bubble boi
Semiconductor Market Commentary, Thru Project Mascot
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.comÂ
Prepaid contracts and demand-paced fabs weaken a broad 2028 memory-glut case. SK hynix says over half its 2027-28 capacity is signed with deposits; 'sold out through 2029' includes negotiations. Samsung targets 60-70% contracted capacity. Micron's three projects begin output in 2H28; tooling follows demand. Reuters says CXMT's 20-30%-yield HBM3E may ship this year to export-restricted Chinese buyers, but performs like older HBM3. Commodity DRAM risks 2028-29 oversupply; NAND could ease in 2H27. … Show more
Data-center compute's growing share of leading-edge wafer starts and harder silicon-to-systems engineering could drive wafer-fab equipment and Applied Materials sharply higher. At Applied Materials' EPIC event, a slide projects data-center wafer starts gaining share in leading-edge logic and DRAM through 2030; the EPIC unveiling adds to the excitement. x.com x.comÂ
High-bandwidth memory supply may not keep pace with AI demand, supporting the memory makers. Anthropic's S-1 says SK hynix, Samsung and Micron are expanding capacity, but warns that large buyers or competitors could reserve much of the available output. x.comÂ
AI data-center investment needs outside AI spending to rise 33-fold by 2030 to break even. Hyperscalers hold $4.5 trillion in debt and future liabilities, but corporate and SaaS token spending outside circular deals is only $60 billion annualized against a $2 trillion break-even ARR. SK hynix says memory shortages may last until 2031, raising capex and CoreWeave default risk; Mitsubishi warns turbines are booked through 2031 and cost 400% more, delaying CoreWeave and Oracle data centers. x.com … Show more
Rack-scale AI pods need more wafers and advanced packaging, favoring Applied Materials' equipment portfolio. Applied's slides model two to four times as many wafers per agentic pod; the company says it leads HBM packaging process steps and expects over $2.4 billion in advanced-packaging revenue in 2026. Wiring uses about half of chip power, adding a need for equipment innovation. Applied's WFE share should grow; its optics opportunity depends on scale toward the end of the decade. x.com x.com … Show more
Micron's ordinary memory could see tighter supply because HBM consumes far more factory capacity than conventional DRAM. Meeting AI demand could thus extend Micron's opportunity beyond AI memory. x.comÂ
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.comÂ
Leading-edge wafer demand is bullish for semiconductor equipment makers because TSMC is already at over 100% utilization on those nodes. Per Ben Bajarin, rack-scale pods drive more wafers across categories. Close collaboration at the leading edge also gives equipment makers and TSMC visibility into plans years ahead. x.com x.comÂ
The OpenAI ecosystem is lifting a broad supply chain, from Microsoft's $300B and Oracle's $200B backlogs to Nvidia and AMD, whose MI450 has a massive customer. Micron trades at 7x earnings that are still rising, Korean memory is fully sold out and TSMC is king. ASML has no capacity for the next 12 months, Lam is the best memory supplier and KLA the best equipment maker, while Applied Materials depends on China returning given 40% exposure. Optical, chemical and test suppliers also gain. x.comÂ