Insights
Quarterly Commentary
3Q 2026
Since it burst into public awareness in 2023, artificial intelligence has become the driving force of both the US economy and the US stock market. Chipmakers, data center builders, software companies and energy generators have alternately soared and swooned based on investors’ perceptions of their ability to benefit from AI’s extraordinary growth.
The world’s largest technology companies are committing staggering sums to data centers, advanced chips, computing infrastructure, and the energy required to power it all. The underlying thesis has been compelling: AI will transform industries, boost productivity, and create entirely new markets.
A recent publication from JP Morgan showed that a basket of 42 AI stocks accounts for over half of the S&P 500’s price appreciation, earnings growth and capital spending since the start of 2024. Such remarkable performance attracts ever more investors who want to get in on the action. We are mindful of history, specifically how past transformative innovations – from railroads to the internet – have played out. We don’t pretend to know how AI will ultimately impact markets or our lives, and we prefer to take a cautious approach to investing in the narrative.
Market Review
Stocks performed well until mid-August when surging oil prices accompanied by rising inflation caused a reconsideration of risks. Rising interest rates highlighted the huge and growing national debt. Worried investors pared holdings of smaller and international equities. That caution did not extend to the AI basket. The seemingly inexorable drive to build data centers, produce more chips, and harness AI to make our lives better continued to attract stock buyers as well as lenders eager to finance the AI buildout. The performance disparity added to investors’ anxiety about the possibility of getting left behind in an AI-dominant world. Market concentration spiked with the top 10 constituents of the S&P 500 comprising over 40% of the index. And while tech stocks were driving the S&P 500 to new highs, 7 of the 11 sectors were down on the quarter.
While gasoline prices make for good headlines, they aren’t as important to the US consumer as they once were. However, the longer fuel prices remain elevated, the more inflation spreads through the economy, primarily via higher transportation costs for all goods. Widespread expectations for strong US GDP growth, largely a result of AI data center investments, are also a factor in higher inflation and interest rates. The scenario we’re all hoping to avoid is one in which economic growth fades and inflation remains.
The Way Forward
Under the surface, there are signs of turbulence in the AI trade. While Nvidia and some of the other large AI stocks finished the quarter near all-time highs, many of the smaller AI participants are showing signs of distress (in both equity and credit markets). These developments suggest that investors are scrutinizing the scale of AI-related borrowing, the commitment of future cash flows that until now have supported tech stock valuations, and the assumptions that infrastructure spending will yield a sufficient return on investment.
If AI adoption accelerates and most consumers and/or businesses are willing to pay premium prices for its capabilities, today's spending could look visionary. If adoption disappoints, pricing falls, or computing capacity outpaces demand, some of that infrastructure could become an expensive burden.
For the time being, we’re operating under the assumption that AI will be a great productivity tool, and in some industries, much more. However, we also think many risks have been underappreciated and see inconsistencies in the market narrative. For example, if non-AI businesses are going to pay large sums for AI tools that will greatly increase their efficiency (and hopefully, profits), why aren’t investors interested in their shares? It seems that it can’t be true that the trillions of dollars invested in AI is justified yet there won’t be any beneficiaries other than those directly involved in the buildout. How and when this apparent contradiction gets resolved is hard to predict.
AI may ultimately rank among the most consequential technologies in modern history, but technological significance and investment performance are two different things. The next phase of this boom will be defined less by how much companies spend and more by what they earn. Investors will increasingly need evidence of profitable adoption, durable cash flows, and returns that exceed the cost of capital.
Closing Thoughts
Besides being thoughtful about our level of commitment to stocks dependent on astronomical growth assumptions, we will be watching the timeline of the data center buildout and considering what a slowdown would mean for the broader economy. Will financing come through as expected? Will local or perhaps even national opposition to data centers, driven by environmental, utility bill, or other concerns, cause politicians and regulators to slow construction? Do we have sufficient understanding and trust in AI, the tech goliaths, and the government to give these tools access to our private information?
