Goldman Sachs CEO David Solomon on Running a Bank in the Age of AI
Odd LotsAI Summary
→ WHAT IT COVERS Goldman Sachs CEO David Solomon discusses AI's impact on banking jobs, Goldman's hiring trajectory, the Alphabet $90B equity offering Goldman led, the SpaceX IPO mandate won over 20 years of relationship-building, current market valuations versus the 1990s dot-com era, and why human relationship skills become more valuable as AI commodifies knowledge work. → KEY INSIGHTS - **AI Job Displacement Reality:** Solomon argues the widely cited 16% decline in entry-level white-collar hiring conflates a narrow set of industries with the broader labor market. Goldman currently onboards approximately 2,500 interns and 2,400 permanent hires annually — similar to pre-COVID levels, down from 3,000-plus in 2021 — with headcount expected to contract modestly over three years, not dramatically. - **Measuring AI Productivity at Scale:** Goldman's most quantifiable AI gains come from reengineering operational processes, not incremental analyst efficiency. The client onboarding, anti-money-laundering, and KYC workflow previously involved 3,800 people in partial-time roles; the rebuilt process will require a few hundred. Revenue-per-banker and earnings-per-banker metrics serve as the proxy for less measurable productivity gains in client-facing divisions. - **Clean Data Sets Determine AI Value:** AI models perform extraordinarily well against clean, proprietary datasets and produce unreliable outputs when scraping unstructured internet data. Goldman's 40-year trading dataset from its SecDB system gives the firm a structural AI advantage. Solomon illustrates the risk with a real example: a leading AI model incorrectly omitted Tiger Woods from a list of back-to-back Masters winners until corrected by human judgment. - **Relationship Capital Compounds Over Decades:** Goldman secured the SpaceX IPO mandate through 20 years of continuous engagement, not a six-month pitch. The firm's first contact with Elon Musk came through SolarCity financing, followed by the Tesla IPO. Dan Dees built the ongoing relationship over roughly 12 years. The actionable principle: consistent, non-transactional client coverage across leadership transitions outperforms any single competitive pitch process. - **Equity Issuance Replacing Debt for AI CapEx:** The Alphabet $85-90B secondary offering — the largest follow-on equity raise in history, led exclusively by Goldman for five months — signals a structural shift. Solomon expects multiple large tech companies to follow, issuing equity alongside debt to fund multi-year AI infrastructure capital plans. Companies with high multiples and voracious capital needs face leverage risk if they rely solely on debt financing. - **Market Valuations: Elevated But Not 1999:** The top 10 S&P 500 companies trade at roughly 30-33x forward earnings versus 45-50x during the late-1990s internet boom. The remaining 490 companies trade at 17-20x forward earnings. Solomon's framework: the 17x cohort looks attractive if AI-driven efficiency improvements accelerate earnings growth across the broader index over five years, though he acknowledges crowding into a narrow group of stocks reflects fear-of-missing-out behavior. → NOTABLE MOMENT Solomon recounts asking a top AI model how many golfers had won back-to-back Masters titles. The model returned Jack Nicklaus and Nick Faldo but omitted Tiger Woods entirely — only correcting itself after Solomon pushed back. The model then admitted it could not learn from the error for future queries, underscoring the current ceiling of AI reliability without human oversight. 💼 SPONSORS [{"name": "VanEck (RAAX ETF)", "url": "https://vaneck.com/raaxpod"}, {"name": "Public", "url": "https://public.com/market"}, {"name": "Chase for Business", "url": "https://chase.com/podcastbizoffer"}, {"name": "Venture Global", "url": "https://ventureglobal.com"}] 🏷️ AI Labor Markets, Investment Banking, Equity Capital Markets, Market Valuations, Data Infrastructure, Relationship-Driven Sales
