Where Are We Really In The AI Cycle? A Shale Boom Parallel

The US shale revolution gives us an interesting lens to frame where we stand today with AI.

Back in the mid-2000s, shale promised to unlock vast energy resources and reshape geopolitics. Capital flooded in. Drilling rigs multiplied. Everyone talked about energy independence.

But here's what actually happened: euphoria ran hard, companies burned cash recklessly, the cycle peaked, and reality finally caught up. The winners? Usually the last ones standing—the operators who survived the consolidation, not the early hype-riders.

AI feels similar now. The narrative is intoxicating. Capex is scaling vertically. Every firm is scrambling to deploy models. But ask yourself: how many of these deployments will actually generate ROI? How many will become white elephants?

Shale teaches us that bubbles don't invalidate the underlying thesis—energy demand was real, and shale does matter. The issue is timing and execution. Most didn't make money on shale. A few did.

The same math likely applies here. AI won't disappear. But the current pricing and investment frenzy? That's worth watching. History suggests we're somewhere in the middle chapters, not the finale.