As of Sep 22, 2026
All companies

Legal AIFounded

Harvey

Helps lawyers research and draft legal documents.

Winston Weinberg
Winston Weinberg
Gabriel Pereyra
Gabriel Pereyra

Business performance

Read the revenue model
Valuation
$15.50B
ARR
$400.00M +
Annual revenue
Not disclosed
Customers
1K +
Net income
Not disclosed
Valuation / ARR
38.8×

ARR

USD millions
View observations and sources
PeriodARRSource
2025-12-31$190.00M +linkedin.com
2026-09-09$400.00M +nonbillable.co.uk

Reading the numbers

ARR grows through legal-organization contracts. Seats and workflow coverage affect deal size beyond customer count. A $15.5 billion valuation against $400 million ARR is about 38.8 times ARR, not an earnings multiple.

Multiple calculation

Sep 9, 2026 $15.50B ÷ Sep 9, 2026 $400.00M + = 38.8×

01What it sells and who buys it

Harvey supports legal research, document analysis, drafting and workflow automation for law firms and in-house teams. It integrates AI into work involving an organization's legal knowledge and documents.

02How the business makes money

Revenue comes from paid organizational deployments tailored to legal workflows and security requirements. Broader internal adoption and additional use cases create expansion opportunities.

03Marketing and customer acquisition

Harvey built evidence inside a major law firm and expanded through professional-service networks.

  1. 01

    Test tens of thousands of real legal questions

    Allen & Overy reported a trial beginning in November 2022, with about 3,500 lawyers asking roughly 40,000 questions by February 2023 before rollout to 43 offices.

    Source
  2. 02

    Enter a professional-services distribution network

    Trials inside major law firms helped refine the product with professional users. A strategic PwC alliance created access to a professional-services network spanning more than 100 countries. Enterprise purchasing and deployment support are central to distribution.

    Source
  3. 03

    Financing the next stage

    In September 2026, Harvey raised $550 million co-led by Diffusion and Lightspeed at a $15.5 billion valuation. It reported adoption by 80% of the Am Law 100 and tied the financing to legal intelligence and organizational expansion.

    Source

How the growth mechanism works

  1. Test real legal work
  2. Collect organizational feedback
  3. Win professional partners
  4. Expand across offices

Legal AI requires review and deployment conditions beyond fluent answers. The initial law-firm trial provided an environment to refine those requirements. PwC offered distribution at a different organizational scale from individual lawyer acquisition.

04Funding and expansion

In September 2026, Harvey raised $550 million co-led by Diffusion and Lightspeed at a $15.5 billion valuation. It reported adoption by 80% of the Am Law 100 and tied the financing to legal intelligence and organizational expansion.

More about the founding and key events

Lawyer Winston Weinberg and AI researcher Gabriel Pereyra founded Harvey in 2022. Weinberg had worked in litigation. One founder understood the buyer’s daily work while the other brought model expertise.

Professional services organizations became a distribution channel. In March 2023, PwC announced a partnership offering access to its tax and legal professionals across more than 100 countries. Harvey entered an organization with an existing network of professionals and clients. That potential reach was broader than the number of active users established by the announcement.

Allen & Overy’s account shows early adoption in practice. It began testing Harvey in November 2022. By February 2023, about 3,500 lawyers had submitted roughly 40,000 questions, and it rolled the product out across 43 offices. A single large customer generated substantial real work and feedback. The firm also required lawyers to review the output.

Harvey’s value includes placing research, document review and drafting inside the law firm’s workflow and enabling organizational adoption. It announced a $15.5 billion valuation in September 2026. The authority of early clients may have supported later sales, although the sources do not isolate that effect.

Further reading

  1. 01linkedin.comRead source
  2. 02nonbillable.co.ukRead source
  3. 03harvey.aiRead source
  4. 04aoshearman.comRead source
  5. 05pwc.co.ukRead source
  6. 06harvey.aiRead source
  7. 07harvey.aiRead source
  8. 08fortune.comRead source
  9. 09harvey.aiRead source