Is the market still pricing SanDisk (SNDK.US) as a "cyclical stock"? Analysts say: Micron's financial report confirms a memory supercycle, and the August financial report may be the trigger point.

San Disk (SNDK.US) is becoming a direct beneficiary of the exploding demand for AI inference. The critical role of enterprise SSDs in large-scale storage workloads is increasingly prominent, and the company's management is replacing volatile spot pricing with long-term supply agreements, significantly enhancing revenue visibility and reducing the cyclical nature of traditional NAND business. The upcoming August financial report may serve as an important catalyst - if management confirms a strong demand for enterprise SSDs, expands long-term agreements, and raises guidance, the stock price is likely to receive further support. Financial analyst Yiannis Zourmpanos stated that with the commercialization of the Stargate QLC SSD and the advancement of future High Bandwidth Flash technology, San Disk is well positioned for the next phase of AI infrastructure development. Despite a cumulative 34% increase in stock price since the last coverage, the market is still using an outdated cyclical valuation framework, and San Disk's valuation may still be underestimated. Investment logic: Micron's financial report confirms the supercycle, AI hardware changes memory economics Micron's latest financial report fully demonstrates how AI hardware infrastructure is reshaping the economic model of the memory industry: companies with top-tier enterprise SSD product lines and good supply discipline are reaping rich rewards. SanDisk is evidently a beneficiary. While the market's focus is still on GPUs and HBM, the impending AI bottleneck will be storage. Agentic AI's reliance on high-performance NAND, enterprise SSDs, KV caches, and storage is increasing day by day. Micron has acknowledged this trend, and despite SanDisk's 34% increase in stock price since the last coverage, I believe the market still underestimates its potential. Micron's performance far exceeded expectations: NAND revenue surged by 360%, supply constraints may continue until 2027 The market generally expected strong performance from Micron, but the actual data was even more impressive: revenue reached $41.5 billion, with NAND business growing by over 360% year-on-year; enterprise SSD revenue exceeded $5 billion for the first time, more than doubling from the previous quarter, signaling rapid growth in the storage sector as part of AI infrastructure. However, the most crucial point in Micron's financial report was that management completely denied a traditional paradigm of the memory industry. Micron did not indicate that supply shortages would normalize over time, but pointed out that the industry's tight supply situation may continue until 2027. This judgment may be more important than the quarterly performance itself as it overturns one of the basic premises of semiconductor investment. In the past, each peak of profitability would stimulate memory manufacturers to expand production, eventually leading to a deterioration in pricing environment. The situation is now completely different: the construction of advanced NAND capacity requires massive capital expenditures, and most of the industry's capacity is shifting towards HBM production. It is not oversupply, but the combination of manufacturing complexity and strong AI demand that is causing supply constraints. The industry environment formed as a result is much healthier than the traditional cyclical nature investors are used to. AI infrastructure surpasses GPUs: The era of inference calls for massive storage When it comes to AI infrastructure, most investors first think of NVIDIA Corporation accelerators. Model training requires GPUs and HBM, but inference brings new challenges: to efficiently process millions of queries, large language models require massive storage support. KV caches, retrieval databases, and persistent model storage are becoming key to the economic feasibility of inference. SanDisk's management recently discussed this at a conference. CEO David Goeckeler stated that customers are no longer just concerned about computing demands, but are seeking the optimal mix of computing, DRAM, HBM, and NAND in inference workloads. This is a significant shift in industry logic - storage has become an indispensable part of AI architecture. Enterprise SSDs currently account for about a quarter of SanDisk's business. More crucially, the new QLC Stargate platform has just begun commercial deliveries, amidst Micron's strong performance in enterprise SSDs. Enterprise SSDs do not compete with HBM but complement it, providing a scalable storage layer for inference. Business model quietly transitioning: From spot pricing to long-term agreements, reducing profit fluctuations Perhaps the most underestimated change is not related to technology. For decades, NAND manufacturers have produced chips for transactional markets, with quarterly pricing fluctuations directly affecting manufacturers' profits, resulting in huge performance volatility and suppressing the valuation levels of memory manufacturers. SanDisk is quietly transforming its business model: management has signed multiple long-term supply agreements. Such agreements set price floors while allowing participants to share the benefits of rising market prices. As CFO Luis Visoso explained, the company's goal is not to sacrifice pricing for the sake of contract duration but to establish continuous supply, benefiting manufacturers and super-scale customers alike. The agreements include fixed price and price range clauses, protecting all parties from the impact of price fluctuations. It is worth noting that Micron is also advancing the same strategy through "strategic customer agreements." When the two industry leaders adopt the same business model, it is no longer an individual experiment but a trend across the industry. The industry is moving towards a new structure with lower volatility and stronger revenue predictability. If this model continues to be adopted, NAND is likely to gradually evolve into an infrastructure business - long-term customer agreements are sufficient to support valuations far higher than those of traditional cyclical commodities manufacturers. Why might SanDisk's August financial report be more important than expected? Micron's financial report has increased the anticipation for SanDisk's quarterly performance, to be announced in August. The market consensus already expects a set of impressive data, but the performance itself may be only a secondary reason for excitement. Zourmpanos pointed out that investors should focus on three main aspects of the financial report: First, management's comments on demand for enterprise SSDs will clarify whether Micron's strong growth is a common industry trend or a unique case for the company. Based on recent discussions by management, there is full confidence in strong industry demand. Second, any progress on additional long-term agreements should be monitored. Each additional agreement will reduce profit fluctuations and improve revenue visibility. Third, progress in the commercialization of the Stargate platform and the development of high-bandwidth flash technology will provide more clues about the direction of evolution in AI infrastructure. High-bandwidth flash is still in its early stages of development, but management already views it as a long-term solution to the challenges of AI inference - this technology combines significantly higher storage capacity with greatly increased bandwidth, surpassing traditional NAND. Commercialization may take some time, but it represents an important option value not yet factored into the current valuation of SanDisk. Valuation: Depending on a core question Despite its remarkable rise, Zourmpanos believes that SanDisk's valuation is still attractive. The market is still using old memory industry multiples rather than viewing it as an AI infrastructure company. At first glance, the company's valuation may seem high: with a forward P/E ratio of 32 times and a forward P/S ratio of 15.6 times, both significantly higher than Micron's corresponding multiples (15 times and 10 times). However, the following factors greatly reduce these multiples: First, market consensus expects the company's revenue to jump from $19.6 billion in FY2026 to $43.5 billion in FY2027. This means the P/S ratio will compress from 15.8 times to 7 times in one year, and the forward P/E ratio in the second year will drop to 11 times - such rapid contraction of multiples is rare among AI infrastructure leaders. Second, Micron's latest financial report signals more robust than expected demand for NAND pricing, enterprise SSDs, and AI inference storage. Therefore, there is a significant possibility of SanDisk's performance being revised upwards. If management announces better-than-expected results and raises guidance in August, the current valuation may appear extremely cheap relative to long-term earnings potential. However, there are several risks to watch out for: execution risk - the lengthy qualification process for enterprise SSDs and delays in customer deployment may drag down growth; technological risk - technologies like high-bandwidth flash are still in the early stages of development and require ecosystem adaptability to generate revenue; high expectations risk - even if the stock price rises significantly, investors may be disappointed if performance does not exceed expectations. Zourmpanos believes that investors are still evaluating SanDisk with an old memory paradigm. The company has transformed into an AI infrastructure provider, revenue visibility continues to improve, the exposure to enterprise business is expanding, and it holds core technology aimed at the next wave of inference. Market sentiment will eventually follow the structural shift in fundamentals.