Harvey’s latest funding round tells you most of what you need to know about where legal AI stands. On September 9, according to TechCrunch, the company raised $550 million at a $15.5 billion valuation, in a round co-led by Diffusion and Lightspeed Venture Partners.
In December it had been worth $8 billion. Few corners of enterprise software have repriced that fast. The stop in between is easy to forget. In March, Harvey closed $200 million at an $11 billion valuation, which at the time looked like a peak. Six months later it looks like a way station. The company has now raised more than $1.55 billion in total and has nearly doubled its value in roughly nine months.
Investors aren’t paying for a demo. They are paying for adoption inside institutions that historically moved at the pace of a partnership vote. Harvey’s annual recurring revenue is above $400 million.
Those figures would be strong for a company selling to marketing teams. For one selling to general counsel and litigation partners, whose default answer to new software has long been a request for another security questionnaire, they mark a genuine change in behavior.
None of this settles the harder question the industry keeps circling, which is what happens to the billable hour. One projection has the legal AI software market growing from $5.21 billion in 2026 to $40.94 billion by 2034, with most of that spending still aimed at speeding up existing legal work rather than replacing hourly billing. That arrangement can hold for a while.
A firm can bill fewer hours on a diligence review and win more reviews. But once clients know a first-pass contract summary takes minutes, the invoice that describes it as six hours of associate time gets harder to defend, and the pricing conversation moves from the firm’s conference room to the client’s. The company isn’t running alone either. Swedish rival Legora was reportedly in talks to raise at more than $10 billion, up from $5.6 billion in March. Two companies adding billions in value within the same half-year suggests buyers have stopped asking whether legal AI works and started asking which vendor they want to be locked into.
The profession that spent a decade treating the cloud as an open question has, in about eighteen months, made AI a line item. That’s the part of the Harvey round worth sitting with.
While these companies may be getting many of the headlines, much of the practical change is happening via other AI solutions. Altorney, for one, was founded by brothers Shimmy and Rachi Messing (article’s featured photo) in late 2021, and has since grown dramatically.
Its MARC platform runs local open-weight models inside a client’s own environment, so review can happen without sending data to OpenAI, Google or Anthropic.
According to the company, “With decades of expertise in eDiscovery, compliance, and litigation technology, they set out to create smarter, more efficient tools to meet the industry’s evolving needs. Built on deep industry insight, Altorney is driving the future of legal innovation—streamlining complex processes, enhancing transparency, and empowering legal teams to work smarter and faster.”
Upfirst AI, on the other hand, works at the opposite end of the market, on the phone line of the small firm. Built by an attorney, it answers calls around the clock, runs intake scripts by case type, and handles more than 35 languages.
Between them, the two show AI reaching both ends of the profession: the discovery room at a large corporate legal department and the front desk of a two-lawyer practice that used to lose clients to voicemail.
