RESEARCH

Harvey AI SEC Form D Review: $550.3M Sold at a $15.5B Valuation as Legal AI Moves Beyond OpenAI

Harvey AI SEC Form D Review: $550.3M Sold at a $15.5B Valuation as Legal AI Moves Beyond OpenAI

HARVEY AI SEC FORM D REVIEW

Harvey AI Corp's September 2026 Form D provides a more precise picture of its latest financing than the round-number headlines. The Delaware corporation reports an indefinite Rule 506(b) equity offering, a first sale on August 28, 2026 and exactly $550,299,446 sold to 28 investors. The filing reports no sales commissions or finder fees, a $0 minimum investment and no planned payments of offering proceeds to the listed related persons. Harvey declined to disclose its revenue range. Three days earlier in practical fundraising terms, the company announced what it described as a $550 million round at a $15.5 billion valuation, co-led by Diffusion and Lightspeed Venture Partners, with participation from existing investors including Sequoia, Kleiner Perkins, Andreessen Horowitz, Coatue, Conviction, Elad Gil, Evantic, GIC, Goldman Sachs Alternatives, Verified Capital and WNDR, plus new investors including Sapphire Ventures and Whale Rock. The SEC and company disclosures are therefore consistent at the headline level, but investors should preserve the distinction between the SEC's exact amount sold, the company's rounded financing announcement and the private valuation assigned to the company.

The filing history also gives Harvey a unusually clear corporate identity chain. The issuer was incorporated in Delaware in 2022 and previously filed under the legal name Counsel AI Corp before becoming Harvey AI Corp, all under CIK 0001974654. Earlier Form D filings date back to 2023, with further offerings in January and July 2024, February 2025 and April 2026 before the September 2026 financing. The current filing identifies co-founders Winston Weinberg and Gabriel Pereyra as executive officers and directors; Sequoia partner Pat Grady and Kleiner Perkins partner Ilya Fushman as directors; John LaBarre as an executive officer and director; and Alan Ghelberg and Katie Burke as executive officers. Harvey's origin story is similarly traceable: Weinberg came from legal practice involving antitrust and securities litigation, while Pereyra had worked on large language models at Google Brain and Meta. The company began by adapting frontier generative AI to legal workflows rather than building a generic productivity chatbot and initially worked closely with OpenAI to develop custom legal capabilities.

The scale of Harvey's commercial adoption has changed dramatically over a relatively short period. In early 2025, Harvey reported that its customer count had grown from 40 to 235 during 2024 while ARR increased approximately fourfold. By August 2025, the company said it served more than 500 customers in 54 countries, generated more than $100 million in ARR, employed roughly 350 people and was used by 42% of Am Law 100 firms. In 2026 the expansion accelerated further: Harvey said one quarter alone added more than $100 million in ARR, while its September financing announcement stated that approximately 80% of the Am Law 100 were using Harvey alongside five Fortune 10 companies. Current company materials go further, describing more than 200,000 lawyers across more than 2,400 organizations in 70 countries. Named deployments include major law firms and enterprises such as A&O Shearman, Dentons, CMS, Reed Smith, O'Melveny, PwC UK, Deutsche Telekom, GE Aerospace and others. These numbers are company-reported rather than audited SEC financial disclosures, but together with named customer deployments they provide significantly more operating evidence than ARR estimates circulating only through venture-market sources.

The technical story is also evolving beyond Harvey's original dependence on a single frontier-model provider. OpenAI was foundational to Harvey's early product development and publicly documented a custom-trained case-law collaboration, but Harvey now operates a broader multi-model architecture. Its current subprocessor disclosures list OpenAI, Anthropic, Google Cloud, Microsoft, AWS, Mistral, Baseten, Fireworks and other providers, while Harvey customers can access multiple frontier models depending on jurisdiction, security settings and workflow. Harvey has also begun moving further down the model stack itself. In 2026 it introduced Harvey Tenet, its first post-trained open-weight model, built from an open-weight foundation model and trained for legal work with infrastructure support from Fireworks. The company also launched Harvey LAB, a Legal Agent Benchmark for testing end-to-end legal tasks, alongside its existing BigLaw Bench evaluation framework. This shift matters because Harvey's long-term defensibility may depend less on simply providing access to models from OpenAI or Anthropic and more on proprietary legal datasets, workflow orchestration, evaluation systems, customer-specific agents, embedded implementation expertise and the ability to post-train models around highly specialized professional work.

