A Rotation Is Not a New Earnings Model

I can see how Wednesday becomes a tidy narrative: the election changed Washington, money rotated, and the market chose a new set of winners. The price record supports the middle part. It does not finish the last part. DIA rose while SMH fell. That is evidence about what investors paid for different baskets by the close. It is not a new set of operating forecasts for every business inside them.

A political development can change the range of plausible regulation, public spending or taxes. A quarterly report can change the expected economics of producing and selling a specific product. I want those risks kept separate. If I cut a semiconductor estimate because I expect fewer units to ship, I should name that assumption. If I instead raise the discount rate because the range of policy outcomes widened, I should not disguise it as weaker customer demand. Either change can reduce a valuation; they would be tested with different evidence.

AMD’s own guide gives a concrete case: approximately $17 billion of next-quarter sales alongside an adjusted gross-margin estimate of about 55%. That is a forecast for AMD. I cannot transfer it to Micron merely because both companies sit in a semiconductor basket. Memory pricing, processor shipments and the cost of assembling a system are related, but they are not the same earnings driver. A strong forecast at one supplier may leave a customer with a higher bill.

I would write down what must happen for each revised assumption to be wrong. Does an order have to be canceled? Does a rule have to be enacted? Does a supplier have to report a different product mix? If the only answer is that the stock would need to recover, I have described a price opinion, not a business test. Wednesday’s unusually wide benchmark split makes that distinction harder to ignore. It does not relieve me of making it.