TSMC says energy efficiency now drives AI chip design

TSMC Says Energy Use Is Forcing a Rethink of AI Chip Design

Surging electricity demands from artificial intelligence are making energy efficiency — rather than raw computing power — the primary constraint shaping future chip development, a senior executive at TSMC said on Thursday.

Efficiency Over Performance

Kevin Zhang, Senior Vice President of Business Development at the world’s largest contract chipmaker, told reporters at a conference in Amsterdam that customers ranging from smartphone makers to AI data center operators are increasingly prioritizing performance gains that do not drive up power consumption.wtvbam

“The area customers most want improvement in is energy efficiency. This is true across the board, whether you are the edge guy, smartphone, mobile, IoT application, or high-performance AI data center,” Zhang said.wtvbam

TSMC manufactures AI chips for Nvidia and AMD, as well as custom AI processors for major cloud companies including Google, Amazon, Meta, and Microsoft.reuters

Beyond Transistor Density

The shift marks a broader turning point for the semiconductor industry, where simply packing more transistors onto chips is no longer sufficient to sustain performance gains for energy-hungry AI workloads, according to Reuters. Zhang said transistor density improvements remain central to TSMC’s roadmap, but other approaches — including advanced packaging, chip stacking, and photonics — are becoming increasingly important to boost efficiency.wtvbam

He said TSMC expects its chips to cut power consumption by up to 30% between its current N2 technology and its A14 generation, due around 2028, while delivering more than 20% higher computing performance.wtvbam

Competitive Landscape

The comments come as rivals explore alternative paths to chip improvement. Chinese competitor Huawei unveiled its “Tau Scaling Law” plan this week to improve performance by speeding up data movement within chips — an approach Zhang described as largely dependent on integrating components more closely through techniques like 3D stacking.wtvbam

TSMC said in April it would delay adoption of ASML’s next-generation extreme ultraviolet lithography technology for several years, underscoring how design features improving energy efficiency are becoming more urgent than smaller circuitry for its coming generation of AI chips.wtvbam