Global 300mm semiconductor fab equipment spending is projected to grow 18% to $133 billion in 2026, then rise about 14% to $151 billion in 2027, with continued gains to $155 billion in 2028 and $172 billion in 2029, led by AI-driven data-center demand.

SEMI notes 404 facilities and lines globally in the Fab Forecast, with 198 updates and nine new fabs/lines since December 2025, signaling ongoing capacity expansion.

The 300mm Fab Outlook reiterates the 404 facilities and lines, incorporating 198 updates and nine new fabs/lines since late 2025, as part of SEMI’s Fab Forecast collaboration.

The data comes from SEMI’s 300mm Fab Outlook and Fab Forecast datasets, with further details available through SEMI Market Intelligence and related press materials.

From 2027 to 2029, a new growth cycle in memory and continued advanced-node expansion are anticipated, aided by policy incentives and supply-chain localization.

Memory, led by DRAM and 3D NAND, is expected to rank second in spending, totaling about $175 billion from 2027 to 2029 as AI training and inference boost demand for HBM and storage capacity.

SEMI directs readers to its Market Intelligence team for data and subscriptions, and provides contact information for subscribers and reporters.

The forecast comes from SEMI’s Fab Forecast database and Fab Outlook, which track hundreds of facilities and production lines globally to inform market intelligence and strategy.

AI and advanced-node demand are resetting investment scale, with the Logic & Micro segment leading capacity expansion and driving sub-2nm technology investments from 2027 through 2029.

Regional investments from 2027 to 2029 will be broadly distributed: Taiwan leads in advanced-node manufacturing, Korea pushes memory capacity upgrades, China expands domestic capacity, and the Americas, Europe, the Middle East, and Southeast Asia pursue growth with policy incentives.

AI-driven demand supports memory investment, with increased HBM use for training and higher NAND storage for inference, sustaining investment across memory supply chains.