When a company beats earnings estimates by 26%, raises full-year guidance, and still loses more than 13% of its market value in a single session, the market is not reacting to the quarter. It is reassessing the decade.
Gartner closed November 3, 2026 at $204.22, down sharply, despite beating analyst expectations on both earnings and revenue and raising its full-year revenue forecast. Adjusted earnings came in at $3.59 per share against a consensus of $2.84, and revenue of $1.5 billion edged past the $1.49 billion forecast. On paper, a clean beat. The stock told a different story.
The anxiety behind the selloff is not mysterious. Gartner’s core product is advice: structured, credentialed, subscription-packaged advice sold to CIOs, CEOs, and senior technology buyers. That is precisely the category that large language models have made cheap. Questions on the earnings call centered on the slowdown in ex-federal contract value growth and whether Gartner’s sales headcount is sufficient for future growth. Neither concern has an easy answer right now.
Gartner has spent decades building its brand around independent research and proprietary frameworks, but after years of exceptional growth, contract value is now expanding much more slowly. FX-neutral contract value growth has slowed from double-digit levels in 2022 to low single digits in recent quarters. Management pointed to sequentially improving numbers, and contract value growth did accelerate for the third consecutive quarter, driven in part by demand for AI research itself. But sequential acceleration from a low base is not the same as structural resilience, and investors apparently drew that distinction sharply.
Not All Information Services Are Equal
The selloff invites a more important question: within the information-services space, which businesses are genuinely substitutable by AI, and which ones are not? The answer splits the sector cleanly.
At one end sits Gartner. Its value proposition is research synthesis and advisory judgment. Both are things a capable AI model can approximate at a fraction of the cost of an analyst seat. Morningstar, reviewing FactSet earlier this year, noted that the downgrade to no moat reflects a dynamic and uncertain future as generative AI is adopted, and that FactSet’s workstation and software offerings face meaningful competitors as the rise of generative AI may alter the competitive and client landscapes. Morningstar also downgraded its moat rating to none from narrow for SS&C.
At the other end sit the ratings agencies. Moody’s and S&P Global have traded this year as a blended average of two different businesses: a deeply moated ratings franchise that is largely insulated from AI disruption, and a data and analytics operation that is squarely in its crosshairs. Credit ratings are often based on nonpublic information provided by management teams, which no AI model can access or replicate on its own. In credit ratings, Moody’s, S&P, and Fitch collectively account for roughly 90% to 95% of global ratings, with regulatory licenses acting as barriers to entry. An AI can run a ratings methodology. It cannot replace the regulatory standing that makes a rating legally meaningful.
After considering factors such as competitive positioning, pricing models, and the prevalence of proprietary data, Morningstar maintained wide moat ratings for Broadridge, Fair Isaac, Moody’s, MSCI, S&P Global, and Verisk. Gartner was not on that list. The distinction is not arbitrary.
The Verdict
In an AI era, models are only as good as the data they are trained and grounded in; curated datasets only become more valuable, and the competitive moat is built not on the ratio of proprietary to public data, but on the enrichment, curation, and context layered on top of raw data, which competitors cannot easily replicate. That argument holds for ratings, indices, and insurance contributory data. It holds less convincingly for general technology advisory research.
Gartner has said its retention in its research businesses remains high, evidence that its research remains deeply embedded in customer decision-making. That is real. High retention buys time. But it is not the same as structural protection, and the market, correctly, is no longer treating them as synonymous.
The lesson from Gartner’s November 3, 2026 session is not that beating estimates means nothing. It is that the market is now doing what disciplined investors should have been doing all along: separating information businesses that own something irreplaceable from those that sell something AI increasingly gives away for free. That work is not finished, and the reassessment may not be either.