The asset managers I’ve recently spoken to have an odd silence in their offices, the kind that usually occurs when something obvious is happening and no one wants to be the first to call it strange. AI is present in every fund prospectus, every earnings call, and every polite but excessively lengthy conversation during cocktail hour.
However, the more you dig, the more you see that the majority of investors aren’t genuinely placing bets on AI. They are placing bets on about seven businesses that just so happen to enable AI.
| Topic Snapshot | Details |
|---|---|
| Subject | Artificial Intelligence as an investment theme and an economic force |
| Current Phase | Shifting from experimentation to real-world deployment in 2026 |
| Market Concentration | S&P 500 weighting in top names near historic highs |
| Notable Non-Tech Adopters | John Deere, Siemens, UPS, FedEx |
| Historical Parallel | The electricity and internet revolutions of past eras |
| Key Voice | Joe Davis, Vanguard’s Global Chief Economist |
| Microsoft Perspective | Aparna Chennapragada, Chief Product Officer, AI Experiences |
| Theme for 2026 | AI shifting from instrument to collaborative partner |
| Risk Factor | Stretched valuations among mega-cap AI leaders |
| Opportunity Factor | Productivity gains for companies applying AI outside tech |
The similarities to other periods in market history are difficult to ignore. The utilities humming in the background did not build the fortunes when electricity rewired American industry a century ago. They were constructed by General Motors and Ford, two businesses that rebuilt their entire purpose in the face of the new force. This is something that Vanguard’s Joe Davis has stated time and time again, and the pattern appears in his charts in a way that seems almost unyielding. The story of picks and shovels is reassuring. It’s also lacking.
Though we often forget it, the internet boom unfolded in a similar manner. The routers and servers weren’t the best performers between 2000 and 2010, once the bubble had finished embarrassing everyone. Retailers, broadcasters, logistics firms, and other companies that figured out how to use technology for their own ends were among them. The second act, which proved to be the longer and more lucrative one, was largely missed by investors who left during the crash.
If you know where to look, you can see the change if you drive by a John Deere field today. Herbicide use is being reduced by up to 90% thanks to the “See & Spray” system, which scans row after row of soybeans. That is a margin story, not a detail from a press release. In order to identify flaws that a human eye would miss, Siemens installed AI-powered vision systems on its manufacturing floors. FedEx and UPS are discreetly rerouting trucks using models that learn from the traffic, weather, and errors of the past. These businesses don’t trade at multiples of chip stock. The front page does not feature any of them. Maybe that’s the point.
It is difficult to hide the risk in the current market. The price-to-earnings multiples on the AI leaders are at levels that require near-perfect execution for years, and the S&P 500 is more concentrated in its top holdings than it has ever been in living memory. Looking at the numbers, it seems like investors have factored in both the revolution and its successful conclusion. Once, Tesla heard the same chorus. Cisco did the same. While the chorus isn’t always incorrect, it’s also not always correct.
A different phase, which Microsoft’s Aparna Chennapragada characterizes as collaboration rather than response, appears to be emerging in 2026. She discusses three-person teams managing international campaigns and AI as a digital coworker that manages the tedious tasks while humans direct the creative process. “The future isn’t about replacing humans,” she recently stated. “It’s about amplifying them.” It’s still unclear if that turns out the way she envisions or in a messier form that no one has yet foreseen.
From a distance, the safer bet feels more like identifying which dull, unglamorous company is suddenly operating with a third fewer employees and twice the throughput than picking the next Nvidia. The next ten years will most likely hide there. Investors will eventually figure it out. Usually, they do, albeit a bit later than they ought to.