AI chip race lifts semiconductor equipment sales to record US$36.55bn

The surge in equipment spending is a structural shift from 'algorithm-first' to 'hardware-first' AI infrastructure. Upstream EDA and materials suppliers are racing to support sub-3nm logic and CoWoS-type advanced packaging, while cloud providers lock in tool capacity to secure delivery timelines. Tightening U.S. export controls have forced global vendors to reconfigure production footprints across Taiwan, China; South Korea; and mainland China, adding 5–8% to operational costs. ASML, Lam Research, and Tokyo Electron will counter geopolitical fragmentation with modular tool architectures, while Chinese players like AMEC and NAURA gain ground in mature-node equipment. Over the next 18 months, record orders will mask looming overcapacity—any delay in HBM4 adoption or AI server procurement could trigger an inventory correction by Q2 2027.