Goldman Sachs Just Replaced 200 Analysts With One AI System, The Results Are Complicated

These days, Goldman Sachs’ analyst floors at 200 West Street have a certain stillness that wasn’t present two years ago. The desks remain in place. The terminals at Bloomberg are still glowing. However, anyone who walked those floors in 2022 would immediately notice that the bodies, the rows of twenty-three-year-olds eating depressing desk salads at 11 p.m. while constructing DCF models, have thinned out. According to people familiar with the deployment, the bank has discreetly implemented an internal AI system that has taken on the responsibilities of about 200 junior analyst-equivalent positions. Depending on who you ask, the results are either spectacular or quietly concerning.

Part of the point is that the AI’s work is not glamorous. accounting for trade. onboarding of clients. The first eighteen months of an analyst’s career used to be devoted to the never-ending reconciliations, pitchbook formatting, and footnote checking. It didn’t require genius. Someone who was prepared to stay until two in the morning was needed for all of this. The new system doesn’t cost six figures plus bonus, manages it across multiple time zones, and doesn’t miss weekend deadlines. From the standpoint of pure margin, this type of math is what causes CFOs to recline in their seats.

However, the image isn’t as clear-cut as the headlines imply. Only roughly one in five companies that have replaced human roles with AI report that the transition went smoothly, according to industry surveys recently cited by consulting analysts and Medium contributors. The others are coping with more subdued issues. code produced by AI that has minor mistakes. Financial results that appear correct but are based on assumptions that no one at the company was able to fully trace. Work that was previously triple-checked by a human chain is now being double-checked by senior bankers. The fact that Goldman hasn’t revealed which side of that ratio it is on is instructive.

All of this has a generational component that isn’t given enough attention. For many years, the investment banking analyst program served as the harsh training ground for the next generation of MDs. You endured two years of tedious labor, acquired the muscle memory of finance, and inadvertently picked up the unspoken guidelines for how transactions are actually completed. What happens if the apprenticeship is eliminated? It’s still unclear if senior bankers in the future will have acquired the same skills as those in the present or if they will be completely different, managing AI systems that they don’t fully comprehend.

The pattern is not limited to Goldman. While its stock was rising, Block laid off about 40% of its employees in February 2026. In order to integrate engineers directly into its lending and onboarding processes, Customers Bank recently announced a collaboration with OpenAI. AI-driven automation is expected to generate £100 million in value for Lloyds this year. According to a Goldman research report, AI is expected to displace 300 million jobs worldwide. CEO David Solomon has openly assured reporters that people will adapt and that the economy is flexible. Investors appear to have faith in him. Less so for junior bankers.

The irony in all of this is difficult to ignore. Goldman, the company that used to define what it meant to work in finance and where rejection letters were gathered and framed, is now developing a system that subtly eliminates the lowest level of its own hierarchy. The decision, according to the bank’s leadership, frees up workers for higher-value tasks. You get a different feeling as you watch this happen. It’s more akin to a gradual rewriting of what finance is and who gets to do it than a catastrophe. It will likely take another ten years to determine whether that rewrite is beneficial for the industry as a whole or just for one quarter’s earnings.