HBM shortages drive up chip prices: Is the memory supercycle continuing?
Volatility in the US memory storage sector has intensified recently. Many investors jumped on the bandwagon after hearing about SanDisk's $94 billion long-term order news, but few truly understand the substantive value behind this headline.
The traditional NAND industry is trapped in a cyclical pattern: rising prices prompt manufacturers to expand capacity, leading to increased supply and a rapid drop in prices, repeatedly punishing investors. SanDisk aims to change this dynamic by disclosing that it has signed eight long-term supply agreements with AI clients.
Based on the floor price stipulated in the agreements, the minimum contract value is approximately $94 billion, with an average term exceeding four years and $16.5 billion in financial safeguards. The guaranteed gross margin under these agreements is approximately 80%.
⚠️ Key highlights to avoid common market misconceptions:
The $94 billion figure does not equate to recognized revenue, nor does the 80% gross margin reflect the company's current actual profitability.
This represents only the minimum baseline within the contract framework. Whether it translates into reported profits depends on actual customer uptake, product mix, production costs, and yield rates. Long-term contracts can mitigate risks associated with demand and price volatility but cannot entirely eliminate operational uncertainty.
Essentially, this signals a new industry trend: AI cloud providers are willing to pay a certainty premium for stable storage capacity.
SanDisk sacrifices some potential upside from future price increases in exchange for stable capacity utilization and cash flow; buyers secure supply to prevent shortages during surges in AI inference and data storage demand.
To verify whether these long-term contracts will deliver value, focus on three core metrics:
1. Order Coverage Ratio: The company expects long-term contracts to cover more than 50% of bit shipments in FY27 and nearly two-thirds in FY28. If customer uptake falls short of expectations, the substantial contract value will be significantly discounted, and the impact of spot prices on performance will not diminish quickly.
2. Costs and Product Technology: While an 80% guaranteed gross margin appears high, NAND operations still bear wafer costs, depreciation, and R&D expenses. This figure applies only to specific products covered under the agreement and should not be equated with the company's entire business.
Meanwhile, the company is planning its next-generation BiCS10 (332-layer) flash memory, which aims to increase wafer output by 65% and double bandwidth compared to the previous generation, with High Bandwidth Flash (HBF) products scheduled for launch in 2027. Successful mass production and yield rates are critical. While long-term contracts lock in commercial partnerships, they do not mitigate the risks associated with technological iteration. If the rollout of new technology is delayed, customers may well switch to other suppliers.
3. Cash Flow Changes: The RMB 16.5 billion in security deposits does not equate to immediate cash inflow. Future analysis should focus on changes in advance payments, operating cash flow, and accounts receivable reported in financial statements. Only sustained improvement in book cash balances can demonstrate that the contracts have genuine binding power.
The current market is clearly divided into two camps:
Bulls argue that long-term agreements give NAND stable returns akin to infrastructure assets for the first time, thereby dampening cyclicality;
Bears contend that as long as industry capacity continues to expand, cyclical fluctuations will not disappear entirely.
Objectively speaking, SanDisk will develop a dual business structure in the future: one segment with more stable order-based revenue, while the remaining standard NAND business will continue to experience significant volatility tied to spot prices.
This implies that our analytical framework for memory stocks needs updating: we cannot rely solely on spot quotes, inventory levels, and capital expenditure. We must also continuously monitor contract coverage ratios, customer concentration risks, and renegotiation clauses in long-term agreements.
There is also an easily overlooked risk: long-term contracts impose "bidirectional constraints." They protect SanDisk's profits during market downturns; however, if memory prices surge significantly in the future, the contracts will also limit the company's ability to capture the full upside. If cost reductions fail to keep pace with contracted pricing, the apparent high gross margins may prove difficult to realize.
Follow-up Watch List (Key verification points for the next 4-6 quarters)
1. Can the proportion of shipments under long-term contracts for NBM steadily increase?
2. Synchronous improvement in advance payments and operating cash flow.
3. Mass production yield and capacity release progress for BiCS10.
4. Customer concentration risk.
5. The gap between the company's overall gross margin and the guaranteed minimum gross margin stipulated in agreements.
In brief:
Do not simply interpret this as "storage will maintain high gross margins indefinitely." A more accurate assessment is that demand for AI storage is increasingly shifting toward long-term contracts, thereby redistributing cyclical volatility. While long-term contracts raise the floor for profits, upside potential remains constrained.
All paper contracts ultimately require validation through cash flow and financial statement data. Blindly going long or short based solely on news headlines makes it easy to fall into traps in the current volatile US memory stock market.
Customer concentration risk also warrants caution: a few large AI clients signing substantial long-term contracts enhances stability but also strengthens buyer bargaining power. The longer the contract term, the lower the tolerance for error in betting on technological pathways.
Even if long-term contracts are successfully executed, it does not mean the cycle has completely ended. Guaranteed minimum contracts cannot shield against risks such as a sharp decline in end-user demand, customers switching suppliers, or substitution by new technologies. Investors must still maintain a margin of safety.
Risk Disclaimer: The above content only represents the author's view. It does not represent any position or investment advice of Futu. Futu makes no representation or warranty.Read more
Comments
to post a comment
3