Thomson Reuters delivered solid growth, but the market focused on the bill for artificial intelligence. In the third quarter of 2026, the group reported revenue of $1.833 billion, up 10%, and raised its full-year organic growth outlook to 8.5–9%. Yet its shares fell after the release: investors are weighing whether AI spending can turn into recurring revenue and durable margins.
Real growth, with several engines
Results released on November 3 show that expansion does not rely on a single technology bet. Recurring revenue, which accounted for 88% of quarterly sales, rose 9%, while transactional revenue grew 27%. The company’s three major customer segments — legal professionals, corporates, and tax and accounting — each recorded 10% organic growth.
Management also raised its expected 2026 organic revenue growth range, from about 8% to 8.5–9%. This outlook excludes the effects of foreign exchange, acquisitions and disposals. It is an encouraging signal about underlying activity, but not a guarantee: the company itself points to risks involving the economy, interest rates, geopolitical developments and the execution of its investments.
CoCounsel puts the AI promise to the test
In the legal segment, revenue grew 9% at constant currency, driven in part by Westlaw and CoCounsel Legal. The suite combines legal research with generative AI features for professionals. Reuters reports that the company presented enhanced versions of CoCounsel, designed to help prepare briefs and motions. The ambition is clear: to sell more than access to a database by embedding assistance into tools already used by law firms and their teams.
But revenue growth alone does not prove that the new model is already profitable. The legal segment’s adjusted EBITDA margin fell from 49% to 46.6%. Thomson Reuters attributed the decline to higher technology and other costs. Reuters also cited $22 million in severance costs and the acquisition of legal data analytics platform Noetica among the factors weighing on the margin.
The distinction matters for customers as well as investors. A legal assistant can attract attention in a product demonstration, but professional adoption depends on repeatable accuracy, dependable sourcing and fit with established review procedures. Firms also need to understand how confidential material is handled and which tasks still require professional judgment. Those operational tests take longer than a launch announcement. They will determine whether users accept premium pricing and whether early experimentation develops into renewals that support the infrastructure and research costs behind the tools.
AI monetization is not yet a straight line
The Reuters division grew 13% at constant currency, including 12% organic growth. The company’s release attributed part of the increase to generative AI content licensing, noting that this revenue was largely transactional. In other words, it may depend on specific agreements or usage and does not necessarily offer the same visibility as a recurring subscription.
This detail helps explain market caution. Editorial content can become a valuable resource for training or powering AI systems, but rights holders still have to determine how to price those uses, track reuse and turn agreements into predictable revenue. Reuters reports that the underlying contract value associated with generative AI reached 34% in the quarter, up from 32% in the second quarter and 24% a year earlier. This internal measure indicates growing adoption; by itself, it does not establish the products’ ultimate profitability.
Why the shares fell despite the numbers
Reuters reported a 7% share-price decline on the day of the release, even as quarterly revenue and adjusted earnings per share increased. The gap between operating performance and the market reaction partly reflects investor expectations: technology companies are now judged on AI sales and on the cost of developing and delivering them. Rising revenue can coexist with pressure on margins, development spending and cash flow.
The metrics also need to be distinguished. Adjusted earnings per share reached $0.81, compared with $0.71 a year earlier, a 14% increase. The group’s adjusted EBITDA margin, meanwhile, edged down to 35.5% from 35.8%. Diluted earnings per share under IFRS fell to $0.78, in part because operating profit was lower than in the comparable period. These indicators tell different stories and should be read separately.
The real test: durable revenue, not just a demonstration
Thomson Reuters now has to show that its AI tools justify their price, retain customers and improve productivity without eroding margins. The company says it has significant investment capacity and highlights its legal, tax and accounting products. Customers will judge value through concrete outcomes: faster research, answer quality, reliable citations, data protection and integration into everyday work.
The quarter therefore neither conclusively validates nor disproves the AI strategy. It shows a company whose recurring core continues to grow, while some newer AI-related revenue remains transactional and technology costs weigh on key margins. In professional software, the next contest will not simply be to launch an impressive assistant. It will be to prove, quarter after quarter, that the assistant becomes an indispensable and profitable service.
Sources
- Thomson Reuters, third-quarter 2026 results, November 3, 2026.
- Reuters, revenue growth and market reaction to investment costs, November 3, 2026.
