Looking Back and Looking Forward
Across the globe, it was a solid Q3 for equities. While the U.S. continues to lag other major markets on a year-to-date basis, it saw a revival in Q3 on the prospect of Fed rate cuts in the context of a still-acceptable, albeit weakening, U.S. economy. The main driver of U.S. equity markets, however, continues to be (surprise surprise) AI-related stocks.
Going back in time, it’s worth noting that AI has driven the majority of S&P 500 returns over the past three years. As J.P. Morgan Asset Management recently noted, 75% of S&P 500 returns since ChatGPT launched (Nov 2022) have come from just 41 AI-related stocks. Those 41 stocks also account for 80% of corporate earnings growth and 90% of growth in capital expenditures since Nov 2022. The other 459 members of the S&P 500 have massively underperformed over the past three years. In short, it’s been all about AI.
Recently however, we’ve seen some clear signs that point to a bubble, and we are no stranger to bubbles, having seen more than one inflate and then deflate over the past thirty years (NASDAQ and sub-prime being the two most notable). We see stark similarities today. Valuations are high and market breadth is narrow, as just discussed. Markets participants are buying AI stocks as if every company will win, while history tells us that only a handful of companies will ultimately come out on top. It’s not even clear that the massive spending on AI will pay off, as highlighted in the recent MIT study that showed 95% of generative AI projects fail to show any measurable impact on profitability. Companies are clearly spending a ton of money in the hope that AI delivers outsized benefits – but what if it doesn’t?
If we’re right and this is, in fact, an AI bubble, it’s impossible to predict when and how it ends. We do know that we’re heading into a seasonally strong period in markets, and we expect corporate earnings in the U.S. will continue to grow in Q4 (despite tariff and other headwinds). Even though the market “Generals” (aka the Mag7) have gone from being big generators of cash to large spenders of cash (via their AI initiatives), investors may continue to support their valuations given their dominant stature, and this could buoy the overall market. To summarize, markets could conceivably hang in longer and rise further, but cracks are starting to show.
And if it is a bubble, what ultimately pops it? The most obvious candidate is the U.S. Federal Reserve. How might that look? By several measures, financial conditions today are loose, and the Federal Reserve is poised to loosen even further by cutting rates at its October and/or December meetings. The response from the bond market to such rate cuts could be similar to what happened in the fall of 2024. Back then, the Fed cut the Funds rate by a total of 100 basis points, yet the 10-year Treasury yield rose (yes rose) around 100 basis points. The simple message of the bond market was “Fed, you don’t care enough about inflation, so we’ll build higher inflation expectations into the long end”. In retrospect, one could argue the Fed made a policy error by lowering rates. We could see a similar episode if the Fed follows through with additional cuts in October and December. Higher 10-year Treasury yields would stress the economy, the equity market, and any bubbles that have formed.
We’ll close by simply stating that we hope we’re wrong. It would be better for all, especially our clients, if things play out calmly and smoothly. But as the saying goes, if it looks like a bubble…and it acts like a bubble…
Our Core Model Portfolios
In our core capital growth models, we remained underweight U.S. equities and overweight non-U.S. equities (e.g. Canada, Europe, Japan, China), which have broadly outperformed the U.S. year-to-date. As we’ve mentioned in the past, we could be in the early stages of a multi-year rotation away from U.S assets and toward non-U.S. assets. We also continued to hold significant exposure to precious metal-related securities, which have been one of the top-performing asset classes in 2025.
In our core income models, we sold a small portion of short-duration fixed income and added to our exposure in Canadian dividend-growth equities. We tend not to hold commodity-related securities (like precious metals) in our income models to substantially eliminate the volatility inherent in commodity-based businesses.
At time of writing, we don’t anticipate significant adjustments to our core model portfolios for the remainder of 2025, as the models hold well-diversified equity exposure in combination with portfolio stabilizers like short-duration bonds and precious metals assets. However, given the “bubblish” environment we described earlier, we stand ready to act with purpose and speed, if and when conditions fundamentally change.