LSEG was supposed to be one of the financial data groups most exposed to AI. Its first-quarter numbers tell you a less convenient story.
London Stock Exchange Group has spent the past year being treated as if AI were coming for its best business. Cheap models would answer market questions. Analysts would need fewer terminals. A data incumbent built around Refinitiv, FTSE Russell and Reuters would suddenly look old. That was the fear. The latest figures make it harder to keep saying it with a straight face.
Financial News reported that LSEG posted £2.4 billion of first-quarter 2026 revenue excluding recoveries, up 9.8% at constant currency and ahead of the 8% expected in a company-compiled analyst consensus. The company also lifted its 2026 guidance to the upper half of its 6.5% to 7.5% organic constant-currency growth range. For a business being talked about as a possible AI casualty, that is a stubborn set of numbers.
The growth was not only an AI story. LSEG's markets division grew 16% to £987 million, helped by record volumes at Tradeweb, the electronic trading platform in which LSEG owns 51%. Its fixed income, derivatives and other unit rose 18.4% to £452 million, while equities, including the London Stock Exchange and Turquoise, grew 11%. The subscription businesses investors worry about most, data and analytics, FTSE Russell and risk intelligence, still grew 6.3%.
But AI is where the argument has moved. David Schwimmer, LSEG's chief executive, said on April 23 that more than 150 customers had connected or were onboarding to the company's Model Context Protocol server, which it set up last year to link LSEG products to external AI systems. You don't need to pretend that 150 customers settles the matter. It doesn't. It does tell you that big financial institutions are not simply abandoning licensed data because a chatbot can produce a confident paragraph.
MCP matters because it gives AI systems a way to pull from approved outside data sources. Anthropic introduced the open standard in 2024, and LSEG is using it to make its data available inside client workflows without turning every answer into an untraceable scrape. That is the whole point. A hedge fund, bank or compliance team can't build production work on a number whose origin it can't explain to a regulator, a client or its own risk desk.
Financial News also reported in December 2025 that LSEG partnered with OpenAI to make ChatGPT Enterprise available to 4,000 staff and selected clients, with users who have LSEG credentials able to access licensed market data and news through ChatGPT. LSEG had already tied itself to Microsoft through a 10-year strategic partnership in 2022, under which Microsoft took a 4% stake. Anthropic, Snowflake and Databricks sit in the same orbit. LSEG is not trying to beat the model companies at building models. It wants to be the data layer they cannot do without.
Frankly, that is a better position than the panic implied. AI can summarize a company filing, draft a market note or help an analyst query a dataset more quickly. It cannot magically create the rights, provenance and audit trail attached to licensed exchange data, Reuters news, Lipper fund data or FTSE Russell indices. If you work in a regulated market, the answer is only useful if you can trust what sits underneath it.
That does not make LSEG untouchable. Bloomberg, FactSet and other data providers can build their own connectors, and MCP is an open standard rather than a private moat. A well-funded startup can also move quickly around a narrow workflow. LSEG's advantage is that its content is already inside the institutions that matter, and switching away from trusted data plumbing is rarely as easy as a pitch deck makes it sound.
The market pressure is still real. The Guardian reported in February that Elliott Management had built a significant stake in LSEG after concern over AI disruption and weak listings weighed on the shares. The same report noted that LSEG now gets almost half its revenue from data and analytics after the 2021 Refinitiv takeover, while the stock exchange itself is a much smaller piece of the group. That is why this debate matters. Investors are not arguing about a side project. They are arguing about the center of the company.
The cleaner reading is this: LSEG is not a finished AI winner, but it is no longer an easy AI victim. The business still has to prove that MCP connections and Workspace AI tools become durable revenue, not just useful demos and client meetings. For now, the Q1 numbers and the customer uptake give Schwimmer something better than a slogan. They give him facts.
Also read: Congress is circling the health data business model that AI startups built their futures on, Taiwan Raided Super Micro's Offices Today as the Nvidia Chip Smuggling Case Reaches a New Front, Arena hit $100M in annualized revenue by letting AI companies pay to be evaluated, and that is exactly the problem
