The world’s largest contract chipmaker sees efficiency, not just transistor scaling, driving the next phase of semiconductor growth.
TSMC has said that the rapid growth of artificial intelligence is making energy efficiency a bigger priority than raw computing power in the development of future semiconductors.
Speaking at a conference in Amsterdam, Kevin Zhang, Senior Vice President of Business Development at TSMC, said customers across smartphones, IoT devices, edge computing, and AI data centres are increasingly demanding chips that can deliver higher performance without significantly increasing electricity consumption.
The comments reflect a major shift in the semiconductor industry as AI workloads place growing pressure on global power resources. Data centres running advanced AI models require huge amounts of electricity, forcing chipmakers to focus more on reducing energy usage while maintaining processing performance.
TSMC, the world’s largest contract chipmaker, supplies processors for major technology companies including Nvidia, AMD, Google, Amazon, Meta, and Microsoft. Zhang said that while increasing transistor density remains important, technologies such as advanced packaging, 3D chip stacking, and photonics are becoming critical to improving chip efficiency.
The company expects its future A14 process technology, planned around 2028, to cut power consumption by up to 30% compared with its current N2 generation, while also delivering more than 20% higher computing performance.
The remarks come as global semiconductor companies search for new methods to sustain performance gains beyond traditional chip scaling. Chinese technology company Huawei recently introduced its “Tau Scaling Law” strategy aimed at improving processor performance through faster internal data movement and tighter component integration.
Zhang said similar concepts have existed in the industry for years and are largely based on advanced chip integration technologies like 3D stacking.
Huawei’s efforts also highlight the impact of U.S.-led export restrictions that limit Chinese companies’ access to advanced EUV lithography equipment from Dutch supplier ASML. TSMC itself recently delayed adoption of next-generation EUV tools, underscoring the industry’s increasing focus on energy-efficient chip design for AI applications.
