31-year-old ex-Morgan Stanley analyst earns $12k a day doing latest fashionable job

Former Morgan Stanley analyst Dave Wang has pivoted from M&A to crypto and now AI, reportedly earning $25k a day with a company that trains bankers and hedge fund staff to use large language models. The piece highlights growing demand for AI education in financial services, including a Jane Street posting for a $300k machine learning educator. Overall it is a career-and-trend article with limited direct market impact.

Analysis

The deeper signal here is not just that AI is monetizing banker-skills; it is that the labor curve inside finance is getting steeper, faster. Firms are starting to pay up for people who can translate frontier models into repeatable workflows, which compresses the economic value of junior “glue” functions while raising the premium on staff who can audit outputs, embed guardrails, and sell transformation internally. That is a second-order positive for the large platforms and consultancies that own enterprise workflow distribution, not for the banks whose internal margin structure is most exposed to headcount leverage.

For the covered banks, the near-term P&L impact is modest, but the strategic impact is asymmetric. MS, GS, and DB all face rising pressure to automate analyst-heavy work faster than their compensation structures can adjust, which is incrementally bullish for efficiency ratios over 12-24 months, but only if management actually cuts roles rather than simply redeploys them into more expensive control functions. JPM’s negative read is more governance-oriented: the article reinforces that the largest, most process-heavy bank will likely be forced to spend more on AI controls and model-risk oversight before it fully harvests productivity gains, delaying operating leverage.

The contrarian view is that AI education is a highly visible, low-moat layer of the stack. As soon as banks codify prompt libraries and internal playbooks, the standalone value of “teaching bankers AI” should decay quickly; the durable profit pool migrates to enterprise software, data infrastructure, and model hosting. So the current enthusiasm may overestimate the longevity of this service model while underestimating the long-run winner: vendors that sit underneath bank workflows and become embedded defaults.

Catalyst-wise, the next 3-6 months matter more than the next 3 years for the pure training businesses, because procurement cycles and internal policy rollouts will determine whether this becomes a repeatable budget line or a one-off novelty spend. A sharper risk is reputational: if a bank’s employees rely on brittle model outputs and a trading, advisory, or compliance error follows, budgets can swing from experimentation to restriction overnight. That makes the tradeable question less about AI demand broadly and more about which institutions can convert AI into measurable headcount savings without creating control failures.

Key Decisions for Investors

• Long MS/GS vs short JPM on a 3-6 month horizon: the former are more likely to monetize AI through advisory, trading, and market-facing productivity gains, while JPM should carry a larger compliance/control drag. Use a modest spread size; risk is that JPM announces a credible enterprise-wide AI roll-out that narrows the gap.
• Avoid chasing standalone AI-training/education vendors at this stage; if exposed, use any post-announcement strength to trim. The moat is weak and procurement can internalize the capability within 1-2 budget cycles.
• Long enterprise workflow beneficiaries with durable distribution, not prompt-layer winners: prefer MSFT and NOW on 6-12 month pullbacks if bank AI adoption accelerates. The risk/reward is superior because banks will standardize on embedded tools rather than bespoke trainers.
• For DB, treat AI as a marginal positive but not a catalyst for rerating. If the stock rallies on cost-cutting/automation headlines, sell into strength; the time to close the efficiency gap is measured in years, and execution risk remains high.
• Options expression: buy 6-12 month call spreads on selected AI infrastructure names rather than banks, funded by writing upside in weaker commercial-bank proxies. The thesis is that monetization accrues to the toolchain, while bank beta is capped by governance and labor-friction constraints.