OpenAI's new project could make junior bankers redundant

OpenAI has quietly hired over 100 former bankers and consultants from Morgan Stanley, JPMorgan Chase, and other top firms as part of a new initiative to train its AI for advanced financial modelling and consulting automation, WSJ reported on Tuesday.

The project, codenamed "Mercury," is designed to replicate and ultimately automate much of the grunt work typically done by junior investment banking analysts, including building IPO models, restructuring analyses, and leveraged buyout projections.

According to multiple reports, OpenAI is paying these ex-bankers around $150 an hour to help refine its models. The consultants contribute by writing prompts, testing model accuracy, and providing expert feedback on AI-generated financial outputs, helping the system learn to perform with the precision expected on Wall Street.

Also Read: 'Expect some bad stuff to happen’: OpenAI CEO Sam Altman's comment sparks concerns over AI risks

The hiring effort is part of a broader enterprise strategy that places OpenAI in direct competition with consulting and advisory giants. The company has been expanding its bespoke AI consulting arm, embedding engineers and domain experts directly into client operations across finance, government, and large enterprises.

Through the Mercury project, OpenAI is teaching its AI to build and revise complex Excel-based transaction models, analyse financial statements, and generate detailed pitch decks, tasks that often consume hundreds of analyst hours. These tools are already being tested in partnership with financial institutions, allowing clients to generate and validate deal analyses, statements, and presentations with minimal human input.

Early access programs with clients like Morgan Stanley, which already uses OpenAI’s technology within its Wealth Management division, are showcasing how AI can be integrated into banking workflows. Advisors can now draw on AI-powered insights for client portfolios, investment ideas, and market summaries, combining the firm’s proprietary data with OpenAI’s generative models, per reports.

Also Read: OpenAI's first-half revenue rises 16% to about $4.3 billion: Report

The move deepens OpenAI’s ties with major banks. The company recently secured a $4 billion credit facility from JPMorgan Chase and others, underscoring its growing financial partnerships. Meanwhile, JPMorgan itself is advancing its own AI agenda, aiming to become the world’s first "fully AI-powered megabank."

Experts believe OpenAI’s banker-led AI training could dramatically boost productivity in investment banking, enabling analysts to offload repetitive modelling tasks and focus on client strategy and execution. The project also signals a shift in how financial services view AI from an experimental add-on to a core productivity tool.

By embedding industry veterans into the training process, OpenAI’s tools are being shaped around the best practices and compliance standards of top-tier banks. The resulting products are expected to set a new benchmark for accuracy and reliability in financial automation, an area where generic AI tools have historically fallen short.

Also Read: OpenAI looks to build data centre in India: Report

As OpenAI expands its consulting and enterprise offerings, Mercury could become a blueprint for how AI reshapes white-collar work, not just in banking but across professional services where precision and process discipline are key.

With ex-Morgan Stanley and JPMorgan talent now training OpenAI’s systems, Wall Street’s future might soon run on prompts instead of spreadsheets.Read about our editorial guidelines and ethics policy