AI Tokenomics Come For Wall Street
Financial data companies such as FactSet and Moody’s appear to have made progress reassuring investors that they won’t be replaced by finance-focused AI tools. But investors may have new questions these days around how the uncertainty of AI costs will affect usage of these businesses’ products.
Investors were spooked when Anthropic and others touted AI products for finance and other professions earlier this year. Companies like FactSet, Moody’s, S&P Global and LSEG, which all make money in part by selling information like market data, company financials and risk analytics, were hit by a market selloff.
Stocks have recovered at least partly since then, and the firms made a pretty compelling argument that rising usage of AI tools doesn’t hurt the value of their data. Banks, asset managers and other regulated customers still need licensed, current information that they can easily verify and audit.
These firms have also added AI features to their products and made data available in formats customers can access through their own AI tools. Overall, they say AI adoption is positive for demand, because an AI agent can consume far more data than any human analyst.
But things aren’t proving quite that simple. In recent months, businesses have become more focused on the rising cost of AI usage and many companies are adopting measures to keep AI spending under control. That’s raising a fresh set of AI-related questions for data sellers.
A big question is who pays the cost of the tokens needed to run AI-powered data analysis. The answer depends on how the access to the data is structured, and whether a customer is using the data through their own AI tools or not. How much more customers are willing to pay for AI to run data analysis is another question. And like AI costs themselves, data consumption by AI tools can be unpredictable, because an agent might rerun queries or try lots of data sets. The overall picture might be unnerving for investors in a business where they’ve long prized relatively steady revenue streams.
On Wednesday, FactSet CFO Joshua Warren said AI tokens “were not a line item that we really thought about in 2025.” Those costs are showing up in FactSet’s own spending as it uses AI internally and builds AI-powered products for customers. FactSet says it has internal controls like budgeting and model routing, and a recently expanded Google partnership gave it preferential token pricing.
When it comes to customer demand, FactSet executives noted that AI products were proving to be a source of growth and commanding richer contracts, but they also said customers are still not clear on what their own consumption will be, so want flexibility in pricing.
“In an AI world, right now, there is a lot of value that clients are placing on that,” FactSet CEO Sanoke Viswanathan said, and the company is adding provisions for “new consumption patterns” as customers experiment.
Moody’s, for its part, appeared to be getting ahead of both sides of the cost question. In a June investor Q&A with a JPMorgan research analyst, a Moody’s executive noted that customers bear the AI token costs when they use its data through tools like Claude Enterprise, ChatGPT or Microsoft Copilot. They added that Moody’s can help customers keep a lid on those costs, including by packaging its data with instructions that help AI agents use it quickly and efficiently. Moody’s also builds token costs into contracts with customers that are using its own AI tools, but it says it closely monitors those costs through routing and other measures.
To be sure, there isn’t much evidence yet that AI has moved the needle either way. Financial data businesses aren’t exactly blockbuster growth stories, but their recent growth rates still look broadly similar to before the AI scare.
If anything, the most noticeable impact from the AI boom might be outside these companies’ data subscriptions. LSEG, which also owns the London Stock Exchange and other markets businesses, has suggested that AI is accelerating decision making and activity in general, which translates into more trading activity and risk management needs. Markets revenue jumped in the first quarter, helped by strong trading volumes.
Meanwhile Moody’s and S&P Global have both said that a surge in debt issuance tied to AI infrastructure spending has boosted their credit ratings businesses, which make money in part when companies issue new debt and pay to have it rated.
The next question is how far into finance these companies can reach with their AI businesses. FactSet executives said on Wednesday that the wealth management industry is still in the early stages of AI adoption, but financial advisers present a big potential customer pool. That could also turn into client-facing tools for people using wealth and consumer financial services.
New types of users could well have a very different take on AI costs and payoffs than the big existing financial customers, which could introduce further complications around fees and pricing. Meanwhile, the rest of the financial data crowd are due to report results in the coming weeks. We’ll see how the narrative continues to change.
New From Our Reporters
