OpenAI's ChatGPT for Wall Street Targets Junior Bankers' Hours

OpenAI launched ChatGPT for Financial Services on Sept. 10, a version of ChatGPT Work built around licensed financial data and GPT-6 Astra's reasoning, aimed at the research and modeling work junior investment bankers do.

OpenAI said in its announcement that the product was shaped by a design partnership with Morgan Stanley and Evercore, which steered its starting point toward investment banking and equity research. CNBC reported the same day that OpenAI demoed the tool analyzing acquisition targets, extracting financial figures and generating PowerPoint decks matching a bank's own style guide.

What Morgan Stanley and Evercore shaped

The product ships with premium datasets from Daloopa, PitchBook, LSEG News and Crunchbase available immediately, OpenAI said, with no separate contracts to negotiate. Subscription integrations cover S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva and Moody's, and more than 50 additional connectors include Datasite, Box, Preqin, FactSet and Intapp. Teams can build financial models, run LBO analysis, banking shorthand for calculating the returns on an acquisition financed mostly with debt, and produce client materials using a firm's own templates.

Category	Partners named by OpenAI
Premium datasets	Daloopa, PitchBook, LSEG News, Crunchbase
Subscription integrations	S&P Capital IQ, LSEG, MSCI, Dow Jones Factiva, Moody's
Additional connectors (50+)	Datasite, Box, Preqin, FactSet, Intapp

The hours it's aimed at

CNBC reported that junior bankers routinely work 100-hour weeks on exactly the tasks the product automates: company research, data analysis, and pitchbook creation. Nick Turley, OpenAI's vice president of product, framed the launch to CNBC as an efficiency gain rather than a reduction in headcount. CNBC also reported that OpenAI's enterprise business generated more revenue than its consumer business as of August, without giving a specific figure for either.

Goldman's 'cognitive atrophy' warning

Chris Churchman, a Goldman Sachs partner overseeing an AI project at the bank, told CNBC that automating the research and modeling tasks junior bankers use to learn their trade risks "cognitive atrophy," his term for skills that stop developing because the entry-level work that used to build them has been handed to software. CNBC framed the product as raising questions for an industry built on an apprenticeship model, where junior staff learn by doing the grinding research work that ChatGPT for Financial Services now automates.

What's locked down

OpenAI said the product supports SAML single sign-on, SCIM provisioning and role-based access controls, and that business data is not used to train its models by default. It listed encryption at rest and in transit, configurable data retention, log export through a Compliance Platform for audit workflows, and the ability to set up separate workspaces as information barriers between teams that are not supposed to share data.

OpenAI did not say how much the product costs or whether banks beyond Morgan Stanley and Evercore have access yet. The company's other enterprise pushes this month include a public beta of its Agents API and the Preparedness Framework rollout behind GPT-6 Astra, the same model now doing the financial reasoning.