Me Memory Loading... : Investor Sentiment and Bull/Bear Views

Author validates prior bottom-call on memory names, citing 80% rally and recovery to ATHs, referencing market makers hedging autocallable flow as the driver. Implicitly endorses holding memory longs through the recovery.

There is quite clearly a shortage of memory, both RAM memory and flash memory. Near term, everyone in the industry is needing to raise prices, and Google will also need to raise its prices like other industry players.

Chinese AI models like Moonshot AI's Kimi K3 achieve extreme memory efficiency through software innovation, creating a clear negative for the ongoing bull market in memory chips. This efficiency pushes the memory market toward DRAM/DDR5 segments where China is more competitive and can bring new capacity to bear, threatening oversupply and weighing on memory pricing and demand.

All memory tiers look exceptional right now, driven by strong HBM demand for AI workloads; memory is a very bullish sector.

More deleveraging lies ahead for the momentum factor, particularly in semiconductors and memory, suggesting the recent one-day bounce was not a near-term bottom.

Memory is a capital-cycle sector where current shortages and rising prices could be undermined by Chinese supply ramping; Kyle is not involved long or short but says anyone in memory needs to study Chinese production closely.

Power constraints are easing because turbines and medium-speed reciprocating engines can be brokered or added, but leading-edge fabs and memory cannot be quickly expanded. Memory makers have not built new fabs since 2022, and new capacity takes years; TSMC's leading-edge expansion is also limited. The bottleneck shifts back from power to semiconductors by 2027, favoring TSMC and memory suppliers.

Memory is becoming a strategic asset and critical enabler of AI rather than just a component. AI adoption is broadening from data centers to edge devices, PCs, smartphones, self-driving cars, and eventually humanoids, all of which require more and faster memory. Because new greenfield fabs take years to build and supply is already tight, this creates a long-term structural demand and tight-supply environment that supports memory value and pricing.

Memory has moved from a smartphone-driven famine/feast cycle to an AI-driven paradigm shift. As more tokens are created and data must be stored and moved, there is not enough storage, and high-performance DRAM/storage is becoming the bottleneck. Over the next 12 months, memory and storage costs in expensive AI racks should account for more than compute. Because incremental capacity takes about two years to add, prices are rising more than expected, and memory manufacturers should preserve or expand margins in 2026.

Nvidia's CES comments on memory demand give the bullish data storage/memory narrative more duration, suggesting demand can extend beyond calendar 2026 into calendar 2027 and making the story a multiyear one with improved visibility.

Kristina says that with Nvidia lacking a next big catalyst and investors taking profits in many AI plays, stockholders are rotating into other AI picks and shovels, especially memory, which has already seen a gigantic surge.