Article Directory
The Quiet Plan to Disassemble Wall Street's Entry Level
The news landed with the kind of sterile precision you’d expect from a press release, but the source was a leak: Sam Altman’s OpenAI is coming for Wall Street’s grunt workers as AI continues to transform the entry level. The project (code-named "Mercury") involves a cohort of over 100 former analysts from the likes of Goldman Sachs and JPMorgan, meticulously teaching an AI the drudgery that has defined the first two years of a Wall Street career for decades.
The official narrative, spun by consulting-class economists, is one of serene transformation. Shawn DuBravac of the Avrio Institute suggests this isn't about replacement, but a shift in skills. He predicts that within a year, firms will automate 60% to 70% of the time analysts spend on "lower-level tasks" like cleaning spreadsheets, building financial models, and assembling pitch decks. The freed-up junior bankers, the story goes, will then ascend to more "sophisticated" work, tackling complex quantitative analysis usually reserved for more senior roles.
This is a clean, palatable story. It suggests a future where the grueling 80-hour workweeks—spent staring at the unforgiving grid of a Microsoft Excel window under the hum of fluorescent lights—are simply replaced by more interesting, more valuable labor. The junior analyst is not fired; they are elevated.
But I've looked at efficiency models for years, and this narrative has a significant data discrepancy. When you propose eliminating a majority of a role's core function, you aren't just changing the job description. You are fundamentally altering the economic equation of the role itself. What happens when a machine can do 70% of a $150,000-a-year employee's job? Do you really just invent 50 hours of new, "sophisticated" work for them each week, or do you recalibrate the number of employees you need?
The Flawed Analogy of the Spreadsheet
The most common analogy being deployed to soothe anxieties is the advent of the electronic spreadsheet. "Just like Excel spreadsheets did back in the day," DuBravac states, "[AI] will streamline some work." This comparison is, from my perspective, a critical misreading of the technological leap we're facing.
The spreadsheet was a tool that amplified the productivity of a single analyst. It allowed one person to do the work previously done by a roomful of bookkeepers and clerks with ledgers and calculators. It didn't eliminate the analyst; it eliminated the analyst's support staff. The spreadsheet made the analyst exponentially more powerful. This new wave of AI isn't a tool in the same way. It isn't a better calculator. It is a process replacement. It aims to perform the entire workflow of building a discounted cash flow model or a leveraged buyout analysis from start to finish.
This is the part of the public analysis that I find genuinely puzzling. The comparison to Excel conveniently ignores the scale and scope of the disruption. Excel was about calculation; generative AI is about generation. It’s the difference between giving a carpenter a nail gun and giving them a factory that pre-fabricates the entire house. You still need the carpenter to assemble the final product, but you certainly don’t need as many of them on the framing crew.
And this brings us to the numbers that get buried at the end of the reports. A McKinsey study from March revealed that only 38% of organizations using AI expect it to have "little effect" on their workforce size in the next three years. Let's reframe that for clarity. A striking 62%—a clear majority—of organizations implementing this technology do anticipate an impact on their headcount. They are not planning for a one-to-one replacement of old tasks with new ones. They are planning for a structural change.
The question isn't whether junior bankers will have different jobs. The real question is, how many of them will be needed at all when a single senior banker, armed with a sophisticated AI, can generate and check the output of what used to be a three-person analyst team?
The Headcount Equation
Let’s be precise. The narrative of "transformation" is a public relations strategy designed to soften the landing of a brutal economic reality. The objective of automating 70% of a job function is not to enrich the employee's work life; it is to reduce the cost of labor. Wall Street firms are not sentimental. Their largest and most unwieldy expense is compensation. Project Mercury is not a philanthropic endeavor to save analysts from burnout. It is a direct investment in labor arbitrage.
The logic is inescapable. If one senior banker can supervise an AI that produces the work of five junior analysts, the firm doesn't hire four new analysts and assign them "more complex" work that doesn't yet exist. It hires one analyst and reaps the efficiency gain. The value proposition for the banks isn't employee satisfaction. It's shareholder value, driven by a leaner, more productive, and drastically smaller junior workforce. The grunt work is being automated away, and it was the grunt work that justified the army of grunts.
