Micron plans $24 billion Singapore expansion as memory-chip shortage persists

Micron Technology is doubling down on Singapore as a core manufacturing hub, announcing plans to invest roughly $24 billion to build an advanced wafer fabrication facility over the next decade. The company said the project is designed to expand output for NAND flash memory, a component that sits at the heart of everything from consumer electronics to data-center storage—especially as AI workloads push infrastructure providers to buy more hardware, faster, and with less tolerance for supply interruptions.

Reuters reported that Micron expects wafer output from the new facility to begin in the second half of 2028. The company described the buildout as a way to meet “growing market demand” tied to data-centric applications and AI. The announcement lands in an environment where memory markets have swung between booms and busts for years, and Micron’s decision signals that, at least for now, management sees demand as durable enough to justify large-scale capacity additions.

Micron’s Singapore footprint is already significant, with Reuters noting the company makes the vast majority of its flash memory chips there. In parallel, Micron has been building an advanced packaging plant for high-bandwidth memory used in AI chips, highlighting how the company is trying to cover multiple bottlenecks in the AI supply chain: not just chip fabrication, but also the packaging and integration steps that can constrain shipments even when wafers are available.

The Straits Times added that the Singapore expansion is expected to increase cleanroom space and create new jobs, reflecting a broader competition among countries to attract semiconductor investment through stable infrastructure, skilled labor pipelines and government support. In the U.S., Micron has also outlined major domestic spending plans, but the Singapore project underscores how global the memory supply chain remains—and how companies are trying to diversify capacity to reduce geographic risk.

For the tech sector, the key takeaway is that the AI buildout is no longer an abstract story about model training and cloud services; it is driving tangible, long-dated industrial commitments. If memory shortages persist, they can raise costs for data-center operators and slow deployment timelines. If capacity comes online faster than demand, the cycle can flip into oversupply. Micron is effectively betting that the AI-era demand curve is steep enough to absorb the new output—while trying to expand carefully enough to avoid repeating the industry’s classic mistake of building too much, too soon.