Taiwan Semiconductor Manufacturing Co has projected that global semiconductor revenue will surge past US$1 trillion this year and reach about US$1.5 trillion by 2030, driven largely by accelerating artificial intelligence demand and the rising need for advanced compute power across data centers and digital infrastructure.
Speaking at the company’s annual technology symposium in Hsinchu City, TSMC deputy co-chief operating officer Kevin Zhang said the industry is still in the early stages of an AI-driven transformation. He noted that while smartphones defined semiconductor growth over the past decade, AI applications are now becoming the dominant force shaping long-term demand across the global chip ecosystem. Zhang added that AI and high-performance computing are expected to account for roughly 55 percent of total semiconductor revenue by 2030, marking a structural shift away from traditional consumer electronics-led demand.
TSMC data presented at the event showed that AI inference workloads are set to overtake training workloads as the primary driver of chip demand over the coming years. The company also estimated that its ecosystem contributions helped generate more than US$350 billion in semiconductor revenue last year alone, with expectations that this figure could exceed US$1 trillion within four years as AI adoption accelerates across industries.
To meet this surge in demand, TSMC is rapidly expanding its manufacturing and advanced packaging capacity. The company is scaling up production of 2-nanometer chips across multiple facilities in Hsinchu City and Kaohsiung, targeting steep growth in output over the coming years. It is also aggressively increasing production of its advanced chip-on-wafer-on-substrate packaging technology, known as CoWoS, which is essential for integrating high-bandwidth memory used in AI accelerators.
Company executives said CoWoS capacity has been expanding at an annual rate approaching 90 percent to meet surging demand from AI chip designers. The latest generation of the technology has entered mass production, enabling significantly larger chip integration and higher memory density, while future versions are expected to scale even further, supporting increasingly complex AI workloads. TSMC also highlighted its emerging COUPE photonics technology, designed to improve chip-to-chip communication efficiency and reduce energy loss in data center systems.
Together, these advancements underscore how AI is reshaping the semiconductor industry’s trajectory, with TSMC positioning itself at the center of a global transition toward high-performance, energy-efficient computing architectures that are expected to define the next decade of technological growth.
