees.
What the company actually does is the least fashionable work imaginable. It automates leasing enquiries, resident communications, maintenance requests, billing and renewals for property managers, and it has expanded into healthcare operations.
Co-founder and chief executive Minna Song put the strategy in one sentence: the industries where AI matters most are still not the ones getting the most attention, housing has enormous problems to solve, and the company grew by going deeper with its customers until it solved the root causes.
That is a different business model from the one most AI companies are pursuing, and the difference is worth naming precisely.
The crowded approach is horizontal. Build a capable general tool, sell it to knowledge workers everywhere, and let the customer work out how to apply it. The market is enormous, the competition is brutal, and the product is comparatively easy to switch away from because nothing about it is specific to the customer’s situation.
The approach here is vertical and narrow. Pick an industry with real operational pain, learn it in detail, and build something that only makes sense inside that industry. The addressable market is smaller on paper. The thing you build is almost impossible to displace, because displacing it requires the competitor to learn the same industry at the same depth.
Bessemer’s Sameer Dholakia, who joined the board, credited the combination of AI research with a detailed understanding of how properties actually operate. That is the whole thesis in a clause.
The financing context sharpens it further. Global proptech venture funding ran at about 4.53 billion dollars across 231 deals in the first half of 2026, roughly 65 per cent below the 2021 and 2022 peak. This single round equals close to 8 per cent of that entire half-year total. Investors are writing fewer and larger cheques, and they are writing them for depth rather than breadth.
There is a version of this available to almost any operator who serves a trade. Knowing one industry’s problems better than anyone else is a defensible position that requires no technology at all, and it compounds, because every year inside it widens the gap.
The unglamorous industries are unglamorous because nobody wants to learn them.
That is precisely why they pay.
