According to Financial Times News, OpenAI has hired more than 100 former bankers from firms including Goldman Sachs and JPMorgan, paying them $150 per hour to train artificial intelligence systems that can perform the financial modeling work traditionally done by junior bankers. The author, a former global head of equity capital markets at Bank of America, notes the irony that these bankers now earn more teaching AI replacements than they did in their actual banking roles, where a junior banker making $200,000 annually works roughly 80-hour weeks for about $50 per hour. This development threatens to upend investment banking’s traditional apprenticeship model, where technical proficiency and endurance have become prized traits over the past two decades. The transition suggests a potential return to valuing judgment, credibility, and storytelling abilities over pure technical skills.
Table of Contents
The Unraveling Apprenticeship Model
What makes this development particularly disruptive is how it attacks the fundamental economics of investment banking’s talent pipeline. For decades, banks have operated on a pyramid structure where junior analysts endure brutal hours performing repetitive modeling work while senior bankers leverage this analysis for client relationships. The implicit bargain was clear: suffer through 2-3 years of Excel hell, and you’d emerge with valuable skills and exit opportunities to private equity or hedge funds. Now, AI threatens to collapse this entire structure by automating the very work that served as both training ground and filtering mechanism.
The $150/Hour Training Paradox
The compensation differential reveals a fascinating market inefficiency. OpenAI paying $150/hour versus the $50/hour effective rate for junior bankers suggests either massive productivity gains from AI or a temporary market dislocation. More likely, it reflects the premium for domain expertise in training foundational models. These former bankers aren’t just teaching Excel formulas—they’re imparting the nuanced judgment about which assumptions matter in different scenarios, what clients actually care about, and how to structure analyses that drive decisions. This represents a fundamental shift from business school education focused on technical mastery to AI training focused on decision patterns.
Broader Financial Services Implications
The ripple effects extend far beyond investment banking. Private equity firms, hedge funds, and corporate development teams all rely on banking-trained analysts who’ve been through this crucible. If the training pipeline constricts, these organizations will need to develop their own training methods or increasingly rely on AI systems themselves. We’re already seeing early signs of this with quantitative hedge funds developing their own AI systems, but the advisory side has been slower to adapt. The real disruption may come from new entrants who build leaner, AI-first advisory firms without the legacy cost structure of traditional banks.
The Human Advantage Redefined
As technical skills become automated, the human skills that become most valuable are precisely those that have been hardest to teach: client intuition, negotiation tactics, reading subtle cues in meetings, and building trust relationships. The most successful future bankers will likely be those who can effectively manage AI outputs while focusing on the human elements of deal-making. This represents a return to banking’s roots, where relationships and judgment mattered more than spreadsheet prowess. However, it also creates a new challenge: how to identify and develop these skills in junior talent when the traditional proving grounds are disappearing.
Implementation Challenges Ahead
The transition won’t be seamless. Financial modeling AI faces significant hurdles around regulatory compliance, audit trails, and explainability. Banking regulators will demand transparency about how AI systems reach their conclusions, particularly for material transactions. There’s also the challenge of edge cases—unusual deal structures or market conditions where historical training data may be insufficient. Early adoption will likely focus on standardized analyses with human oversight, gradually expanding as confidence in the systems grows. The hybrid “managing analyst” role mentioned in the article represents a practical intermediate step, but the long-term trajectory points toward significant workforce transformation.
Timeline and Adoption Curve
Based on current AI capabilities and banking industry conservatism, I expect meaningful automation of junior banking tasks within 2-3 years, with widespread adoption taking 5-7 years. The first movers will likely be bulge bracket banks with dedicated AI budgets, followed by mid-market firms, and finally boutique advisors. The most immediate impact will be on pitchbook creation and basic valuation work, with more complex modeling taking longer to automate reliably. What’s certain is that the banking talent market of 2030 will look fundamentally different from today’s, with fewer entry-level modeling roles and greater emphasis on AI management and client relationship skills from day one.
