Legal AI Startup Harvey Hits $11 Billion Valuation

Investment Pours Into Legal AI Companies

Harvey, a legal AI company founded in 2022, announced on March 25, 2026 that it had raised $200 million at an $11 billion valuation. GIC and Sequoia Capital co-led the round. About three and a half years after its founding, the company has become a decacorn — a private company valued at over $10 billion.

Harvey builds AI services that lawyers use to review contracts, search case law, and draft documents. It’s a company that helps handle legal-knowledge-intensive work, tailored to each client firm’s materials and workflows.

What interests me is how law firms’ billing practices will change as this company grows. When an organization that gets paid by the hour adopts a tool that cuts down on working hours, how will it divide the value of the time saved with its clients?

Valuation and Customer Numbers Grew Together

Harvey’s valuation, as recognized in successive funding rounds, climbed fast. Here’s the timeline of major announcements.

• December 2023, Series B: $715 million
• July 2024, Series C: $1.5 billion
• February 2025, Series D: $3 billion
• June 2025, Series E: $5 billion
• December 2025 investment: $8 billion
• March 2026 investment: $11 billion

Compared to December 2023, that’s a jump of more than 15x in roughly 2 years and 3 months. The business itself scaled up too. Harvey announced in August 2025 that its annual recurring revenue (ARR) had passed $100 million, and on March 24, 2026, Fast Company reported it had crossed $190 million. ARR is a metric that annualizes recurring contract revenue — it’s distinct from revenue or profit actually recognized within a given year.

According to a company announcement on March 25, Harvey’s customers include a majority of the AmLaw 100 (the 100 largest U.S. law firms), more than 500 in-house legal teams, and 50 asset managers, spanning customers in 60 countries. In the same announcement, Sequoia’s Pat Grady noted that more than 100,000 lawyers now use Harvey. The fact that Sequoia has co-led three separate funding rounds also signals investor confidence in the company’s long-term growth.

The English draft matches the Korean source accurately with correct numbers, glossary terms, structure, and no Hangul remaining. No corrections needed.

Document work and cost pressure are driving AI demand

In legal tech company Clio’s 2024 Legal Trends Report, 79% of surveyed legal professionals said they were using AI in some form. The previous year’s figure was 19%. Since the survey asks about individual AI use, this doesn’t mean 79% of law firms have adopted AI systems firm-wide.

This survey alone can’t tell us whether legal adoption is faster than in other industries. What it does confirm is that demand to try AI in legal work is growing.

I understand that demand along three dimensions.

First, there’s a lot of document work involved. M&A due diligence¹, contract review, case law research, and compliance checks all require comparing and organizing large volumes of material — tasks where a language model can draft text or locate relevant passages. Still, experts need to verify whether statutes and precedents were applied correctly.

Second, legal service costs are high. According to the 2026 Report on the State of the US Legal Market by Thomson Reuters and Georgetown Law, the average standard hourly rate at AmLaw 100 firms in 2025 exceeded $1,000. This figure aggregates the rates firms quote — it differs from actual realized revenue after discounts, or from the rates individual junior associates charge. For corporate legal departments, whether costs can be lowered while maintaining review quality becomes a key adoption criterion.

Third, recurring procedures can be organized by client. Examples include finding specific clauses in contracts or summarizing due diligence materials into a standard format. Harvey announced in March 2026 that customers had built more than 25,000 custom AI agents on its platform. That number reflects agents created — not a performance measure of how many complete legal work without human review.

Law firms are spending money on technology too. The same Thomson Reuters/Georgetown report put the year-over-year growth in 2025 tech spending at 9.7%. This figure is based on data through November, and the report characterized it as a very rapid rate of increase.

When Working Hours Drop, What Happens to Legal Fees

One issue worth examining with AI adoption is how much firms charge clients.

The method widely used at law firms is the billable hour². Firms multiply time spent on a client’s matter by an hourly rate to generate the bill. Thomson Reuters’ 2026 report notes that roughly 90% of legal spend tracked in Legal Tracker is still billed on a time basis. Increased adoption of technology hasn’t immediately changed how firms bill.

Let’s run a simple hypothetical. A lawyer billing $300 an hour who spends 25 hours drafting a brief would charge $7,500. If the same work gets done with AI in 10 hours and the rate stays the same, that comes to $3,000. To keep the per-matter charge at $7,500, the hourly rate would need to rise to $750. This calculation is an illustration based on the assumption that only the working hours change — it’s not experimental evidence that every type of legal work can be shortened this much.

Say, for instance, that drafting time for an NDA³ drops. A client might well ask whether that reduced time gets reflected in the bill. The firm, for its part, would likely want the fee to reflect the accuracy of the review, the liability involved, and the difficulty of the case. Either way, rather than setting a discount rate simply because AI was used, both sides need to look at the actual scope of work and the terms of the engagement.

