rubicon59
Technology & Semiconductor Equity Commentary
Agentic AI's repeated context favors cache memory and storage demand for Micron, SK hynix and Sandisk over fresh GPU computation. Per @demian_ai, about 95% of input tokens for DeepSeek V4.1 Flash on OpenRouter were cache hits; storage matters unless routing and caching keep almost everything in GPU memory. New products such as 3D DRAM and HBM-on-logic could be the final trigger for a full memory-stock re-rating. x.com x.comÂ
SanDisk's data-center flash growth shows AI inference is becoming a larger storage customer. Data-center revenue reached $4.8bn, up 61% in a quarter and over 40% of sales; total revenue of $11bn and EPS of $49.51 beat expectations. Guidance of $12.4bn revenue and $56 EPS suggests feared supply normalization has not yet eroded pricing. Overweight, with a $2,700 target. x.com x.comÂ
Networking, power and compute look more durable than cyclical memory in the AI buildout. Lumentum's revenue compounded year after year across the 2010s and 2020s, unlike Micron's; AWS revenue also compounded. Lumentum, Nebius, Bloom Energy and Credo should heavily outperform SanDisk and Micron from here, though AI itself remains heavily doubted. x.com x.comÂ
SanDisk CEO David Goeckeler's report of customer signals beyond fiscal 2030 adds storage to the long-duration AI hardware-demand case. He says those signals come through new-business-model partnerships; Micron and SK hynix in memory and Lumentum in photonics were already cited as signs of demand beyond 2030. x.comÂ
SanDisk benefits as agentic AI explodes context data. Most of that KV cache cannot live in HBM, so it spills to QLC NAND, which is slower but the only tier economical enough at that scale. x.comÂ
SanDisk's investor-day projections look too rich, as forecasting four years out on flash offload in AI inference is speculative. Measured against an 18TB attach-rate max bull case, now more like 8-10TB, they imply SanDisk assumes KV cache grows with users and context, many HBF customers, which looks unlikely now, or rock-solid four-year long-term agreements. The core disagreement is how big and how fast flash offload becomes ubiquitous. x.com x.comÂ
NAND-based memory hierarchies that handle ever larger AI context and per-organization model tuning are driving Kioxia and SanDisk higher, and Phison, which plays well in the space, could follow. Tiered architectures are already happening. People use Phison's aiDAPTIV with SSDs for post-training, and its NAND is built for massive endurance, slow but long-lasting. Some data must be instant while other data can wait seconds or longer, so memory needs span a wide continuum below HBM. x.com x.comÂ
A worsening memory shortage in 2027 could favor Micron, SanDisk and SK hynix as AI stocks. Samsung warns that the shortage will worsen then. x.comÂ
SanDisk's 2030 guidance has zero credibility, like Salesforce's or ServiceNow's long-range targets, and looks like its Oracle-RPO moment. A potential compute overbuild could arrive well before then, along with massive new memory capacity from everyone, including China. The guidance is still good for a big one-day gain. x.com x.comÂ
Sandisk benefits because AI needs more storage, and long-term customer commitments could make NAND demand and pricing more predictable. x.comÂ