Verisk's 45% Drawdown Masks a Pure-Play Insurance Data Machine Quietly Shipping Agentic AI

Verisk Analytics has shed nearly half its market cap in 12 months on P&C market normalization fears and AI disruption anxiety, yet the company just posted a consensus beat, is integrating with Anthropic's Claude via MCP connectors, and runs 400+ catastrophe models on an open exchange platform that's becoming the Bloomberg Terminal of insurance risk.

VRSK · Industrials · May 27, 2026

S&P 500 Position

Classified under Industrials / Research & Consulting Services, Verisk sits alongside peers like Gartner, Dun & Bradstreet, and IHS Markit (now S&P Global). Within the narrower insurance analytics vertical, its direct competitors — CoreLogic (private), CCC Intelligent Solutions (~$7B market cap), Guidewire (~$17B) — are either smaller or focused on different layers of the insurance stack. Moody's RMS is the nearest competitor in catastrophe modeling but operates within a much larger parent.

Index Weight: ~0.05% | Rank: Approximately 350-400 in the S&P 500 by market capitalization (~$22.5B)

Company Overview

Verisk is now a pure-play insurance analytics platform after completing a multi-year divestiture arc that shed Wood Mackenzie ($3.1B to Veritas Capital, February 2023), its financial services arm, environmental health & safety businesses, and Atmospheric and Environmental Research (December 2024). What remains is a deeply embedded data infrastructure layer for the global P&C insurance market — every top-100 U.S. P&C insurer for the lines Verisk serves is a client, and approximately 70% of revenue flows from U.S. primary carriers. The business model is subscription-first (>80% of revenue), regulatory-grade, and sticky: Verisk maintains the statutory datasets that underpin rate filings, ISO circulars, and loss-cost calculations that insurers legally need. The competitive moat is not any single product but the contributory data flywheel. Insurers feed claims, premium, and loss data into Verisk's databases; Verisk normalizes, enriches, and redistributes that data as actuarial products, catastrophe models, and underwriting signals. CoreLogic competes on property data, CCC Intelligent Solutions owns auto claims workflows, Guidewire and Duck Creek run core policy administration, and Moody's RMS contests catastrophe modeling — but none of them replicate the breadth of Verisk's regulatory and statistical position across lines. The disintermediation risk from Guidewire and Duck Creek bundling analytics into their platforms is real but slow-moving: regulatory data dependencies and model validation cycles create switching costs that pure software vendors struggle to replicate. The stock's 45% drawdown over the trailing twelve months reflects three converging anxieties: P&C premium growth decelerated to an annualized 3.8% in early 2026 (from 6.5% in late 2025), a sector-wide software and data analytics multiple compression hit the entire group, and investors fear that large language models could commoditize Verisk's data advantage. CEO Lee Shavel has pushed back on the AI disruption thesis, citing proprietary data moats, but the market is unconvinced so far.

Products & Revenue

Verisk operates as a single Insurance segment with two revenue solution lines: Underwriting Solutions and Claims Solutions. Underwriting Solutions — rate filings, actuarial data, catastrophe models, commercial property analytics, and life/annuity producer licensing — drives roughly 70% of revenue and grew 7.7% in FY2025. Claims Solutions — Xactimate property estimating, auto claims workflows, fraud detection, and subrogation analytics — accounts for the remaining ~30% and grew 4.1% in FY2025. Subscriptions represent over 80% of total revenue, with transactional revenue from per-claim Xactimate usage and event-driven catastrophe model runs making up the balance. The divestiture of Verisk Marketing Solutions removes approximately $68M of revenue from the 2026 run rate.

Underwriting Solutions (70.5%): ISO statistical and rating content, actuarial data products, catastrophe modeling (AIR and the Verisk Model Exchange with 400+ peril models), commercial property underwriting tools, life insurance producer licensing (SuranceBay), and regulatory filing services. This is the regulatory-grade data engine.

Claims Solutions (29.5%): Xactimate/XactAnalysis property estimating platform (industry standard for homeowners claims), auto physical damage and casualty claims analytics, fraud detection, and subrogation services. XactAI and XactGen represent the agentic AI layer being built on top of this workflow.

Segment percentages based on Q1 2026 10-Q filing ($552.1M Underwriting, $230.5M Claims, $782.6M total). FY2025 full-year revenue was $3,073M per the 10-K.

Leadership

Lee M. Shavel

CEO since 2023. Shavel became President in May 2022 and CEO in January 2023 after serving as CFO. He previously held senior roles at Nasdaq, Inc. and MSCI. His strategic thesis is pure-play insurance analytics at scale — he orchestrated the Wood Mackenzie sale, the financial services divestiture, and the aggressive $1.5B accelerated share repurchase that pushed equity into deficit. Total compensation is approximately $12.07M, 91.7% performance-linked.

