Episode 390: The "AI Bubble" and Stock Market Concentration

Episode Summary

Executive Summary: The episode explains why today’s U.S. stock-market worries—high concentration in the Magnificent 7 and elevated valuations—should be interpreted through a long-term, evidence-based lens. Using historical examples from Nortel, the dot-com era, Canada, and Japan, the hosts argue that concentration is less predictive than valuations, and that diversification and discipline remain the best response.

Main Topics: Podcast identity and community framing (Priority: 3/5): The hosts introduce the show’s mission, audience, and format, emphasizing evidence-based investing, longitudinal discussion, and the value of their community-driven approach. Current U.S. market concentration and AI-bubble concerns (Priority: 5/5): The episode opens with client concerns about the U.S. market being dominated by a handful of stocks and whether AI-related enthusiasm has created a bubble. Valuations vs. concentration as predictors of future returns (Priority: 5/5): The hosts distinguish between market concentration and market valuations, arguing that high valuations have a much clearer relationship with lower future returns than concentration does. Historical bubble patterns and productive bubbles (Priority: 4/5): They review past technology-driven bubbles—canals, railroads, the internet—to show how speculative booms can fund useful infrastructure even while harming investors. Nortel as a Canadian cautionary tale (Priority: 5/5): Nortel’s rise and collapse is used to show how extreme single-stock concentration can hurt a market, while also illustrating how diversified and value-oriented investors fared better. U.S. lost decade and the importance of diversification (Priority: 5/5): The dot-com bust and subsequent weak U.S. returns are contrasted with better outcomes in U.S. value stocks, foreign markets, and globally diversified portfolios. Practical portfolio lessons and critique of active management claims (Priority: 4/5): The episode concludes that investors should stay diversified and disciplined, noting that active managers’ ability to exploit concentration is limited and often overstated.

Key Arguments: U.S. market concentration is historically extreme, but concentration alone has not shown a strong or reliable relationship with future returns. High valuations are a more meaningful warning signal than concentration, though even valuations cannot predict timing. The current AI spending boom resembles earlier technology revolutions that produced both investor losses and real-world infrastructure gains. Nortel demonstrates that a single stock can distort a national market and cause severe losses, yet diversified investors recovered much faster than concentrated ones. U.S. investors who ignored valuation and concentration concerns in 2021 would have missed substantial gains, showing why market timing is unreliable. Global diversification protects investors from being trapped in one market’s extended underperformance, as seen in Japan after 1989 and the U.S. after 2000. Value and small-cap value stocks often held up better during market drawdowns, but they are not a guaranteed fix and can also underperform for long periods. The best response to concentration and valuation risk is not prediction but disciplined adherence to a globally diversified plan.

Data Points: Magnificent 7 share of S&P 500: 36% - Current concentration level cited for the seven largest U.S. stocks within the S&P 500. Magnificent 7 share of U.S. total market: 32% - Share of the total U.S. market when using a broader total-market index. Oldest concentration data referenced: Since 1927 - The hosts say current U.S. concentration is the highest in their data going back to 1927. AI-related stocks' share of S&P 500 returns since ChatGPT: 75% - JPMorgan report cited for the period since ChatGPT launched in November 2022. AI-related stocks' share of earnings growth since ChatGPT: 80% - JPMorgan report cited for the period since November 2022. AI-related stocks' share of capital spending since ChatGPT: 90% - JPMorgan report cited for the period since November 2022. Nortel share of Canadian market index peak: Over 36% - Nortel’s weight in the Canadian TSE 300 at its peak around August 2000. Canadian market CAPE at peak: 60.62 - Nortel-era Canadian market valuation peak, described as unprecedented. TSE 300 decline after Nortel peak: 43% - Drop from September 2000 to September 2002 after Nortel’s collapse. Canadian market recovery time after crash: By July 2005 - Time it took for the Canadian market to recover to pre-crash levels. U.S. market recovery time after dot-com peak in USD: About a decade - Time for the U.S. market to recover from the March 2000 peak in U.S. dollar terms. U.S. market recovery time in CAD terms: Until July 2013 - Recovery timeline when measured in Canadian dollar terms. Average top-seven concentration in 10 largest non-U.S. markets (2015): 40.94% - Average concentration across the 10 largest non-U.S. stock markets as of November 2015. Most concentrated non-U.S. market: Switzerland: 60.11% - Top-seven share in Switzerland as of November 2015. Least concentrated non-U.S. market: Japan: 16.91% - Top-seven share in Japan as of November 2015. Equal-weighted return of 10 non-U.S. markets (2015-2025): 8.44% - Average 10-year return from November 1, 2015 to November 26, 2025 in USD. Active manager outperformance when concentration rising: 30% - Share of U.S. mutual funds outperforming from 1960 to 2023 when concentration was rising, per Michael Mobison’s analysis. Active manager outperformance when concentration falling: 47% - Share of U.S. mutual funds outperforming when concentration was falling, per Michael Mobison’s analysis.

Pivotal Quotes: "the main lessons are diversification and discipline" — Benjamin Felix: Summarizing the episode’s conclusion on how investors should respond to concentration and valuation worries. "If you love everything in your portfolio, you're not diversified enough." — Dan Bordolotti: A behavioral rule-of-thumb used to explain why diversified portfolios must contain both winners and disappointments. "We are not taking position on whether we are currently witnessing a bubble in the U.S. stock market." — Benjamin Felix: Clarifying that the discussion is about historical evidence and risk framing, not a timing call.

Implications: Listeners should not react to concentration headlines by abandoning stocks. The evidence favors staying globally diversified, rebalancing, and maintaining discipline, while moderating return expectations when valuations are elevated.