Everyone is calling the AI trade crowded. The data says otherwise.
The Narrative vs. The Numbers
Every week I see another headline declaring the AI bubble has popped. And every week, the actual spending data tells a completely different story.
Let's look at what's actually happening.
Capital Expenditure Is Accelerating
The hyperscalers (Microsoft, Google, Amazon, Meta) have collectively guided for over $200 billion in capex for 2026. That's not a typo. Two hundred billion dollars, most of it going directly to AI infrastructure.
When companies are increasing capex guidance quarter over quarter, that's not what a bubble looks like. Bubbles are characterized by speculative investment. This is operational investment — these companies are building because they see actual demand.
The Revenue Is Real
NVIDIA just reported another quarter of 100%+ revenue growth. Microsoft's Azure AI revenue is growing triple digits. This isn't 1999 where companies were burning cash with no business model. The AI infrastructure buildout has real, measurable revenue behind it.
What I'm Watching
The names I think still have the most upside from here:
- NVDA — Still the picks-and-shovels play. Blackwell architecture is a step function improvement.
- PLTR — The enterprise AI deployment leader that nobody wants to admit is winning.
- CRWD — AI-native security is becoming non-optional as attack surfaces expand.
Don't let the narrative distract you from the data.