SEMI Forecasts Record Chip Equipment Sales in 2028

SEMI forecasts that worldwide semiconductor manufacturing equipment sales will reach $165.9 billion in 2026, up 23.2% from 2025, and continue rising to a record $229.5 billion in 2028. The industry group published the mid-year forecast on July 14 and attributed the stronger trajectory to investment in AI infrastructure, leading-edge logic, advanced memory, testing and packaging.

The forecast is evidence that the AI buildout is reaching far beyond accelerator designers and cloud operators. It is pulling capital into the tools used to make, assemble and test the chips behind those systems. Still, these are projected equipment sales, not a direct measure of finished-chip output or deployable compute capacity.

Wafer-fab equipment accounted for $116.9 billion in sales in 2025. SEMI expects that segment to grow 23.1% to $143.9 billion in 2026, then reach about $200 billion in 2028. It links the revised outlook to leading-edge logic and advanced-memory spending, including equipment for high-bandwidth-memory-related DRAM.

The application-level forecast shows how concentrated the expansion is. DRAM equipment sales are projected to rise 39.0% to $38.8 billion in 2026 and reach $56.9 billion in 2028. Semiconductor test equipment is forecast to grow 31.0% to $15.3 billion in 2026, after a 55.3% increase in 2025. Assembly and packaging equipment is projected at $6.7 billion in 2026 and $8.6 billion in 2028.

Korea Times reporting based on the SEMI forecast also highlights China, Taiwan and Korea as the three largest equipment-investment markets through 2028. Hyperframe Research interprets the mix as a concentrated cycle led by advanced logic, HBM DRAM, testing and packaging rather than a broad expansion across older semiconductor capacity.

For AI infrastructure teams, the numbers point to investment in several known supply-chain constraints at once: leading-edge wafers, high-bandwidth memory, advanced packaging and more demanding chip testing. That matters because a shortage in any one of those stages can limit how quickly complete accelerator systems are delivered.

The timing remains important. Equipment must be installed and qualified before wafers can be produced, packaged, tested and integrated into systems. SEMI's forecast therefore supports a multi-year capacity expansion story, not a promise that cloud or enterprise buyers will see an equivalent near-term increase in available AI compute.