The strategic tension is that the same frontier-model providers enabling Harvey are also expanding directly into professional services. OpenAI's September 2026 introduction of a legal-focused platform demonstrates how quickly underlying AI suppliers can move vertically into Harvey's market, while Anthropic and other developers continue improving legal reasoning performance. Harvey attempts to mitigate this risk by remaining model-agnostic and by building a substantial application, data and workflow layer around legal teams. The company says customers now operate more than 25,000 custom agents, and it has built embedded legal-engineering teams to help firms design complex workflows rather than merely purchase software seats. Customer research commissioned by Harvey reports high monthly adoption and substantial time savings among surveyed users, but investors should separate those studies from independently audited productivity evidence. Legal organizations must also verify generated work because AI-related errors and fabricated citations remain a documented industry-wide problem; platform sophistication does not eliminate attorneys' professional obligations to review work product.

The resulting investment question is therefore less about product-market fit than valuation durability. Harvey moved from a reported $3 billion valuation in its February 2025 Series D, through later financing milestones, to an $11 billion valuation in March 2026 and $15.5 billion by September 2026. The March round alone added $200 million at $11 billion, while the September financing added another roughly $550 million six months later. Harvey has reportedly raised more than $1.5 billion in total capital, giving it unusual resources to hire, train models, subsidize enterprise deployment and pursue global expansion. Yet a $15.5 billion private valuation implies expectations well beyond establishing a successful legal-software franchise. Investors should therefore examine current ARR, net revenue retention, gross margins after inference expense, implementation and legal-engineering costs, customer concentration, contract duration, model-provider economics, sales efficiency, international regulatory exposure and how quickly proprietary model capabilities reduce dependence on external frontier-model vendors.

KEY FINDINGS Harvey AI Corp filed a September 2026 Form D reporting exactly $550,299,446 sold to 28 investors in an indefinite Rule 506(b) equity offering. Harvey publicly characterized the transaction as a $550 million financing at a $15.5 billion valuation. The company was incorporated in Delaware in 2022 and previously operated legally as Counsel AI Corp, creating a continuous SEC filing trail rather than a newly formed 2026 issuer. Winston Weinberg and Gabriel Pereyra remain the core founders, while investors including Sequoia and Kleiner Perkins have board-level representation. Commercial evidence is unusually extensive: Harvey reports 200,000+ lawyers across 2,400+ organizations in 70 countries, approximately 80% penetration of the Am Law 100 and multiple large enterprise deployments. The major strategic shift is Harvey's move from being primarily an application built on frontier models toward developing proprietary legal models, benchmarks, agents and workflow infrastructure.

CORPORATE AND SEC HISTORY Current legal name: Harvey AI Corp Previous SEC legal name: Counsel AI Corp Incorporation year: 2022 Current SEC headquarters: 201 Third Street, Suite 500, San Francisco, CA 94103 Current SEC telephone: 415-579-0539 Founders: Winston Weinberg and Gabriel Pereyra Latest Form D signatory: John LaBarre, Chief Legal Officer Earlier SEC Form D filings: 2023, January 2024, July 2024, February 2025 and April 2026 Industry classification: Other Technology Investment-adviser registration: Not applicable; Harvey is an operating technology company rather than a private fund or investment adviser

LATEST FINANCING Form D first sale: August 28, 2026 Latest filing: September 2026 Security: Equity Federal exemption: Rule 506(b) Minimum investment reported: $0 Sales commissions: $0 Finder fees: $0 Related-person proceeds: $0 Revenue range: Declined to disclose Company-announced round: $550M Company-announced valuation: $15.5B Round co-leads: Diffusion and Lightspeed Venture Partners Other publicly disclosed participants: Sequoia, Kleiner Perkins, Andreessen Horowitz, Coatue, Conviction, Elad Gil, Evantic, GIC, Goldman Sachs Alternatives, Verified Capital, WNDR, Sapphire Ventures and Whale Rock

