The protocol doesn't care about your revenue guidance unless the code behind it is auditable. Yet the market is treating SanDisk's 2028-2030 CAGR of 15-20% as a bullish signal. Let me dissect why this is a mistake.
Context
SanDisk (controlled by Western Digital) and its joint venture partner Kioxia announced a long-term pricing agreement (LTPA) with hyperscalers, covering enterprise-grade NAND for AI data centers. The market reaction was immediate: share prices jumped, analysts raised targets, and the narrative shifted from "NAND is a commodity" to "NAND has recurring revenue."
But here's the problem: the LTPA is a financial derivative, not a technical guarantee. The underlying product—3D NAND—is still a physical chip with a known failure mode: bit rot. No amount of contractual phrasing changes the physics of floating-gate transistors.
Core: The Structural Flaw in the LTPA Model
Based on my audit experience, long-term agreements in semiconductor supply chains are risk-shifting mechanisms, not risk-reducing ones. The buyer (hyperscaler) locks in a price floor, while the seller (SanDisk) locks in a volume commitment. The buyer gets price certainty; the seller gets revenue visibility. But the seller's risk—capital expenditure recovery—is not eliminated; it's just deferred.
Here's the math. SanDisk's 2028-2030 revenue guidance implies a cumulative bit growth of roughly 60-80% over current levels. To achieve that, they need to ramp up 3D NAND stacking from 218 layers to 300+ layers within two years. The industry average for a new node introduction is 18-24 months. BiCS 8 (218 layers) was introduced in 2023. BiCS 9 (300+ layers) is expected in 2025. The timeline is tight, and the yield curve for 300-layer NAND is unknown.
Trust is a variable we must eliminate, not manage. The LTPA does not eliminate the trust that the product will meet spec. It only manages the price. If SanDisk ships a defective product, the hyperscaler will still pay, but they will also sue. The contract is a legal instrument, not a technical one.
The Contrarian Angle: What the Bulls Got Right
To be fair, the LTPA does improve the revenue composition. If 30% of SanDisk's revenue becomes recurring, the company's valuation multiple should expand. The market is pricing in a transition from a cyclical commodity play to a semi-annuity model. This is logically consistent.
But the blind spot is the assumption that the hyperscalers will not find alternative sources. SK Hynix is already supplying 238-layer NAND to the same hyperscalers. Samsung is at 300+ layers. Both have better bargaining power because they have DRAM and HBM cross-subsidies. SanDisk, with only NAND, is a single-product company. The LTPA locks them into a narrow margin band. If the bit price declines faster than expected, the LTPA becomes a liability.
Hype is just volatility wearing a suit and tie. The market is ignoring the fact that the LTPA is a two-way street. The hyperscaler can demand volume flexibility, but SanDisk's capital expenditure is fixed. Any demand shortfall will be absorbed by SanDisk, not the buyer.
Takeaway
Risk is not a number, it's a structural flaw. The LTPA makes SanDisk's revenue more predictable, but it does not make the company's technology less risky. The real question is: can SanDisk execute on 300-layer NAND with a yield that supports a 15-20% CAGR? I have seen no evidence of that. The market is betting on a contract, not a chip. I am betting on the chip.