How can China's mature 14nm chips approach Nvidia's performance?

By integrating them with high-performance memory and new computing architectures like software-defined near-memory computing.

What is 'near-memory computing' in China's chip strategy?

It's a co-design approach placing processors close to memory to reduce data movement and boost efficiency.

How do US export controls affect China's chip development?

They limit access to leading-edge chip manufacturing capacity, pushing China towards innovative packaging and architecture solutions.

Home / Technology / China's Chip Leap: Mature Tech Meets New AI

China's Chip Leap: Mature Tech Meets New AI

3 Dec

•

Summary

China aims to match Nvidia performance using mature 14nm chips.
Integration with advanced memory and new architectures is key.
This strategy seeks to bypass US export controls and cut costs.

China's Chip Leap: Mature Tech Meets New AI

China is significantly accelerating its drive for chip self-sufficiency by integrating mature technologies with novel computing architectures. This strategic move aims to bridge the performance gap with industry leaders like Nvidia, according to prominent industry expert Wei Shaojun. By combining 14-nanometre logic chips with advanced memory and innovative designs, China seeks to challenge the dominance of cutting-edge AI processors.

Wei Shaojun, vice-president of the China Semiconductor Industry Association, detailed a solution involving "software-defined near-memory computing." This approach utilizes 3D hybrid bonding to stack 14nm logic chips alongside 18nm DRAM. Such a configuration could potentially rival Nvidia's 4nm GPUs, the current standard for AI training, while also offering substantial savings in cost and energy usage.

This technological direction aligns with Huawei's strategy of "stacking and clustering" chips to overcome limitations imposed by US export controls. These controls restrict China's access to advanced chip manufacturing. Wei's proposal underscores a commitment to a "fully home-grown supply chain," vital for securing domestic AI computing power amid ongoing technological competition.