Introduction
Four years. That's all it took for Harvey to reach 80% of the top 100 US law firms and close a $550 million funding round at a $15.5 billion valuation. The round was co-led by Diffusion — a new fund launched by former Coatue partner Kris Fredrickson — and Lightspeed Venture Partners, with Sequoia, Kleiner Perkins, and Goldman Sachs joining as backers. The deal confirms that legal SaaS has become one of the most fiercely contested segments in B2B tech.
Founded in 2022, Harvey has long since moved beyond automating contract drafting and review. It is now crossing a strategic threshold: building its own language model rather than licensing compute from general-purpose AI providers.
Exceptional Market Penetration
Harvey's adoption metrics are remarkable for a four-year-old company. Half of Fortune 10 companies use its tools in-house for their legal teams. Around 20% of the Fortune 500 are customers. The client base has surpassed 3,000 organizations, and ARR has crossed $400 million.
This level of enterprise adoption reflects a genuine product-market fit with the sector's core requirements: data confidentiality, legal precision, and high-volume document processing. Harvey's Vault product — a document repository capable of indexing up to 100,000 documents per client — is purpose-built to operate entirely within a firm's secure environment, a non-negotiable for most large legal operations.
Tenet: Taking Ownership of the Model Layer
The most significant aspect of this round may not be the amount raised, but what it funds. Harvey has unveiled Tenet, its first post-trained LLM, developed in partnership with Fireworks Research. The model builds on Kimi K3 — the open-weight model from Moonshot AI released in July 2026 — a Mixture of Experts architecture mobilizing 104 billion active parameters per inference across a total of 2.8 trillion parameters. Post-trained via asynchronous reinforcement learning on long-form legal tasks, Tenet delivers a 20% performance gain over its base model on contract processing, and outperforms several leading general-purpose models on Harvey's internal evaluations.
Alongside Tenet, Harvey launched Harvey LAB: a proprietary legal benchmark covering approximately 1,200 tasks, designed to measure how well LLMs handle real legal work — not abstract reasoning exercises stripped of domain context.
What Harvey's Trajectory Tells Us About Vertical SaaS
Harvey's growth illustrates a broader inflection point in B2B SaaS: vendors that reach critical mass in a regulated vertical increasingly internalize their AI layer, for three compounding reasons.
First, confidentiality: in legal practice, routing client documents through a third-party API remains a regulatory liability that most firms cannot absorb. Second, performance: a model trained on legal corpora consistently outperforms a general-purpose one on domain-specific tasks where precision is non-negotiable. Third, defensibility: controlling the LLM means controlling the value that competitors — and hyperscalers — cannot easily replicate.
For General Counsel and CIOs navigating digital transformation, the question is no longer whether AI tools belong in the document workflow. It's which vendor to bet on — and what it means to build core legal operations around a platform whose proprietary model is quietly becoming the operational backbone of an entire practice.