SanDisk Details AI Storage Strategy, Signs $93.9 Billion Long-Term Agreement, HBF Samples to Debut in 2027

On August 14, SanDisk stated in its latest investor day materials that the construction of AI data centers is changing the demand structure in the flash memory industry. The company expects the Flash market size to reach approximately 1.2ZB by 2026, with data centers surpassing Edge as the largest source of demand; by 2027, the industry revenue TAM is projected to leap from a historical cycle midpoint of about $60 billion to over $300 billion, approaching $500 billion.

The core of this narrative lies in AI inference. SanDisk believes that long context, RAG, multimodal, and agent applications will drive the demand for KV Cache, with SSDs evolving from low-value storage components to strategic elements within AI infrastructure. The company even refers to SSDs as "Token Batteries," implying that by storing completed computational results, they can reduce redundant calculations, lower energy consumption, and enhance throughput.

In terms of business model, SanDisk emphasizes signing multi-year new commercial agreements with major clients to mitigate the traditional cyclical fluctuations of the NAND industry. The company claims that currently, eight clients are involved in these arrangements, with an average contract duration exceeding four years. By FY2027, approximately half of the bit supply will be incorporated into this model, rising to about two-thirds by FY2028. The total contract value calculated at floor price for these agreements is $93.9 billion, with remaining performance obligations of $91.1 billion and $16.5 billion in financial guarantees supporting it.

The financial targets are equally aggressive. SanDisk expects FY2026 annual revenue to reach $20 billion, a year-on-year increase of 175%; Non-GAAP gross margin is projected at 71.6%, with adjusted free cash flow of $8.7 billion. The long-term model provided for FY2028 to FY2030 indicates that revenue will maintain mid-to-high double-digit growth, with Non-GAAP gross margin around 80%, operating profit margin around 75%, and adjusted free cash flow rate around 50%.

The next further mainline is HBF high-bandwidth flash memory. SanDisk states that HBF will address the "memory wall" issue in large model inference, providing capacities up to 8 to 16 times that of HBM at similar read bandwidths. The company expects the first HBF inference product samples to appear in 2027, which will allow the market to continue reassessing its long-term space through changes in AI memory architecture.