VALUATION AND FUNDING TRAJECTORY February 2025: Harvey announced a $300M Series D at approximately $3B valuation. March 2026: Harvey announced a $200M growth round at approximately $11B valuation, co-led by GIC and Sequoia. July 2026: Harvey announced additional strategic investment from Goldman Sachs Alternatives and J.P. Morgan Growth Equity Partners. September 2026: Harvey announced $550M at a $15.5B valuation. Important interpretation: Private valuation is the financing-round valuation assigned by investors; it is not SEC-certified enterprise value, publicly traded market capitalization or evidence that every prior security has equivalent liquidity.

COMMERCIAL SCALE 2024: Customer count reported growing from 40 to 235; ARR approximately quadrupled. August 2025: 500+ customers across 54 countries; $100M+ ARR; 42% of Am Law 100; approximately 350 employees. 2026: Harvey reported a quarter in which more than $100M of ARR was added. September 2026: Company said 80% of Am Law 100 law firms use Harvey and five Fortune 10 companies were customers. Current customer page: 200,000+ lawyers, 2,400+ organizations and 70 countries. Named customer evidence includes major international law firms, professional-services organizations and corporate legal departments. Important distinction: Customer counts, adoption statistics and ARR figures are company-reported metrics and are not disclosed or verified through Harvey's Form D.

PRODUCT AND MODEL PENETRATION Core platform functions: Legal research, drafting, document review, due diligence, knowledge retrieval, workflow automation and AI agents. Early model relationship: OpenAI custom model and case-law work. Current model architecture: Multi-provider environment including OpenAI, Anthropic, Google, Microsoft/Azure, AWS-hosted services, Mistral and additional inference/model providers. Custom agents: More than 25,000 customer-created agents publicly reported by Harvey in 2026. Harvey Tenet: First internally post-trained open-weight legal model announced in 2026. Harvey LAB: Legal Agent Benchmark for complex end-to-end professional tasks. BigLaw Bench: Proprietary legal-model benchmark used to compare frontier models. Strategic direction: Reduce dependence on any single frontier-model provider while building proprietary domain intelligence, workflows, benchmarks and legal-engineering expertise.

WEBSITE / ENTITY PENETRATION Legal issuer: Harvey AI Corp Brand: Harvey Previous name: Counsel AI Corp SEC legal-name match: Confirmed Founder match: Confirmed San Francisco operating presence: Confirmed SEC phone: 415-579-0539 Primary business: Enterprise legal and professional-services AI RIA / CRD status: Not applicable Major institutional backing: Strong and publicly documented Named enterprise customer evidence: Extensive Public product documentation: Extensive Security / privacy documentation: Extensive Current subprocessor disclosure: Publicly available and regularly updated

MANAGEMENT AND BOARD PENETRATION Winston Weinberg: Co-founder, CEO, executive officer and director. Gabriel Pereyra: Co-founder, executive officer and director. Pat Grady: Director; associated with Sequoia Capital. Ilya Fushman: Director; associated with Kleiner Perkins. John LaBarre: Chief Legal Officer; executive officer, director and latest Form D signatory. Alan Ghelberg: Executive officer. Katie Burke: Executive officer. 2026 senior additions publicly announced include leadership across revenue, product, marketing, security, strategy and research functions as Harvey scales internationally.

COMPETITIVE LANDSCAPE Harvey competes across several overlapping categories rather than against one direct rival. Legal-information incumbents such as Thomson Reuters and LexisNexis possess proprietary content, research infrastructure and embedded enterprise distribution. AI-native companies including Legora compete directly for law-firm workflows. OpenAI and Anthropic are both model suppliers and potential vertical competitors as they expand professional-domain products. Microsoft and other enterprise platforms can bundle AI into existing customer infrastructure. Harvey's defense is therefore built around deep legal workflow integration, proprietary evaluation data, embedded legal engineers, custom agents, customer knowledge integration and the ability to use multiple frontier models rather than depending on one provider.

