Money and business in the Middle East.

AI

Thomson Reuters CEO Bets on Content as Stock Trails S&P 500

Thomson Reuters is accelerating AI spending even as its shares lag the S&P 500 since August 2025, when advances in frontier models raised fears that software and information providers would be displaced.

· Originally published by ontime+ · Source: Reuters · Last verified: 17 Nov 2026

Key Points

  1. Thomson Reuters CEO Steve Hasker says proprietary content, not frontier models, decides whether AI output is reliable.
  2. The stock traded at $109.73 Monday, 22.6% below its 52-week high of $141.77.
  3. The company is adding AI spending while competitors fear information providers get displaced by general-purpose models.

The latest:

Thomson Reuters is accelerating AI spending even as its shares lag the S&P 500 since August 2025, when advances in frontier models raised fears that software and information providers would be displaced. Chief Executive Steve Hasker told The Wall Street Journal he sees the opposite, arguing authoritative content is the decisive ingredient in accurate model output. AI-enabled offerings now account for 34% of revenue, the company said.

Details:

  • The thesis: Hasker said proprietary, highly curated content is the single most important component in the AI ecosystem, ahead of chipsets and energy, because it is what makes output accurate and what he calls fiduciary-grade. Abundant content exists, he said, but differentiation comes from unique data sets others cannot access.
  • The market verdict: Shares of the Toronto-based company traded at $109.73 Monday afternoon on Nasdaq, 22.6% below a 52-week high of $141.77. The underperformance against the S&P 500 dates to August 2025, when apprehension about frontier models extended to Thomson Reuters as a supplier of news, data and services to professional markets.
  • The growth claim: Hasker maintains AI is delivering positive returns, and cited it as a major factor in lifting revenue growth across the legal, corporate, and tax and accounting units from 4% to 5% when he arrived to a projected 9.5% to 10% this fiscal year.
  • The spending plan: The company said in February it would expand profit margins by 100 basis points in 2027, then said on Nov. 3 it would reinvest part of that operating leverage into innovation and growth. The money goes to engineers, salespeople and forward-deployed engineers who implement AI inside client companies.
  • The products: Thomson Reuters released a next-generation version of its model-agnostic CoCounsel Legal this year and launched a fully agentic CoCounsel Tax, Audit & Accounting in October. Both sit under a strategy the company brands Fiduciary-Grade AI, covering research, analysis and document drafting for professionals it says cannot afford to be wrong.
  • The in-house model: One CoCounsel Legal feature runs on Thomson 1.0, the company’s proprietary large language model released in August. To complete it, Thomson Reuters acquired Cambridge, U.K.-based startup Safe Sign Technologies and invested roughly $40 million in training, talent and compute.
  • The training base: The systems were trained on proprietary data from Westlaw, Practical Law, Checkpoint and Reuters. Some 2,600 internal practice-area experts teach the agents to research, verify and reason, and vet their output, alongside human customer service staff clients can reach with questions.
  • The restructuring: Hasker, CEO since March 2020, scrapped the holding company structure for an operating company model so the chief technology officer and the heads of engineering and product could work across the enterprise. Breaking silos let the company move faster, he said; planning cycles shrank from three to five years to six to 12 months.
  • The Reuters role: Hasker said the Reuters unit supplies a verified base of global news and insight, fulfills a societal obligation and helps recruiting. Thomson Reuters was formed in 2008 when Thomson acquired Reuters Group, founded in London as a stock ticker agency in 1851.

Between the lines:

The gap between the 34% AI revenue share and a stock 22.6% below its high frames Hasker’s argument: investors are pricing frontier models as substitutes for curated legal and tax data, while he is spending $40 million-scale sums to prove they are inputs that need it. His own caveat limits the timeline — he said the hard change-management work has barely progressed, and without it companies will not report significant returns.

What’s next

Watch whether the 9.5% to 10% revenue growth target for the Big Three units lands this fiscal year, whether the 100-basis-point 2027 margin expansion survives the reinvestment, and how the AI revenue share moves from 34%.

Read on ontime+ ↗

Thomson Reuters CEO Bets on Content as Stock Trails S&P 500 · INXEN