They booked their highest-ever foundry backlog in the same quarter they admitted the US fab is delayed to 2026.
Thesis: This is a memory recovery trade, plain and simple. Ignore the foundry narrative—it's expensive noise. The core memory business swung to profit (GM 36.2%) and HBM3E is ramping. Supply constraints in legacy DRAM and NAND are creating real pricing power. You buy this for the memory cycle upswing and the 80% growth in server SSDs. The foundry bleed is priced in; the memory margin expansion is the alpha.
Verdict: LONG — Conviction: MEDIUM
Catalyst: Mass production and shipment of HBM3E 12-high in Q2 2024.
Key Risk: Mobile channel inventory correction in H2 due to Q1 'aggressive selling' (channel stuffing).
The Tell: Management explicitly admitted the mobile weakness they created: 'For mobile, due to aggressive selling by customers in the first half of the year, there is a bit of increase in distribution inventory. So this may act as a constraint on set demand growth in the second half.' They stuffed the channel.
Friction Level: MODERATE_FRICTION — Bulls see a structural AI memory supercycle. Bears see a foundry money pit that can't catch TSMC despite burning billions in capex.
Report not found
The report data is no longer available. Please return to the archive.