Harvey AI
Legal AI platform for law firms and corporate legal departments. On Sept 9, 2026 it closed a $550M round at a $15.5B valuation co-led by Diffusion and Lightspeed, with Sapphire and Whale Rock joining and Sequoia, Kleiner Perkins, a16z, Coatue and GIC returning, which confirms the raise reported as in talks in August and lifts total funding above $1.5B. Co-founder Winston Weinberg put ARR above $400M, more than double the $190M of January 2026, with 3,000+ customers (from about 1,300 in March), about 80% of the AmLaw 100, five of the Fortune 10 and roughly 20% of the Fortune 500. It shipped Tenet, its first post-trained legal model built on Moonshot's open-weight Kimi K3, and bought agent-testing startup Guardrails AI, its fourth acquisition of 2026, to cut reliance on OpenAI and Anthropic, both of which now sell into legal directly. California SB 574, chaptered Sept 30, bars lawyers from delegating the practice of law to generative AI and requires personal verification of AI-supplied citations. The round prices the company at roughly 39x ARR, down from about 58x in March.
Key Investors
Scenarios
Revenue more than doubled in about eight months to $400M+ while the valuation multiple compressed from roughly 58x to 39x ARR, so growth is outrunning the price. Penetration of about 80% of the AmLaw 100 plus a fast-growing in-house base (five of the Fortune 10) is distribution that frontier labs lack, and owning a post-trained model and evaluation stack (Tenet, Harvey LAB, Guardrails AI) moves margin and differentiation away from rented frontier APIs. Rules like California SB 574 that demand verified citations and confidentiality controls favour auditable enterprise legal tools over generic chatbots.
Anthropic's legal plug-ins for Claude (February 2026) and OpenAI's law-firm partnerships attack the same workflow from the model layer, while Thomson Reuters CoCounsel, LexisNexis and Legora compete on content and price. At roughly 39x ARR any slowdown compresses the valuation hard. Building on a Chinese open-weight base model (Kimi K3) may raise procurement questions with government and security-sensitive clients, and SB 574 limits on entering confidential client information into generative AI systems could slow some California use cases depending on how its carve-outs are read. Every metric is company-stated by a private company.
Key Factors to Watch
- ●$550M round at $15.5B valuation closed Sept 9, 2026 (Diffusion and Lightspeed co-led); total funding above $1.5B
- ●ARR above $400M (co-founder statement, Sept 2026), up from $190M in January 2026 and about $350M in August
- ●3,000+ customers (from about 1,300 in March), about 80% of the AmLaw 100, five of the Fortune 10
- ●Tenet post-trained legal model on Kimi K3 plus Guardrails AI acquisition (fourth of 2026) shift the stack toward owned models
- ●Anthropic legal plug-ins, OpenAI law-firm partnerships, Thomson Reuters, LexisNexis and Legora cap pricing power
- ●California SB 574 (Chapter 858, chaptered Sept 30, 2026) bans delegating legal practice to GenAI and mandates citation verification
Score History
| Date | Score | Direction | Note |
|---|---|---|---|
| 2026-10-10 | 8.1 | Very Positive | W41 rotating refresh, HOLD 8.1. The $550M raise at $15.5B reported as in talks in August closed Sept 9 (Diffusion/Lightspeed); ARR above $400M and 3,000+ customers per co-founder Weinberg; Tenet model on Kimi K3 and Guardrails AI acquisition; California SB 574 chaptered Sept 30. Fixed stale $190M ARR, $11B round and 58x multiple (now about 39x). aam 8 to 9 candidate fails the gate on magnitude (+0.10). |
| 2026-08-22 | 8.1 | Very Positive | Score 7.9->8.1 (moat_durability 8->9). ARR nearly doubled to $350M (from $190M in Jan) in ~7 months while in talks (Aug 7, The Information) to raise $500M at $15.5B - 40% up from $11B in 5 months - validating sustained dominance vs Legora, Thomson Reuters, LexisNexis and Anthropic's legal plugin. Note: the raise is reported in talks, not closed. |
| 2026-05-16 | 7.9 | Positive | W20 light refresh — score holds 7.9. $200M raise at $11B valuation (March 2026), 100K+ lawyers across 1,300 organizations, 25K+ custom agents in production. GIC + Sequoia lead. |
| 2026-04-10 | 7.9 | Positive | Score 7.5→7.9 (sb 6→7, md 7→8). $200M Series E at $11B valuation (March 25 2026), ARR $100M→$190M in 5 months, AmLaw 100 majority penetration. Stays positive. |
| 2026-03-08 | 7.5 | Positive | Score 7.3→7.5 (formula reweight: sb 0.25→0.15, are 0.20→0.25, md 0.20→0.25, dr 0.20→0.25, aam 0.15→0.10) |
| 2026-03-08 | 7.3 | Positive | Score 7.5→7.3 (dr 3→4). External research cross-ref: growing competition from Thomson Reuters Westlaw AI, LexisNexis CoCounsel, Legora ($1.8B) |
| 2026-03-08 | 7.5 | Positive | Initial assessment from batch 8 research |
Legal Peers
Last researched: 2026-10-10
This is research and analysis, not financial advice. Scores reflect AI impact potential, not investment recommendations.