Logic folding could create more hybrid-bonding demand for BE Semiconductor than shorter HBM stacks take away. Customers are requesting four- or six-die stacks because HBM is expensive; fewer layers need fewer bonding steps and fit packages more easily with conventional solder, weakening the near-term HBM opportunity. Vertically stacked logic still needs exceptional alignment as it shortens signal paths and cuts energy use. x.com 

Micron cannot scale HBM without crushing its DRAM margins, which is why memory stocks are lagging despite a severe shortage. @CKCapitalxx calls the shortage structural, citing Goldman's view that it is the worst in 15 years and Micron having sold out its entire 2026 production. x.com 

3D-stacked DRAM could constrain bit output even more than today's HBM, preventing memory oversupply beyond 2030. HBM demand already diverts capacity from conventional DRAM; logic-integrated stacks could add pressure if they scale around 2029-30. That comparison depends on the architecture being understood correctly, and memory makers have not settled the DRAM process nodes. x.com 

The memory shortage should hold through this cycle, because every route around DRAM's high margins tightens supply before it loosens it. Memory makers earn roughly 90% gross margins on DRAM, about a 10x markup, and Hot Chips' memory day showed designers engineering around it. But those alternatives are 2-4 years from volume, and the nearest use more wafers per bit: per Micron, HBM uses about 3x DDR5's silicon, and high-bandwidth flash needs roughly 9x the NAND it displaces. x.com 

Memory prices look to be at their peak for SK hynix and Micron as Chinese commoditization catches up. CXMT has made HBM3E and is now only one generation from HBM4, so China will scale its own DRAM and HBM production. Over several years, supply and innovation have always caught up with demand in past cycles, and US work on memory architectures that work around HBM or use less of it is already well underway. That contradicts @jukan05's reading that CXMT's progress tightens commodity DRAM. x.com  … Show more

Memory remains structurally scarce because SK hynix says conventional DRAM and HBM are practically sold out through 2029, while customers request roughly twice the HBM volume it can produce. Long-term agreements cover more than half of 2027 and 2028 capacity, and the company expects the supply-demand imbalance to persist through 2030. x.com  x.com 

Cerebras gains leverage as hyperscaler ASICs tighten HBM supply while its chips skip HBM entirely. Maia 300 uses eight HBM4 stacks per chip; a million chips next year would add about eight million stacks of demand to an already tight 2027 memory market. x.com 

SK hynix looks like the weakest storage short because it has fallen behind. More negative news may follow beyond HBM4; its LPDDR6 is reportedly well behind Samsung's. x.com  x.com 

Memory supply looks constrained through at least 2030 because HBM, vertical HBM and vertical edge-inference memory require more wafers per bit than standard DRAM. Micron's roadmap shows that wafer pressure persisting as those products arrive; its illustrative product timing remains subject to change. x.com 

SK hynix and other Korean memory makers' aggressive price hikes invite customers to optimize memory use, undermining the case for 90% margins despite a long-term shortage. Memory is the only layer raising cost per token. Elon Musk described Micron's future pricing as reasonable and other suppliers' as extortionate, warning that today's customers would not forget; exorbitant long-term contracts compound that risk. x.com  x.com