AI Boom Echoes 1840s Railway Mania, Warns Goldman Strategist
TREE NEWS reports: Goldman Sachs strategist Bobby Molavi has drawn a stark parallel between the current AI-driven market frenzy and the 19th-century British railway mania, warning that the builders of transformative technology often fail to capture its long-term value. In his latest “Ruminations” note, Molavi questions who will ultimately reap the rewards of the AI boom, suggesting it may not be the companies currently spending billions on infrastructure.
Nvidia’s market capitalization surged from $1 trillion to a peak of $5.8 trillion before falling back to an 18-month low, a trajectory that reminded Molavi of the railway bubble that gripped Britain in the 1840s. During that episode, 500 new railway companies raced to build 8,000 miles of track—20 times the length of England—spending more than the entire national income. Today, US data center capital expenditure runs into hundreds of billions annually, and cloud giants have already exceeded their full-year debt issuance forecasts with seven weeks left in 2026.
Bubbles and Utility Are Different Things
Molavi emphasizes that technological innovation can be epoch-defining without immediately creating value for its early backers. The internet, railways, and even tulips survived their respective bubbles, but the value ultimately accrued to different parties: the tulip industry as a whole, the application layer of the internet, and the broader economy that leveraged the infrastructure.
“Early capital that chased these bubbles got hurt… especially when speculation and hope translated into ‘valuations’ decoupled from fundamental revenue and cash flow,” Molavi writes. In other words, you can be completely right about the technology and still lose money on the stocks.
The internet bubble followed the same script: Cisco and Qualcomm peaked and fell, while 360networks, AltaVista, and Webvan flamed out entirely. The fiber was laid, the network was built, and a decade later, a different set of companies profited from that infrastructure.
This Time Is Different—and Riskier
Molavi argues that the current cycle carries greater risk because technology is now embedded in every sector. “Every industry is tech, and tech is every industry,” he notes, pointing to a rare degree of circular dependency: Nvidia’s deals with OpenAI, Nvidia’s stake in Intel, Oracle purchasing Nvidia GPUs, and hyperscalers holding stakes in AI startups.
“We’ve been in a massive FOMO trade for much of this year—a gold rush in the form of AI capex and technological supremacy. The fear of ‘not moving’ has far exceeded the risk of ‘moving too early or too much.'”
Market structure amplifies these dynamics. In a market increasingly driven by passive funds, ETFs, and systematic strategies, price momentum itself becomes a fundamental input. With the S&P 500 dominated by just a handful of stocks—over 60% of this year’s gains came from six stocks, and 35% from just one—rising prices force more money into these names, creating a feedback loop.
Valuations: Cheaper, But Not Cheap
Molavi compares current forward P/E ratios to past market bottoms: 10.4x in March 2009, 13.3x in March 2020, 15.7x in October 2022, and roughly 18x now (if last week marked a low). His verdict: “Cheaper, but not truly cheap.” He acknowledges that US corporate earnings are still growing at over 30%, but warns that if the AI capex cycle turns, much of that growth could evaporate.
Key Takeaways for Investors
- Technology ≠ Returns: Being right about AI’s transformative potential does not guarantee stock market profits. Timing and valuation matter.
- Circular Dependencies Amplify Risk: The interconnectedness of AI supply chains and investments means a downturn could have far-reaching effects.
- Momentum Cuts Both Ways: Passive flows and systematic strategies can inflate bubbles and accelerate declines.
- Debt-Fueled Buildout: Unlike the dot-com bubble, this AI infrastructure boom relies increasingly on debt and vendor financing, potentially spreading losses beyond equity holders to bondholders.
As Molavi concludes with cautious agnosticism: the value of AI is not in doubt; what’s questionable is the price and timing of returns. Tulips survived, the internet survived, railways survived—and GPUs will survive too. The question is who holds the bill when the music stops, and whether this time it lands on the entire market’s balance sheet.
