An opinion piece on Micron and SK Hynix, following on from SK Hynix's recent IPO and recent publications on this subreddit.
**The tell in Micron's contracts nobody seems to be reading**
SK Hynix's CEO said last week that the memory shortage will run past 2030 — right after raising $26.5B, the largest US listing ever by a foreign company, to build the fabs that would end it. Everyone read those two facts as confirming each other. I think they contradict each other, and the mechanism is the oldest one in economics: scarcity raises the price, the price funds the capacity, the capacity ends the scarcity.
But the detail that actually changed my mind is in Micron's prepared remarks. Its sixteen new take-or-pay agreements — running to 2030, \~$100B in remaining performance obligations — carry a **price ceiling set at current market prices**, covering roughly 40% of company revenue. At the peak of the greatest pricing environment memory has ever seen, Micron has agreed to cap what it can charge its largest customers for five years.
You don't sell five years of upside cheaply in a market you think stays this good. You sell it because you want the floor, and the floor has to be paid for. Nvidia hasn't done this. TSMC hasn't. ASML hasn't. Companies with moats don't need to buy stability — they have it.
The other number worth sitting with: last quarter's record came almost entirely from price, not volume. DRAM bits shipped rose \~3%. DRAM prices rose \~62%. Same company, same factories — gross margin went from **-9.1% in FY2023 to 84.9% last quarter.** Nothing about the business changed. The price of DRAM did.
The other number worth sitting with: last quarter's record came almost entirely from price, not volume. DRAM bits shipped rose \~3%. DRAM prices rose \~62%. Same company, same factories — gross margin went from **-9.1% in FY2023 to 84.9% last quarter.** Nothing about the business changed. The price of DRAM did.
I wrote it up properly with the numbers here: [https://wonderfulatafairprice.substack.com/p/micron-stock-and-the-265-billion](https://wonderfulatafairprice.substack.com/p/micron-stock-and-the-265-billion)
Happy to be argued with — the bull case on HBM is genuinely strong and I may be wrong about the mix shift.
