Harvey closed a $550 million growth round on September 9, 2026, at a $15.5 billion post-money valuation, up from its previous $11 billion mark. Diffusion and Lightspeed Venture Partners co-led the raise, which included participation from Sequoia Capital, Kleiner Perkins, Andreessen Horowitz, Coatue, Conviction, and GIC. The company claims the platform is now used by 80 percent of Am Law 100 firms and five of the Fortune 10 in-house legal teams. These vendor-stated figures have not been independently audited, but they support a thesis that investors are pricing Harvey as load-bearing infrastructure for legal work rather than a swappable productivity add-on. At roughly 39x annual recurring revenue, the multiple demands that enterprise legal teams consolidate onto a single AI provider instead of fragmenting across use cases. The real test is whether firms renew and expand their usage over the next 12 months, or if any major client publicly exits the platform.
Custom Models and Deep Integration
Harvey partnered with OpenAI to build a custom-trained model specifically for case law research because standard retrieval-augmented generation systems failed to provide the depth required for complex legal arguments. The team injected the equivalent of 10 billion tokens of data into the base model, starting with Delaware case law before expanding to all U.S. jurisdictions. This collaboration allowed Harvey to move beyond simple question answering toward generating "ammo" for legal arguments, a task where general foundation models lack the necessary specialized knowledge. The partnership extends to using GPT-6 Astra for drafting, which improves document formatting and context awareness by processing more information simultaneously. A memory panel lets lawyers encode preferences like prioritizing EDGAR sources or using numbered lists, ensuring outputs align with individual workflows.
The platform integrates deeply with existing firm infrastructure through Vault, which syncs with iManage, SharePoint, Box, and Google Drive to ground answers in specific precedent banks. A strategic alliance with LexisNexis, announced in June 2025, brings Shepard’s Citations and primary law directly into Harvey via the "Ask LexisNexis" feature. Firms like Stinson LLP adopted this combination firmwide in February 2026, while DLA Piper tested it early. In October 2026, the companies released an agentic Motion to Dismiss workflow that reads complaints, proposes ranked arguments, and drafts motions with lawyer checkpoints. This integration addresses a critical gap where generic AI tools often produce confident but inaccurate citations. However, Harvey does not publish pricing, and third-party estimates suggest mid-market deals cost $1,200 to $1,500 per seat monthly with a 20-seat minimum. That opaque commercial structure excludes smaller firms and forces buyers to negotiate multi-year terms without public benchmarks. Will the high switching costs created by deep document integration keep large firms locked in, or will the premium price drive them back to cheaper, self-serve alternatives?