Saurabh Khemka, President, Underwriting Solutions: Leads the largest revenue line (~70% of total), overseeing Core Lines, Global Underwriting Data and Analytics Solutions, Global Life Solutions, and the now-divested Marketing Solutions business. He owns the catastrophe modeling and ISO statistical product roadmap.

Pradip Patiath, Board Director (appointed May 2026): The latest addition in a board refresh that has seated nine new directors since 2022. His appointment follows extensive shareholder engagement on AI oversight, board composition, and governance reform.

The AI Angle

Agentic AI Meets Regulatory-Grade Insurance Data

Verisk has more than 35 AI-powered projects and solutions deployed across internal operations and client-facing products. The flagship AI initiative is the XactAI / XactGen pipeline: XactAI launched in September 2025 as a computer-vision and ML layer that accelerates property damage estimation inside the Xactimate platform. XactGen followed as the agentic AI evolution — it autonomously gathers content from multiple sources (imagery, contractor pricing databases, policy terms) and generates near-complete exterior and interior estimates for faster claim settlement. This is not a chatbot bolted onto a dashboard; it is an autonomous workflow agent operating within the most widely used property claims estimating system in the U.S. The second major AI product is the Commercial GenAI Underwriting Assistant, launched September 2025 as a cloud-based generative AI tool designed to modernize commercial property underwriting. It synthesizes Verisk's proprietary hazard data, building characteristics, and loss history into underwriter-ready risk assessments. The product targets the commercial lines segment where manual data gathering and PDF parsing still dominate workflows — a classic GenAI use case where retrieval-augmented generation over structured proprietary data delivers immediate productivity gains. Verisk's infrastructure strategy leans toward partnership over foundational model training. In May 2026, the company deployed Model Context Protocol (MCP) connectors that make Verisk's insurance analytics available directly inside Anthropic's Claude. This enables underwriting and claims professionals to query regulatory-grade data through conversational natural-language interfaces within enterprise AI environments. The MCP approach is architecturally interesting: rather than building a proprietary LLM, Verisk positions its data as a first-class tool that frontier models can invoke — similar to how Stripe or Twilio became API primitives for application developers. The risk profile is two-sided. On one hand, Verisk's proprietary contributory databases — built over decades from insurer submissions — cannot be replicated by an LLM trained on public internet data; regulatory filings, ISO circulars, and loss-cost calculations require structured, audited datasets. On the other hand, the market is pricing in a scenario where frontier AI reduces the marginal value of Verisk's analytics layer by enabling insurers to build more in-house. CEO Shavel has addressed this directly, emphasizing data moats, but the stock's 45% drawdown suggests investors want proof, not promises.

Financial Snapshot

Revenue (TTM): $3.10B — TTM ending Q1 2026 | Net Income: $910M — TTM net income

Margins: EBITDA margin 55.7% (Q1 2026), net margin 29.3% (TTM)

Verisk's capital allocation is unusually aggressive for a data analytics company. The $1.5B accelerated share repurchase plus $126M of open-market buybacks, funded by new senior notes and a $500M term loan, pushed total debt to $4.48B and shareholders' equity into deficit. This explains the negative debt/equity ratio and stratospheric P/B. Free cash flow was $1.19B in FY2025, giving ample coverage for debt service and dividends. The subscription-first model (>80% recurring) provides cash flow visibility that supports this leverage posture. FY2026 guidance of $3.19B–$3.24B revenue and $7.45–$7.75 adjusted EPS was reaffirmed after Q1, implying mid-single-digit growth despite the Marketing Solutions divestiture headwind.

1-Year Performance

VRSK trades at $171.50, down 45.3% over the trailing twelve months. The stock hit a 52-week low of $161.67.

Three catalysts drove the decline: an October 2025 guidance cut (FY2025 revenue lowered to $3.05B–$3.08B from $3.09B–$3.13B, triggering a 10%+ single-session drop), a March 2026 sell-off of 5.2% on P&C premium growth deceleration data, and persistent AI disruption anxiety that has compressed the multiple from historical 40x+ P/E to 26x. The Q1 2026 beat ($1.82 adjusted EPS vs. $1.77 consensus) and guidance reaffirmation have not yet reversed sentiment. Institutional selling is evident — King Luther Capital sold ~40K shares and Legal & General sold ~44K shares in the most recent reporting period.

Recent News

Fun Fact: The Verisk Model Exchange, which now hosts 400+ peril models from 20+ third-party providers, functions as a vendor-neutral catastrophe risk marketplace — essentially an app store for natural disaster math. Every model runs on a consistent financial engine so insurers can compare, say, a KatRisk flood model against a Verisk AIR model side-by-side with identical assumptions. The platform emerged from Verisk's realization that its own proprietary catastrophe models would gain market credibility by inviting competitors onto the same infrastructure — a counterintuitive bet that openness strengthens the platform's lock-in.