The law firm that takes three months to finish a memo is about to take a startup public in three weeks.
The Summary
- Harvey, the legal AI startup valued at $15.5 billion, has started preliminary talks with Latham & Watkins about an IPO, though no listing is imminent
- The company crossed $400 million in annual recurring revenue in September, selling AI software to law firms and corporate legal departments
- Harvey is watching how Anthropic's potential Thanksgiving-week IPO performs before deciding whether to move forward
- A four-year-old company going public would've seemed absurd a year ago, but SpaceX just landed the largest IPO in history and OpenAI could follow in early 2027
The Signal
Harvey represents the first wave of vertical AI companies testing whether enterprise buyers will pay enough to justify billion-dollar public valuations. Legal was always going to be the beachhead. Lawyers bill by the hour, and AI agents do research faster than associates. The business model writes itself.
What's remarkable is the speed. Harvey launched in 2023. It's now doing $400 million ARR. For context, Snowflake had generated $242 million in revenue when it went public in 2020. Affirm was at $509 million before its 2020 listing. Harvey is playing in the same stadium, just three years after founding.
"A four-year-old startup making its market debut would have seemed far-fetched just a year ago."
The timing tells you everything about the shift happening in public markets. Anthropic could IPO the week before Thanksgiving. OpenAI might follow in Q1 2027. If those listings hold, the door opens for application-layer AI companies to test investor appetite. Harvey would be among the first pure vertical plays to try.
But here's the strategic read: Harvey is watching, not rushing. The Anthropic IPO is the signal they need. If it trades well, Harvey has a clear path. If it stumbles, they stay private longer and keep the $550 million they raised this fall. The company has leverage. Law firms are locked in. Corporate legal departments are expanding usage. The revenue is real and recurring.
The deeper implication is what this says about agent deployment at scale:
- Legal research is now automated for major firms
- Contract review happens faster than humans can read
- Due diligence that took weeks now takes hours
- Associates bill fewer hours, partners bill more
The Implication
If Harvey goes public in 2027, it validates the thesis that vertical AI agents can build billion-dollar businesses faster than any software category in history. Watch the Anthropic listing. If it prices strong and holds, expect a wave of four- and five-year-old AI companies to file in the first half of 2027.
For law firms, this is the beginning, not the end. The firms buying Harvey now are building moats. The ones waiting are watching their most profitable work get commoditized by competitors with better tools. The IPO is just Harvey proving that selling automation to the people who bill by the hour is the best business model since the billable hour itself.