Micron plans $24 billion Singapore memory-chip expansion as AI demand tightens supply
Micron says it will invest about $24 billion over the next decade in a new advanced wafer fabrication facility in Singapore, aiming to boost NAND output as AI-related demand strains memory-chip supply.
A major bet on memory as AI infrastructure expands
Micron Technology announced plans to invest roughly $24 billion over the next decade to build an advanced memory-chip manufacturing facility in Singapore. The project centers on expanding wafer fabrication capacity to meet demand for NAND flash memory, which is increasingly tied to AI data centers, cloud infrastructure, and data-heavy computing.
The company’s statement frames the expansion as a response to an acute shortage across memory categories, as multiple industries race to build AI infrastructure. In this environment, supply constraints can ripple outward, affecting everything from consumer devices to enterprise server deployments.
Timeline and scope: wafer output targeted for 2028
Micron said wafer output from the new facility is expected to begin in the second half of 2028. The company described a large cleanroom footprint and positioned the investment as part of its long-term manufacturing strategy in Singapore, where Micron already has significant production operations.
The Singapore build also sits alongside efforts to strengthen high-bandwidth memory (HBM) supply chains, a critical component for advanced AI accelerators. While this announcement focuses on NAND, the broader message is that Micron intends to expand carefully while trying to avoid the historical boom-bust cycle that can occur when capacity ramps too quickly.
Why Singapore matters in the global chip map
Singapore has long served as a major semiconductor manufacturing hub, with deep logistics capabilities, engineering talent and established supplier networks. For Micron, building there leverages existing infrastructure and workforce, potentially shortening ramp time compared with a greenfield build in a new geography.
At the same time, the scale of investment underscores a strategic reality: memory is no longer a “commodity afterthought” in the AI era. Memory bandwidth, capacity and reliability increasingly determine how effectively AI systems can be trained and deployed at scale.
What to watch next
- Whether competitors match capacity plans, potentially easing shortages or raising oversupply risks by 2028–2029.
- How quickly Micron can secure tools and materials amid tight equipment supply chains.
- Whether pricing power for NAND and related memory categories persists through 2026–2027.