🚨Tau Scaling : Moore's Law is officially on life support. #Huawei #ai #artificialintelligence #china

追加: 2026-05-27

[00:00:00]Moore's Law is officially on life support, and Huawei just proposed its replacement. For 50 years, chips got faster by shrinking transistors, the tiny switches inside every processor, but we're now at 2 nanometers, the width of approximately 10 silicon atoms. Go smaller and quantum physics breaks everything.

[00:00:21]So, Huawei asked a different question.

[00:00:24]Instead of making transistors smaller, what if we made signals travel faster?

[00:00:29]That's the Tao Tao scaling law, unveiled May 25th at an IEEE symposium in Shanghai.

[00:00:34]Tao Tao means time in physics. The idea, stop obsessing over transistor size and start obsessing over how fast electrical signals move through circuits, chips, and entire computing systems. Think of it like city traffic. Old approach, build more lanes. Tao scaling, redesign the road so cars move faster on existing infrastructure. Huawei also introduced logic folding, restructuring chip architecture so signals travel shorter distances. A chip can perform like a 1.4 nanometers chip without being physically built at 1.4 nanometers. Why it matters for AI. Modern AI isn't just slowed by weak chips, it's slowed by how long data takes to move between chips. Tao scaling targets exactly that bottleneck. Huawei has already shipped 381 [snorts] chips built on this framework. Their goal, match leading-edge chip performance within 5 years despite US sanctions cutting off TSMC access. Bold claim, early days, but the era of just shrink it may be over.