These days, if you walk through any major data center construction site outside of Phoenix or northern Virginia, you’ll see the same scene: cooling units the size of shipping containers waiting to be hauled inside, contractors squinting at blueprints, and trucks idling in the dust. No one is discussing Nvidia. Power loads, copper, switchgear, and whether the transformer order will be delivered before Christmas are all topics of discussion. That is how the AI revolution truly appears on the ground, more so than any earnings call.
For almost two years, investors have been pursuing the obvious names. Anything with a shiny AI deck, including chips and hyperscalers. However, portfolio managers are now having a more subdued conversation—the kind you hear at conferences when someone leans in a little closer. It goes something like this: the businesses wiring the building might provide the next leg of returns instead of the companies creating the brains.
| Snapshot: The Quiet AI Infrastructure Story | Values |
|---|---|
| Focus Area | AI infrastructure, hardware, and software enablers |
| Key Companies Mentioned | Qualcomm, Vertiv, Nextracker, Interactive Brokers, Roku, TE Connectivity |
| Estimated Hyperscaler Capex 2026 | ~$750 billion |
| Notable Recent Move | Qualcomm’s AI200 and AI250 rack-scale inference launch (Oct 27) |
| Sector Trend | Capital rotating from chip leaders to overlooked builders |
| Analyst Source | Marc Chaikin, Joe Austin — Chaikin Analytics research |
| Market Backdrop | Nasdaq Composite riding a 10-day winning streak |
| Risk Watch | Stretched valuations, geopolitical tension, AI-spend fatigue |
A peculiar illustration of how that way of thinking changes is Qualcomm. It was the smartphone chip company that everyone took for granted for many years; it was the kind of stock that stagnated while more eye-catching names surged ahead. The AI200 and AI250, rack-scale inference systems targeted directly at data centers, were then unveiled in late October. After a roughly 30% increase, the stock returned to the $165–$175 range. It’s difficult to dispute Marc Chaikin’s description of that pullback as an appealing setup. A spike that retains its previous floor typically indicates that someone thinks something has changed.
Vertiv and Nextracker belong to the same broad category of businesses whose unglamorous products are necessary for everything else to function. Solar tracking, power, and thermal control. The AI economy’s infrastructure. It appears that the market is finally beginning to price these names more like picks-and-shovels plays for a multi-decade buildout rather than as industrials.
The second category, on the other hand, is more intriguing and more difficult to define. Businesses are utilizing AI instead of marketing it. Because Interactive Brokers has been discreetly automating its operations for so long, its margins are currently close to 86 percent—a figure that would be expected from a software company rather than a brokerage. Natural-language queries that would have required three analysts ten years ago are handled by its IBot assistant. In the meantime, Roku uses machine learning to determine what 90 million households see next within its ad stack.
These businesses don’t promote themselves as AI players. That’s a component of the appeal. The price typically reflects the loudness of a product’s branding. There is still space when it isn’t.
Being truthful about the risks is important. The Iran situation hasn’t changed, tech valuations are stretched, and the astounding $750 billion hyperscaler capital expenditure figure assumes a level of conviction that could falter if even one significant buyer withdraws. The Nasdaq’s 10-day winning run is satisfying, but streaks come to an end.
Even so, it’s difficult to ignore how much of the real revolution is taking place off-camera as you watch this unfold. No one refers to software platforms, cooling loops, or substations as AI companies. The headline stocks might continue to rise. However, the long-term profits may come from the quieter ones.
