AI Bubble vs Dot-Com Bubble: Why the Railroad Comparison Is Different
Railroad, internet and AI comparisons often collapse two questions into one: are investors paying too much for shares, and are businesses building too much productive capacity? Those problems can overlap, but they do not have to.
Two ways a boom can disappoint
A profitable company can trade at a price that requires implausibly strong future growth. Its stock can fall even if revenue keeps rising. Separately, an industry can invest in too much capacity, causing lower utilization and weaker pricing. That damages business earnings as well as valuation.
A July 2026 Federal Reserve research note examines technology-led overinvestment and the earlier investment cycle around the internet boom. It is research context, not a finding that every present AI project is excessive.
Use a two-axis test
Put business economics on one axis and valuation assumptions on the other. A company with durable cash generation but an aggressive multiple belongs in a different category from one with weak project returns at a modest market valuation.
Consider a hypothetical company whose earnings rise from $2 to $3 per share. If the market's multiple falls from 60 to 25, its implied share price falls from $120 to $75. Those teaching numbers describe multiple compression, not a forecast for an AI company.
Why infrastructure adds another layer
Once construction or equipment spending has occurred, management may continue operating because revenue covers incremental costs, even if it never earns an adequate return on the original investment. Cheap services can benefit customers while shareholders absorb the disappointing return.
That is why “people are using it” is not a complete rebuttal to the bubble argument. But “the stock fell” is not proof that the underlying technology failed either.
Ask a more precise question
Before labeling an AI company another dot-com, specify the proposed mechanism: weak demand, poor pricing, excessive leverage or an expensive entry valuation. Each requires different evidence.
Use historical comparisons to improve a financial model, not to replace one with a familiar story.
More in this analysis series
Start with AI stocks versus railroad mania. Continue with AI Bubble Risk: Debt, Refinancing and the Railroad Bond Lesson.
Analysis published October 1, 2026. Historical comparisons and hypothetical examples are not forecasts. Educational information, not investment advice.