#ai-capex

The Fed held at 3.50-3.75% for a fifth straight meeting, but the vote was 9-3 and all three dissents wanted a hike. That is the most dissents since 2016. Underneath it is a question the Fed has openly admitted it cannot answer yet, and it has commissioned task forces to go find out.

Apple passed Nvidia on Monday and touched $5 trillion on Tuesday, up 25% this year against Nvidia's 2.6%. The same week, China started building the lithography tools it was never supposed to have. Both events are the market re-pricing one question: is AI capex a moat or a treadmill?

Hyperscalers depreciate AI servers over five to six years. If the real economic life is shorter, reported profits across Big Tech are too high. Here is how depreciation actually works, what the bears get right, what they get wrong, and how to check the assumption yourself in a 10-K.

Amazon raised at least $25 billion in the bond market to fund AI infrastructure, on top of the $54 billion it already borrowed in March. Hyperscalers used to pay for capex out of cash flow. Now they issue debt, and fixed coupons meet uncertain AI revenue.

The S&P 500 gained 9.6% in the first half of 2026 while the Magnificent Seven, a third of the index, went up 2.6%. That gap is the whole story, and most recaps skip right past it.

Nvidia was the obvious AI trade. The non-obvious one is whoever plugs the grid into the building. Hyperscalers are guiding to $725 billion of 2026 capex, and the binding constraint is no longer chips. It's turbines, transformers, and interconnection queues.

Amazon, Google, Meta, and Microsoft guided to roughly $725 billion of 2026 capex, up 77% in a year. Some of them show you the revenue on the other side of the spend. Some of them call ROI 'a very technical question' and change the subject. The earnings call is where the tell shows up.