AI Adoption in Am Law 100: What 42% Harvey Penetration Tells Us
Half of the largest US law firms use Harvey, but the adoption pattern reveals as much about barriers as it does about momentum.
As Harvey reached its three-year anniversary in August 2025, the company hit significant milestones: $100 million ARR, over 500 customers globally, and 42% penetration among AmLaw 100 firms. By September 2025, that figure reached 50%.
More than half of the largest US law firms now use Harvey. Understanding why half adopted and why half did not provides practical guidance for firms evaluating AI deployment.
The Adoption Numbers
Harvey's Growth Trajectory
Harvey launched in November 2022 with zero revenue. The growth curve:
- End 2023: $10 million ARR
- End 2024: $65.8 million ARR (558% year-over-year increase)
- August 2025: $100 million ARR
- Late 2025: Estimated $195 million ARR (per Sacra)
Customer count expanded from 40 in early 2024 to over 500 customers across 54 countries by mid-2025.
AmLaw 100 Penetration
The 42% (and later 50%) AmLaw 100 adoption figure represents significant market validation. These are sophisticated buyers with resources to evaluate options and significant technology budgets.
Named customers include Latham & Watkins, Willkie Farr & Gallagher, and Duane Morris. CMS expanded its Harvey rollout to 7,000+ lawyers across its global platform in more than 50 countries.
Valuation Context
Harvey's valuation progression in 2025:
- February 2025: $3 billion (Series D)
- June 2025: $5 billion (Series E)
- December 2025: $8 billion (Series F)
Investors include EQT, WndrCo, Sequoia, Kleiner Perkins, and Andreessen Horowitz. The legal AI space attracted significant capital, with Harvey capturing the dominant share among law firm-focused tools.
Use Cases by Practice Area
Where AI Adoption Concentrates
Based on industry reporting and adoption patterns, AI utilization varies significantly by practice area:
High Adoption:
- Contract review and due diligence
- Legal research and memoranda
- Document drafting (initial versions)
- Corporate transactions (standard provisions)
- Regulatory compliance research
Moderate Adoption:
- Litigation document review
- Discovery analysis
- Brief drafting support
- Client communication drafting
Lower Adoption:
- Complex negotiation strategy
- Trial preparation requiring judgment
- Client counseling
- Matters with unusual fact patterns
The pattern reflects AI's current capabilities: strong at volume processing and pattern matching, less suited to complex judgment and strategic decision-making.
Transaction Versus Litigation
Transactional practices have generally adopted AI faster than litigation practices. Possible explanations:
- Contract analysis maps well to AI extraction capabilities
- Due diligence involves large document volumes suited to AI processing
- Transaction timelines create pressure to accelerate review
- Litigation involves more adversarial dynamics where AI limitations carry higher risk
Why 58% (Then 50%) Have Not Adopted
Accuracy Concerns
According to Thomson Reuters research, "demonstrated accuracy" remains the biggest barrier to AI investment. 91% of professionals said computers should be held to higher standards than humans, with 41% requiring 100% accuracy before using AI without human review.
This standard is not currently achievable. AI tools make errors that require human verification. Firms with low tolerance for any error may conclude that verification overhead negates efficiency gains.
Cost and Budget Constraints
Enterprise AI platforms require significant investment beyond subscription fees:
- Platform licensing (often tiered by usage)
- Implementation and integration services
- Training and change management
- Ongoing administration and policy development
- Verification workflow overhead
Not every firm has budget for sophisticated AI deployment. Not every firm concludes the investment delivers sufficient return.
Data Privacy and Security
41% of respondents in Embroker's 2024 survey cited data privacy concerns regarding AI adoption. Larger firms have addressed this through enterprise agreements with tools like Harvey that provide confidentiality protections. Smaller firms often lack leverage to negotiate favorable terms or resources to evaluate data security adequacy.
Confidentiality obligations are strict. Lawyers cannot input privileged information into systems that might use it for training or expose it to unauthorized access.
Ethical and Trust Concerns
Recent sanctions cases (Mata v. Avianca, Morgan & Morgan, Ko v. Li) have created caution. Firms may prefer to wait for clearer regulatory guidance rather than risk being test cases for enforcement actions.
The billable hour conflict also creates institutional resistance. If AI enables completing work in one hour that previously took five, time-based invoices shrink by 80%. Firms whose economics depend on hour accumulation face structural tension with efficiency-enabling tools.
Integration and Implementation Complexity
Even firms that want to adopt face practical barriers:
- Transitioning from legacy systems
- Security credential configuration
- Training program development
- Integration with existing document management
- Policy development and governance
Implementation often proves more complex than anticipated. Firms may start pilots but struggle to achieve general availability.
The Expectations-Reality Gap
Bloomberg Law's 2025 State of Practice Survey found that law firm lawyers reported smaller-than-expected changes from AI in every workload and operational category. The hype cycle created expectations that current tools do not fully satisfy.
MIT economist Mert Demirer observed: "I will expect some impact on the legal profession's labor market, but not major... the law's low risk tolerance, plus the current capabilities of AI, are going to make that case less automatable at this point."
Lessons for Mid-Market Firms
Right-Size Expectations
AmLaw 100 firms have resources for enterprise AI deployment including dedicated technology teams, significant budgets, and negotiating leverage with vendors. Mid-market firms should not assume the same tools at the same scale make sense.
Consider:
- What volume of work would flow through AI tools?
- What verification infrastructure exists or needs building?
- What integration requirements exist with current systems?
- What governance capacity exists to manage AI deployment?
Start with Clear Use Cases
Successful adoption starts with specific use cases rather than general "AI deployment." Identify:
- High-volume, repetitive work that consumes associate time
- Research tasks with clear verification paths
- Document review projects with definable scope
- Drafting needs where AI first drafts reduce time to completion
Build verification workflows and measure results before expanding scope.
Evaluate Total Cost
Subscription pricing represents only part of total investment. Factor in:
- Implementation and integration
- Training and change management
- Verification workflow overhead
- Governance and policy development
- Ongoing administration
For some mid-market firms, the full investment may not generate positive returns at current volumes.
Watch the Large Firm Experiment
AmLaw 100 firms are effectively running a large-scale experiment with legal AI. Mid-market firms can observe:
- Which use cases deliver value versus disappointment
- What verification workflows prove effective
- How governance frameworks mature
- Where regulatory enforcement focuses
The cost of being a fast follower is often lower than being an early adopter.
Key Takeaways
- Harvey reached 50% AmLaw 100 penetration by late 2025, with $100M+ ARR and $8B valuation
- Adoption concentrates in contract review, due diligence, and research; lower in complex litigation and strategic work
- 91% of professionals require computers to meet higher accuracy standards than humans; 41% require 100% accuracy
- Size disparity: 39% adoption at firms with 51+ lawyers versus 20% at smaller firms
- Barriers include accuracy concerns, cost, data privacy, sanctions risk, and implementation complexity
For a deeper analysis of adoption barriers, see Best Law Firms' investigation into what is really stopping firms, and the RSGI/Harvey adoption report examining the impact of legal AI on firm operations.