Five Rules for Investing in AI Infrastructure Without Getting Fooled by the Theme

I've noticed a recurring pattern while analyzing AI-related stocks recently. Investors get one idea right, then invest around it in exactly the wrong way.

They hear that power matters to AI, and then they throw utilities, cooling names, electrical equipment, generators, and every other adjacent stock into one giant bucket. That's not smart. A real bottleneck is not the same as a loose theme.

Owning the wrong stock for the right theme — that's the trap many investors fall into. So here are five rules I follow to avoid that trap in AI infrastructure investing.

1. Start With the Bottleneck, Not the Ticker

If you can't explain the physical constraint in one sentence, you probably shouldn't be buying the stock.

For me, that sentence is simple: "AI demand can be strong, but deployment can still slow down if enough power can't get to the rack."

That's the starting point. Not a stock tip, not a hot name — a physical constraint that limits the system. Understand the constraint first, then find the companies that solve it.

2. Favor Companies Close to Deployment

The most investable businesses are those that help move a project from planned capacity to live capacity.

Apply this filter and your stock list gets much shorter. Being related to AI power isn't enough. What matters is where a company sits in the chain. The closer to the deployment site, the more direct the investment thesis.

3. Separate by Role, Not Category

Utilities, electrical infrastructure, rack support, on-site power — these are all different things. Don't flatten them into one category.

Each area may benefit to some degree, but not in the same way and not with the same magnitude. Your portfolio weight should reflect that difference. Ask: "What exact problem does this company solve?" You need a specific answer. That's how you separate a real bottleneck winner from a stock that merely sounds related.

4. Check That Business Evidence Matches the Story

Investment themes start with compelling narratives, but they only become real when business evidence backs them up — growing backlogs, accelerating orders, time-to-power mentions on earnings calls, rising high-density data center demand.

If this were just a clever story, it wouldn't deserve attention. What makes it worth investigating is that the pressure is showing up in real-world business metrics.

5. Don't Let Theme Excitement Override Valuation

This may be the most important rule. Even if the thesis is correct, a stock can still disappoint if the entry price is wrong.

That's how good ideas become bad trades: getting so excited about the theme that you stop paying attention to valuation, role clarity, and execution quality. This theme is no longer fully hidden. Some of the better names may already carry elevated expectations.

Selectivity, risk control, and price discipline — these complete the framework.

---

Most investors still see AI as primarily a chip story. I think that's too narrow. AI is also becoming a physical infrastructure story. If power emerges as a real gating factor, the companies solving that problem may matter more than the market currently realizes.

That doesn't mean every adjacent stock wins. It doesn't mean the thesis carries no risk. And it certainly doesn't mean you abandon discipline. Widen the lens, but keep the investment principles tight.