Just realized something that most AI investors are completely overlooking right now. Everyone's focused on chip stocks like Nvidia and TSMC, but there's a massive bottleneck nobody's talking about - the actual power needed to run all these AI data centers.

Think about it. Rand Corp. estimates AI data center power demand hits 68 gigawatts this year, then jumps to 327 gigawatts by 2030. That's insane growth. But if you're only buying semiconductor stocks, you're missing half the picture. The real constraint isn't just processors anymore - it's electricity, cables, cooling, infrastructure. That's where the actual money might be.

I've been watching NextEra Energy pretty closely. Most people think of it as just a boring utility company, but what they're doing right now is pretty interesting. They operate Florida Power & Light, serving over 12 million customers, and they've positioned themselves as a major power supplier for hyperscalers building AI data centers. They just signed a deal with Google Cloud back in December to build and power new AI infrastructure. The company's planning to deliver an additional 15 gigawatts of power to data centers by 2035, with 6 GW coming from gas-fired plants. Their CEO even said he'd be disappointed if they don't double that goal to 30 GW.

The numbers are solid too. Last year net income was $2.97 billion versus $2.3 billion the year before. EPS came in at $1.44 versus $1.12 in 2024. Management's guiding for at least 8% compound annual growth through 2032, plus 10% dividend growth this year. For a utility, that's pretty strong.

But here's the other piece - data centers don't just need power, they need connectivity. That's where Credo Technology comes in. They make these Active Electrical Cables that basically move data between chips way more efficiently than traditional copper wiring. They just partnered with TensorWave to supply their next-gen AI cluster infrastructure. Their recent earnings showed revenue of $268 million, up 272% year-over-year, with net income of $82.6 million versus a loss of $4.2 million the prior year. They're guiding for $335-345 million in the next quarter.

What I find interesting is you get two totally different plays here. NextEra's a regulated utility with predictable growth and a solid dividend. Credo's got a specialized product that's critical infrastructure for the whole AI buildout. If that Rand projection is anywhere close to accurate, both of these are positioned to benefit massively from the AI infrastructure wave without being directly exposed to chip competition.

Most people sleeping on the power angle when it comes to AI. Worth thinking about if you want some diversification beyond just chip stocks.