Analysis
Legal AI startup Harvey raised $550 million at a $15.5 billion valuation, TechCrunch reported September 9, co-led by Lightspeed Venture Partners and Diffusion, a new firm co-founded by longtime Harvey backer and former Coatue investor Kris Fredrickson. Sapphire Ventures, Whale Rock Capital Management and a number of Harvey's existing investors also participated.
A valuation climbing in lockstep with disclosed revenue
The new round nearly doubles Harvey's valuation in roughly nine months, according to CNBC's earlier coverage of the intervening round:
“- Total raised to date -- more than $1.55 billion, with disclosed revenue topping $400 million.”
- Dec 2025 valuation -- $8 billion.
- Mar 2026 valuation -- $11 billion.
- Sep 2026 valuation -- $15.5 billion (this round).
- Total raised to date -- more than $1.55 billion, with disclosed revenue topping $400 million.
Harvey said 80% of Am Law 100 firms and five Fortune 10 companies now use its tools for legal research, contract analysis and workflow automation.
Harvey said the fresh capital will go toward building its own AI models rather than relying exclusively on frontier labs like OpenAI and Anthropic for its underlying technology -- a strategic shift that puts Harvey in more direct competition with the same labs whose models it has historically licensed, and mirrors a pattern of vertical AI-application companies increasingly investing in proprietary model development as their revenue scales enough to justify the compute spend. Pulse tracks Harvey's climb here.
The competitive field in legal AI
Harvey's climb has come alongside, not instead of, a broader legal-AI funding wave: competitors including Legora and Eve have each raised significant rounds this year targeting overlapping enterprise legal-research and contract-review use cases, though none has disclosed a valuation approaching Harvey's. The $15.5 billion mark on $400 million-plus in disclosed revenue implies a roughly 39x revenue multiple -- rich by traditional enterprise-software standards, but broadly in line with what other AI-native vertical software leaders have commanded this same month.
The decision to build proprietary models is the more consequential long-term bet than the valuation itself. Training and serving frontier-competitive legal-specific models is materially more capital-intensive than fine-tuning or prompting a third-party foundation model, and it puts Harvey's cost structure on a different trajectory than pure-application companies that stay model-agnostic. If Harvey's in-house models don't meaningfully outperform what OpenAI or Anthropic already offer for legal use cases, the company will have spent a large share of this round on infrastructure that a lighter-weight competitor building on top of the same frontier labs could match at a fraction of the cost.
