Just been tracking something pretty interesting about where the really large tech companies are putting their money right now, and it's worth paying attention to.

So these massive hyperscalers are collectively planning to drop over $650 billion this year on AI infrastructure. That's not casual spending—when companies commit that kind of capital, they're betting on serious returns. Let me break down where the real opportunities are hiding.

First, the obvious play: chip makers. Nvidia's GPUs basically run the AI show, and their CUDA platform gives them a moat that's hard to compete with. AMD's making moves too, especially in AI inference where they've landed some big deals. But here's where it gets interesting—Broadcom is quietly winning with custom AI chips. They helped build Google's TPUs, and now they're working with other companies on their own custom silicon. That's a multi-year growth story.

Then there's TSMC. They've got this near-monopoly on manufacturing advanced AI chips, which means pricing power. Not to overlook: memory makers like Micron are seeing huge demand for high-bandwidth memory. HBM requires about 3x the wafer capacity of regular DRAM, so the entire memory market is tight right now. Micron's locking in long-term commitments, which makes their business way less cyclical than before.

Now for the large tech companies themselves—Alphabet, Amazon, and Microsoft in cloud computing are all spending heavily and expecting strong returns. What's smart here is that all three are embedding AI into their core businesses. Google's using Gemini for search, Microsoft's AI copilots are driving enterprise software growth, and Amazon's using AI and robotics to boost efficiency. Meta's doing similar things with their recommendation algorithm and AI-powered ad tools.

Here's a less obvious angle: energy infrastructure. AI data centers need massive amounts of power, and Energy Transfer's sitting on cheap natural gas assets in the Permian Basin. They're positioned for a ton of high-return projects around data center power needs. Plus they're yielding over 7%.

The takeaway? The infrastructure play around this AI build-out is broader than just the headline names. Chip makers, memory companies, the foundries, and even energy infrastructure all benefit. If you're looking to play this trend, there are more angles than people realize.