OpenAI’s Project Mercury Recruits 100+ Ex‑Bankers

OpenAI has quietly recruited more than 100 former investment bankers from firms including JPMorgan, Morgan Stanley and Goldman Sachs to train its large language models on transaction and financial modeling as part of a secret initiative codenamed Project Mercury. Contractors — including MBAs and ex-bankers — are reportedly paid about $150 per hour, undergo modeling tests and AI-chatbot screening, submit models weekly and receive early access to OpenAI tools while their work is used to teach the AI tasks such as Excel modeling, pitchbook preparation and slide formatting. The project’s stated goal is to automate repetitive, structured “grunt work” performed by junior analysts to free them for higher-value assignments. Economists and industry executives say the likely near-term effect will be a transformation of entry-level roles — shifting required skills and automating much routine analyst time — rather than immediate mass layoffs, though some banks have signaled slower hiring due to anticipated AI-driven productivity gains. Project Mercury highlights accelerating AI adoption across Wall Street and intensifies questions about labor impacts, oversight and the reliability of automated financial workflows.