These memory makers—$MU , $SNDK , and SK Hynix—are being hammered despite their low P/E ratios.
They have a ton of long-term orders on the books, with inventory already sold through 2030, and they’re still getting hammered.
Honestly, it’s pretty stupid, but that’s the market for you—it only looks ahead.
More and more people are using $AMD to train large models, and migrating from CUDA to ROCm has never been this smooth.
They have a ton of long-term orders on the books, with inventory already sold through 2030, and they’re still getting hammered.
Honestly, it’s pretty stupid, but that’s the market for you—it only looks ahead.
More and more people are using $AMD to train large models, and migrating from CUDA to ROCm has never been this smooth.