Micron plans a $24 billion Singapore fab as AI demand tightens memory supply

A major new fab commitment in Singapore

Micron Technology said on January 27, 2026 that it plans to build a new chip manufacturing plant in Singapore with investment totaling roughly $24 billion over the next decade. The project centers on an advanced wafer fabrication facility designed to expand the company’s capacity for NAND memory chips—components that have become increasingly strategic as artificial intelligence workloads drive data-center buildouts.

Micron’s announcement arrives in a climate where memory chips have moved from a cyclical commodity story to a core AI infrastructure constraint. As more computing shifts toward AI training and inference, the memory requirements—capacity, bandwidth, and power efficiency—rise sharply, putting pressure on suppliers and pushing them toward large, long-horizon capital spending decisions.

Timeline and scale: production targeted for 2028

According to reporting, wafer output from the new Singapore facility is expected to start in the second half of 2028. The timeline illustrates how difficult it is to add supply quickly in semiconductors: planning, permitting, construction, tool installation, and qualification can stretch for years, particularly for advanced processes and the cleanroom capacity they require.

The investment also reinforces Singapore’s role as a major semiconductor manufacturing hub, with Micron already producing a large share of its flash memory there. For the company, the expansion is positioned as a way to meet rising demand while trying to avoid the industry’s recurring pitfall: building too much capacity and triggering an oversupply-driven price collapse.

Why investors care: memory is now an AI bottleneck

The market focus on Micron is about more than one new plant. The broader story is that AI hardware demand is spilling beyond GPUs into the memory stack—high-bandwidth memory, DRAM, and NAND—tightening supply chains across the board. When memory is scarce, it can limit how quickly cloud providers and enterprises can deploy AI systems, which then feeds back into chipmakers’ pricing power.

If the shortage persists, large-scale expansions like this could support multi-year revenue visibility and a higher valuation framework than memory makers historically enjoyed. But if demand assumptions cool, these same investments can become a risk, making Micron’s execution discipline and pacing decisions as important as the headline number.