CORE INVESTOR QUESTIONS Investors should reconcile the $15.5B valuation against Harvey's current ARR rather than historical $100M+ figures; obtain current gross margin after inference and cloud costs; determine how much of reported ARR reflects committed multi-year contracts versus shorter subscriptions; measure net dollar retention and expansion within major law firms; identify concentration among Am Law 100 and large enterprise customers; review customer-acquisition cost and embedded legal-engineering expense; quantify OpenAI, Anthropic and other model-provider costs; examine whether Harvey Tenet materially reduces unit inference costs or increases proprietary differentiation; review international data residency and professional-secrecy requirements; determine employee stock-option dilution after repeated financings; and analyze the likely IPO or secondary-market pathway required to generate returns at a $15.5B private valuation.

CORE RISKS Harvey's rapid growth substantially reduces basic product-market-fit risk but increases valuation risk. A company valued at $15.5B must support continued exceptional revenue growth and durable enterprise retention to justify that price. Legal AI also operates in a high-liability environment where inaccurate citations, misinterpretation or confidential-data failures can create serious consequences. Lawyers remain responsible for verifying AI-generated work, meaning automation cannot eliminate professional review. Dependence on third-party model providers creates cost, product-roadmap and competitive risks even as Harvey develops its own post-trained models. Legal-information incumbents control large proprietary content libraries, while frontier-model companies can move directly into vertical legal offerings. Repeated financing rounds at sharply rising valuations may also increase expectations for a very large eventual public-market or acquisition outcome.

SEC SNAPSHOT Previous Name: Counsel AI Corp Latest SEC File No.: 021-598179 Year Formed: 2022 Principal Office: 201 Third Street, Suite 500, San Francisco, CA 94103 Phone: 415-579-0539 Industry: Other Technology Exemption: Rule 506(b) Security: Equity Offering Amount: Indefinite Amount Sold: $550,299,446 Investors: 28 First Sale: August 28, 2026 Minimum Investment Reported: $0 Sales Commissions: $0 Finder Fees: $0 Revenue Range: Declined to disclose CEO / Co-Founder: Winston Weinberg Co-Founder: Gabriel Pereyra Latest Form D Signatory: John LaBarre Company-Announced Financing: $550M Company-Announced Valuation: $15.5B

PRIMARY EVIDENCE REVIEWED SEC Form D for Harvey AI Corp filed September 2026. SEC historical Form D filings for Harvey AI Corp / Counsel AI Corp. Harvey September 9, 2026 $550M financing announcement. Harvey March 2026 $200M financing announcement. Harvey February 2025 Series D announcement. Harvey customer and deployment disclosures. Harvey three-year operating update and ARR disclosure. Harvey 2026 executive and product announcements. Harvey model, benchmark and agent-product materials. Harvey public subprocessor list. OpenAI public case study describing its custom-model collaboration with Harvey. Public reporting on Harvey's September 2026 financing and legal-AI competitive landscape.

IMPORTANT FORM D NOTICE Harvey's latest Form D should not be summarized as a fixed $550 million offering. The SEC filing identifies the total offering amount as indefinite and reports exactly $550,299,446 already sold to 28 investors. Harvey separately announced a $550 million financing at a $15.5 billion valuation. The $15.5 billion figure is a private financing valuation, not a valuation independently determined by the SEC. Likewise, company-reported ARR, customer adoption and usage statistics should remain separate from SEC filing facts. Form D is an exempt-offering notice and does not constitute SEC approval of Harvey, its products, financial performance, valuation or investment merits.

Important Form D notice: A Form D filing is a notice filing for an exempt securities offering. It does not mean that the U.S. Securities and Exchange Commission has approved, licensed, endorsed, or verified the issuer or the offering. Readers should verify information through official SEC sources and conduct their own due diligence.
Verification note: SEC.gov and the relevant regulator's official records remain authoritative. This site's research is independent editorial content.