They raised CapEx to $20B to chase demand they explicitly admit they can't fully meet until 2027.
Thesis: The bottleneck is physical. Clean room lead times are lengthening. HBM demand is cannibalizing DDR5 supply at a 3:1 ratio. This isn't a demand story. It's a supply shock. Customers are signing 'unprecedented' multi-year contracts with locked pricing because they have no alternative. The margin ceiling at 68% isn't a problem. It's a cash flow engine. You own the shortage.
Verdict: LONG — Conviction: HIGH
Catalyst: HBM4 production ramp in CQ2 2026. Yield confirmation will be the trigger.
Key Risk: HBM4 yield ramp failure. If they miss the window, that capacity becomes dead weight while competitors catch up.
The Tell: Sanjay Mehrotra refusing to answer the HBM market share question. 'We are not really going to be specifying the share.' This contradicts prior commitments to match DRAM share. They know they're behind on volume relative to the total market, even if sold out.
Friction Level: MODERATE_FRICTION — Bulls see pricing power in the supply shortage. Bears see a margin ceiling at 68% and execution risk on the HBM4 ramp.
Report not found
The report data is no longer available. Please return to the archive.