SanDisk shares surged 315.3% as AI infrastructure demand reshaped the storage hardware market. Western Digital, SanDisk's parent company, climbed 166.1% on similar tailwinds.
Western Digital shipped over 2.2 million ePMR (energy-assisted perpendicular magnetic recording) drives, reflecting enterprise appetite for high-capacity storage. Data center operators need expanded storage to support AI training and inference workloads.
Micron Technology locked contracts for 100% of its calendar 2026 HBM (high-bandwidth memory) production. HBM chips serve AI accelerators from Nvidia and AMD. The sellout signals tight supply across the memory stack.
Two major players announced $400 billion in combined capital expenditures for semiconductor fabrication. Micron separately committed $20 billion for fiscal 2026, including new data center megafabs. These investments target AI-driven demand for DRAM and NAND flash.
The rally stems from a direct link between AI compute expansion and storage requirements. Each AI cluster needs terabytes of high-speed memory for model parameters and petabytes of storage for training datasets. As hyperscalers add GPU capacity, they must scale storage infrastructure proportionally.
Analyst projections show AI infrastructure capex reaching multi-hundred-billion levels through 2027. Storage and memory suppliers benefit from this spending wave. Stock performance reflects investor confidence that demand will outlast current AI deployment cycles.
HBM supply constraints remain a limiting factor for AI accelerator production. Micron's 2026 sellout indicates memory will stay tight even as fabrication capacity expands. Western Digital's ePMR volumes show traditional hard drives remain relevant for cold storage tiers in AI data lakes.
The semiconductor storage sector now trades at elevated multiples. Valuations price in sustained capex from cloud providers, AI labs, and enterprises building private inference infrastructure. Any slowdown in AI deployment timelines could pressure these gains.
Micron's $20 billion investment spans multiple technology nodes. The company targets both cutting-edge HBM for AI accelerators and mainstream NAND for broader data center needs. This dual approach hedges against concentration risk in AI-specific products.