Siebert says AI trade shifts to structural bottlenecks as memory pricing power faces new capacity risk
SIEB•Siebert said AI investing is shifting toward structural bottlenecks, while new Chinese and Korean memory capacity could weaken DRAM and HBM pricing power in 2027-2028. It cited ASML’s third-quarter sales of €11.7 billion and 57.5% gross margin, with demand expected to exceed supply in 2027-2028.
1. AI bottlenecks and memory
Siebert analysis flagged a shift in AI investing toward structural bottlenecks, warning that temporary shortages can fade as new capacity arrives. A memory-stock selloff was tied to fears that new Chinese and Korean output could erode DRAM and HBM scarcity and pressure pricing power in 2027-2028.
2. Equipment demand outlook
Siebert cited ASML’s record third-quarter sales of €11.7 billion and a 57.5% gross margin, and said demand is expected to exceed supply in 2027-2028. Customers are seeking ASML tool orders for 2029-2030, with no comparable second supplier for the most advanced lithography machines.
3. TSMC pricing test
Siebert described TSMC as the next test of pricing power, saying the focus is on how far out clients book leading-edge capacity rather than near-term profit reports.
