Just caught something that might matter for anyone paying attention to where real money is flowing in the intelligence market right now.

The five biggest tech companies are about to drop over $700 billion on AI infrastructure this year. Amazon, Alphabet, Microsoft, Meta, and Oracle. That's not some projection or analyst guess—these are their actual budgets. Amazon alone is going $200 billion, Alphabet $180 billion, Microsoft $151 billion. The scale is honestly hard to wrap your head around.

Here's what's wild though. Everyone talks about Nvidia and the chip designers making bank off this AI boom. But the real play might be further up the supply chain.

Think about it. Two-thirds of all that capex is going straight to GPUs, CPUs, and AI accelerator chips. That's roughly $450+ billion flowing to semiconductor manufacturing. The companies like Nvidia design these chips, sure. But someone has to actually build them. And there's really only one company that can do it at the scale and sophistication required for cutting-edge AI chips.

Taiwan Semiconductor Manufacturing is basically the only player that matters here. They're not just another fab. They've got the technological edge to print transistors smaller and more efficiently than anyone else. That means they're the bottleneck for the entire intelligence market supply chain. Every major AI accelerator chip that goes into these data centers likely came out of TSMC's facilities.

What's interesting is management already raised their growth outlook before seeing these updated capex numbers. They're now guiding for 25% annualized revenue growth through 2029, up from 20% a year ago. And that was before knowing the hyperscalers would be spending this aggressively on the intelligence market.

The math is straightforward. More spending on AI chips means more demand hitting TSMC. More demand means pricing power. Stronger margins. Faster profit growth. Management's guidance could end up being conservative if this capex surge accelerates as expected.

Stock's trading at all-time highs, which makes sense. But even at 26 times forward earnings, it's not expensive considering the growth runway in the intelligence market. Those earnings estimates might actually be too low given what we just learned about big tech's spending plans.

If you're looking for exposure to the AI infrastructure build-out without picking individual chip stocks, this is probably the most direct way to play it. The company has the technology moat, the scale advantage, and now visible demand visibility that most investors haven't fully priced in yet.