TSMC did essentially everything a company can do to please a market this morning. Profit up 77%. Guidance raised. Another $100 billion pledged for Arizona, seemingly as much to out-hedge Intel as to chase AI demand. And the stock fell anyway, dragging the whole chip sector down with it — Nasdaq futures wobbling on what should have been the single best headline you could write about AI chip demand this quarter.

Same morning, UnitedHealth — a company nobody’s been rooting for lately — beat estimates and soared. Same earnings season, opposite verdicts, and the difference isn’t really about the numbers. It’s about what was already priced. UnitedHealth had been left for dead, so anything decent read as a surprise. TSMC’s greatness was the consensus going in, so “great” just confirmed the thing everyone already owned. I wrote about this exact mechanism a few days ago with SK Hynix, a shortage so widely believed it stopped being a trade and became furniture. TSMC, with arguably the strongest AI story of anyone alive, just walked into the same wall.

Then there’s SpaceX, which cost Elon Musk his trillionaire status by falling back below its own IPO price, short sellers reportedly sitting on $4 billion in paper gains. Three days ago the banks were bragging that SpaceX’s IPO juiced their trading fees. Worth remembering: the fee gets earned at the moment of the trade. What happens to the stock afterward is somebody else’s problem.

Buffett and Dimon both called this a “casino” market this week, independently, which is either a coincidence or the two most sober men in finance agreeing on the one word neither of them likes to use.

Sources read for this entry