Micron Technology outlined a $45 billion capital expenditure plan for fiscal 2027 on Thursday, underscoring the strategic shift in memory as artificial intelligence reshapes demand across the semiconductor industry.
Executive Sumit Sadana told the Six Five Summit that the company’s fiscal 2026 spending will reach about $26 billion, more than double the $13 billion deployed in fiscal 2025, while its U.S. investment commitment has been raised from $200 billion to $250 billion with an accelerated deployment schedule. Revenue growth over the last twelve months reached 167%, and the company’s gross margin stands at 73%, according to the presentation.
Sadana described a persistent shortage of memory across all market segments, noting that Micron does not yet have visibility into when supply will align with demand. He emphasized that AI system performance now depends “to a first order” on memory subsystem performance and capacity, shifting the role of memory from a supporting component to a primary differentiator in customer roadmaps.
The company detailed a global expansion pipeline of 20 projects, including new fabrication facilities and expansions. In Idaho, the first wafer outs from the Idaho One Fab are scheduled for mid-2026, while Idaho Two Fab is projected to begin production near the end of 2028. The Tongluo Fab in Taiwan is expected to start wafer starts in 2027, with additional expansions planned. In New York, ground was broken in January for a four-fab cluster, with first output targeted for 2030. Work is also underway in Japan and Singapore.
Micron’s manufacturing process involves approximately 2,000 steps per wafer, with a total cycle time of about five months from fab start to customer shipment. High Bandwidth Memory stacks can reach 12 die high, and single SSDs can store up to 245 terabytes. The company’s stock has gained 698% over the past year and 229% year-to-date, with a P/E ratio of 20.7 and a return on equity of 67%.
Sadana said customers are embedding Micron’s memory into multi-year roadmaps, focusing on differentiating through memory performance and capacity to gain competitive advantage in the AI era.
