OpenAI and Anthropic Want the Junior Banker’s 2 a.m. Shift

Wall Street’s next generation of AI companies is taking shape around three main types of businesses. Some own financial data. Others turn that data into research and analysis. A third group controls special information that investors use to make decisions.

The growing market includes established providers such as London Stock Exchange Group (LSEG), Bloomberg and PitchBook, as well as startups such as AlphaSense, Rogo, Hebbia, Daloopa, Model ML, Samaya AI and Farsight.

Each company is trying to win a share of the technology budgets at investment banks, companies that manage investments and other financial institutions. But they make money in different ways. This raises a key question: Which companies will ultimately earn the most from AI in financial services?

The competition is growing as AI systems move beyond answering questions. They are starting to build financial models, study deals and prepare investment research. Financial institutions increasingly need trustworthy data, software that can handle complex tasks and special information that general-purpose AI models cannot easily copy.

LSEG, Daloopa and PitchBook Supply Data to the AI Labs

Data is the foundation of this market. LSEG, Daloopa, PitchBook, S&P Capital IQ, FactSet and Moody’s all provide financial data for banking tools built by OpenAI and Anthropic, PYMNTS reported on Sept. 11.

In October 2025, LSEG worked with Microsoft to let banks build AI agents using its data. In December, ChatGPT added access to that data, PYMNTS reported.

Daloopa and PitchBook supply data to both AI labs, and both companies are expanding how widely their information is used. Daloopa raised $47 million in Series C funding in May. In a blog post in September announcing a partnership with Model ML, Daloopa said its data covers more than 6,000 companies.

PitchBook, which is owned by Morningstar, added its private-market data to ChatGPT in December, according to a press release dated Dec. 17. Before that, PitchBook had connected its data to Claude, Perplexity and Microsoft’s Copilot Studio. Rogo, Hebbia, Model ML and Farsight also connect to PitchBook. This means startups that compete with the AI labs also buy data from the same suppliers.

Payment companies are making a similar bet on information they own. Mastercard and Plaid are training foundation models using their own transaction data. This gives them an advantage that companies building general-purpose AI models do not have, PYMNTS reported on April 6.

Rogo and Hebbia Sell the Analyst’s Work

Rogo and Hebbia turn financial data into finished work, including financial models, memos and presentation decks. Both companies use AI models from Anthropic, OpenAI and Google, according to PYMNTS’s Sept. 11 report.

Rogo Co-Founder and CEO Gabe Stengel said in April that much of Wall Street’s financial analysis work is “done by a 21-year-old in tools from 40 years ago at 2 a.m.,” PYMNTS reported on April 29, citing Bloomberg. Rogo’s software is designed to take this work off the analyst’s shoulders.

Banks are paying for these tools. Rogo’s annual recurring revenue, or the yearly income it expects from ongoing subscriptions, passed $50 million. That was up from $15 million at the end of 2025, PYMNTS reported in its Sept. 11 article, citing The Information. According to the April 29 PYMNTS report, Rogo’s customers include Lazard, J.P. Morgan, Bank of America and Wells Fargo.

Hebbia is nearing $50 million in annual recurring revenue. It also has a contract covering about 5,000 investment banking employees at Morgan Stanley, The Information reported on Sept. 3.

The AI labs are now selling tools for the same kind of work. Anthropic introduced 10 AI agents for finance in May, PYMNTS reported on May 5.

In September, OpenAI followed with ChatGPT for Financial Services. Morgan Stanley and Evercore helped shape the product as design partners, according to PYMNTS’s Sept. 11 report. Startups argue that banks still need their services. Farsight CEO Samir Dutta, whose company also competes with Rogo and Hebbia, said most banks still do not have clearly defined use cases that would let them build effective tools themselves.

Bloomberg and AlphaSense Build Their Own AI Interfaces

Companies that own financial content, such as Bloomberg and AlphaSense, are taking a third approach: building their own AI tools for customers to use. In a press release on Oct. 15, Bloomberg announced that ASKB, its AI tool that users can interact with through conversation, was now generally available to financial professionals. ASKB works within the Bloomberg Terminal, a paid service that traders and analysts use to access market data, news and analysis.

ASKB partly uses AI models from Anthropic, Fortune reported in April in an article published on April 28.

AlphaSense follows a similar approach. It offers an AI-powered market intelligence platform built around company filings, transcripts of earnings calls and research from brokerage firms. In June, the company raised $350 million at a company valuation of $7.5 billion.

The boundaries between these three parts of the market are already becoming less clear. In July, LSEG connected its data to Model ML. In September, Model ML partnered with Daloopa to bring Daloopa’s data into its platform. In February, Hebbia partnered with Fitch Solutions to add Fitch’s credit research to its platform.

The biggest deal so far, however, has been in the data business. On Thursday, Oct. 22, LSEG agreed to buy credit data provider Octus for $3.5 billion, PYMNTS reported. LSEG said that 80% of its AI agreements include purchases of additional datasets.