According to Fortune, Goldman Sachs CEO David Solomon delivered cautious optimism about AI’s economic impact during a Thursday conversation with Carlyle Group co-founder David Rubenstein at the Economic Club of Washington, D.C. Solomon revealed that six or seven large companies plan to spend $350 billion combined this year on AI infrastructure, which he identified as a key growth driver for the U.S. economy. While acknowledging the “enormous” opportunity in AI, Solomon emphasized that “there will be winners and losers” and cautioned that much of the capital being deployed won’t produce adequate returns. Drawing historical parallels, he referenced Alan Greenspan’s 1996 “irrational exuberance” warning when the Nasdaq was near 1,300, noting it later rose above 5,000 before significant adjustments occurred. This perspective comes as Goldman Sachs reported stronger-than-expected third-quarter earnings driven by investment banking and trading revenue.
Table of Contents
The $350B Infrastructure Reality Check
Solomon’s specific figure of $350 billion in annual AI infrastructure spending reveals the staggering scale of current commitments, but this number deserves critical examination. This spending primarily represents capital expenditure on data centers, specialized chips, and cloud infrastructure rather than immediate productivity gains. The companies making these investments—likely including Microsoft, Google, Amazon, and other tech giants—are essentially betting that future AI services will generate sufficient returns to justify today’s massive outlays. What Solomon didn’t explicitly state is that this infrastructure buildout creates a “winner-take-most” dynamic where only the best-capitalized players can compete at scale, potentially limiting innovation from smaller entrants who can’t match these investment levels.
Inevitable Boom-Bust Cycles in Tech
Solomon’s reference to Greenspan’s 1996 warning and the Nasdaq’s subsequent trajectory reveals a sophisticated understanding of technology investment cycles that many current AI enthusiasts are ignoring. The pattern he describes—initial excitement, massive capital formation, overinvestment, followed by consolidation and correction—has repeated throughout modern technological history from railroads to dot-com to cloud computing. What makes the current AI cycle particularly risky is the concentration of investment among a handful of mega-cap companies, creating systemic risk if their bets don’t pay off as expected. The investment banking perspective Solomon brings is crucial here—he’s seen how capital flows can distort markets and create bubbles that eventually correct, often painfully for late entrants and marginal players.
The Coming Productivity Paradox
While Solomon expressed confidence in meaningful productivity gains from AI integration, history suggests this transition may be more complex and prolonged than optimists anticipate. The “productivity paradox” observed in previous technological revolutions—where massive investment initially fails to translate into measurable productivity improvements—could easily repeat with AI. Companies may struggle with implementation challenges, workforce retraining, and organizational resistance that delay realizing the promised benefits. Furthermore, the productivity gains that do materialize may be unevenly distributed, primarily benefiting companies with existing scale and technical expertise while leaving smaller businesses struggling to keep pace. This could exacerbate economic inequality and create new competitive dynamics that Solomon’s “winners and losers” framing only begins to capture.
Unmentioned Regulatory and Ethical Headwinds
Notably absent from Solomon’s remarks was any discussion of the regulatory and ethical challenges that could significantly impact AI’s development timeline and profitability. As artificial intelligence systems become more powerful and integrated into critical infrastructure, they will inevitably attract increased regulatory scrutiny around data privacy, algorithmic bias, and market concentration. The European Union’s AI Act and similar legislation developing in the U.S. could impose compliance costs and limitations that affect return on investment calculations. Additionally, public backlash against job displacement or controversial AI applications could slow adoption rates, creating another headwind that current investment models may not adequately factor in.
Wall Street’s Calculated Positioning
Solomon’s balanced perspective reflects Goldman Sachs‘ strategic positioning in the AI ecosystem—participating in the boom while maintaining enough skepticism to avoid catastrophic losses when the inevitable correction occurs. Major investment banks benefit from AI in multiple ways: through advisory fees for mergers and acquisitions, capital raising for AI companies, and proprietary trading in AI-related securities. This diversified exposure allows them to profit from the hype cycle while having protection against downturns. Solomon’s message serves both as realistic guidance for clients and careful risk management for his own institution—acknowledging the transformative potential while preparing stakeholders for the volatility that typically accompanies technological revolutions.