Clio’s 2024 report found that up to 74% of hourly-billed work could potentially be automated. That figure estimates the scope of tasks technology could affect — it shouldn’t be read as meaning 74% of the work, or that much revenue, has already vanished.

The effect of time savings on profit also needs separate scrutiny. If the hourly rate and the number of matters stay constant, reduced billable hours could mean lower revenue. But the outcome changes if firms cut work they previously couldn’t bill clients for anyway, or use the freed-up time to take on other matters. What actually remains as profit has to be calculated including AI usage fees and review costs.

Other billing structures are available too. Firms can set a fixed fee per case or document, or charge a flat monthly rate for a defined scope of advisory work. These alternative fee arrangements are called AFAs⁴. Under a fixed fee, a firm’s profit can rise in proportion to the time it saves — but the firm also bears the risk if the work turns out to be heavier than expected. Fees tied to outcomes, meanwhile, are permitted to varying degrees depending on the country and case type, so they require separate review.

Against this backdrop, Harvey has set out a goal of becoming a platform that handles a wide range of legal team tasks. Rather than stopping at producing a single draft, the direction is toward handling downstream work — document review, collaboration, and more — within the product itself.

Oswarld’s Lens

I think Harvey’s $11 billion valuation reflects high growth expectations. Whether that figure is justified is something I can only judge once I know more about customer retention, costs, and actual profit. Raising funding successfully shouldn’t be read as proof that the business is profitable.

What interests me from a GTM strategy standpoint is the process by which customers keep using this product.

At Harvey, legal engineers work alongside client teams to help them build agents and workflows tailored to their organization. I think this kind of onboarding support can be part of the company’s competitive edge. The more closely the product fits a customer’s document formats and review sequences, the more reason there is to keep using it — and the more re-configuration and retraining is required if they switch to another tool. That said, this doesn’t mean every client firm has an engineer physically stationed on-site.

Law firm staffing models also need to be considered here. When multiple associates⁵ under a partner handle the actual casework — the leverage model⁶ — how much time goes into drafting and preliminary research really matters. If AI cuts down on that time, firms may need to rethink how they allocate work and train junior lawyers. Still, that doesn’t mean we can conclude Harvey is a one-to-one replacement for junior lawyers. Law firms are also Harvey’s customers, and lawyers’ review and accountability still remain in place.

What ultimately matters is how firms use the time they save and what they promise their clients in return. They could charge the same amount for faster review, or lower their fees, or focus their energy on more complex advisory work. Which approach makes sense will depend on the specific matter and contract a client has entrusted to them.

Closing

Domestic legal tech companies and professional service organizations can also draw lessons from Harvey’s case by examining the following:

• Look at whether tasks can be concretely divided between what AI handles and what experts verify.
• You need to calculate not just reduced hours, but usage fees, review costs, and what gets billed to clients—all together.
• Onboarding support that fits the product to clients’ actual documents and workflows matters too.

There’s no fixed answer to whether firms should stick with hourly billing or move to flat fees. I think the organizations that come out ahead will be the ones that clearly explain what work and responsibility they’re providing—so that clients can pay with confidence even after AI enters the picture.

References & Further Reading

• Harvey, March 2026 funding announcement: valuation, customer base, and plans for custom agents and rollout support.
• Harvey, Series B, Series D, Series E, December 2025 investment: funding announcements at each stage.
• Harvey, Three-year anniversary post, 8.4.2025: announcement of surpassing $100 million in ARR.
• Fast Company, How Harvey made its legal AI tools indispensable, 3.24.2026: coverage of product improvements and ARR surpassing $190 million.
• Clio, 2024 Legal Trends Report release: survey on legal professionals’ AI use and the automation potential of their work.
• Thomson Reuters Institute & Georgetown Law, 2026 Report on the State of the US Legal Market: the relationship between hourly rates, tech spending, billing models, and AI adoption.

The author is Oswarld (Kwangseob Ahn). Current roles: Adjunct Professor at Sejong University, Strategy Consultant at INLEVEL9. Career, research, books, and recent work are kept current on the About page. Latest · July 2026: HEMA-2: A Consolidation-Aware Tri-Memory Architecture with Multi-Channel Scheduling for Lifelong Conversational AI.

Footnotes

1. Due Diligence: The process of examining a target company’s finances, legal standing, and operations before an M&A deal or investment. ↩

2. Billable Hour: The time spent on client work that is recorded for billing purposes. Fees are calculated by multiplying this time by an hourly rate. ↩

3. NDA (Non-Disclosure Agreement): A contract stipulating that information shared during negotiations or collaboration will not be disclosed externally. ↩

4. AFA (Alternative Fee Arrangement): A fee arrangement other than hourly billing, including flat fees and monthly retainers. ↩

5. Associate: A lawyer at a law firm who is not a partner. ↩

6. Leverage Model: An organizational and revenue structure in which a small number of partners oversee the work of multiple associates to deliver services. The cost, billing rate, and workload of each staff member affect profitability. ↩