Legal AI FUD Hits The Timeline

TBPN，科技每日秀 Substack 专栏。关注

Happy Tuesday.

We’re live on YouTube and 𝕏.

The current thing in tech and business is startups like Harvey moving to cheaper, open-source models.

Today’s Lineup

Fair Square Medicare Founder & CEO Daniel Petkevich at 12:00 PM
The Horowitz Andreessen Academy CEO Gagan Biyani at 12:10 PM
Snorkel AI Co-Founder & CEO Alex Ratner at 12:20 PM
Swarm Aero Co-Founder & CEO Peter Kalogiannis at 12:30 PM

Run of Show

Legal AI FUD Hits The Timeline

People are reacting severely to some news that legal AI startup Harvey’s gross margin fell from about positive 50% to negative 50% by June. The actual reporting comes from Bloomberg, and explains in more detail. Costs exploded this year, but on the back of more product usage (which is good!). Harvey says token consumption increased 20x this year. Also, the negative 50% number is from June and the company says they are back to positive gross margins after changing model usage and launching Harvey Tenet, their first post-trained open-weight model.

So, negative gross margins in legal software is very, very bad, but these are weird unprecedented times and costs swing wildly. It’s naive to think that gross margins would structurally be negative in this category for any enduring reason. Some quick comps: For Q2 2026, E-discovery provider DISCO generated a 74.4% GAAP gross margin and Thomson Reuters’ Legal Professionals segment posted a 48.1% adjusted EBITDA margin. Even pure-play law firms have healthy margins: PwC reports net profit margins of 41.2% for the top ten firms (partnership profits also compensate equity partners for their work, so not directly comparable with corporate profits).

The obvious good news is that tokens are getting cheaper by the day, there is fierce competition at every layer of the stack, and moving from one model to another is not particularly costly. Despite Grok 4.7 not exactly leaping to the frontier, it did perform extremely well on the Harvey Legal Agent benchmark both in terms of cost and quality.

So with that backdrop, it’s not crazy to see how Harvey has been hovering up capital, despite jitters in the margins. Harvey announced $550m of funding at a $15.5b valuation on September 9, with more than $400m in ARR, 3,000+ customers, and adoption by 80% of the top 100 law firms.

The debate will rage, the AGI-pilled will doom about the fact that Harvey is sort of structurally short superintelligence, but in the meantime, the proof will be in the ARR growth. Law firms are already starting to mull vertical integration at the legal AI layer, and Harvey explicitly says it doesn’t use customer data in post-training their internal models (Tenet), so the product better be really really good to keep lawyers happily paying.

Clip Spotlight: Box CEO Aaron Levie says AI has been “unequivocally” a net positive for software

“You look at infrastructure providers—the Cloudflares of the world—totally on fire, because agents need sandboxes, they need compute, they need network, they need gateways. Great business.”

“For us, we have reaccelerated growth far past our internal plans because it turns out that enterprises need core systems to be able to manage their unstructured data.”

“Agents need to be able to work with that unstructured data to make decisions or move information through a workflow—whether it’s all of your contracts, your research materials, marketing assets, or financial documents.”

“Now imagine an enterprise with 10,000 employees. How do you ensure that those agents continue to go after the actual canonical records, the actual documents that are the real sources of truth and the authoritative versions of that data?”

“That means you need platforms.”

“And all of that is creating more value for these platforms.”