Huawei’s latest chip research has drawn attention because it suggests China may be exploring ways to improve computing performance despite restrictions on advanced semiconductor tools. Rishi Raj looks at the implications for the future direction of the global chip industry
What Huawei has announced
Huawei researchers have outlined a chip design framework that focuses on improving computing efficiency through architecture, packaging, and interconnection technologies rather than relying only on access to the world’s most advanced manufacturing nodes. The company has described concepts such as “LogicFolding” and “Tau scaling”, which seek to improve overall computing performance by redesigning how chips process and exchange data. The proposals have been discussed in academic papers and industry forums rather than through a commercial product launch. The announcement comes at a time when Huawei and other Chinese technology firms remain restricted from accessing advanced semiconductor tools and high-end AI chips because of US export controls.
Why it is being seen as significant?
The semiconductor Industry has largely advanced through transistor miniaturisation. Companies like TSMC, Samsung, and Intel improve performance by fitting more transistors into smaller spaces, from 7-nanometre to 5nm, 3nm, and now 2nm manufacturing processes. That depends heavily on extreme ultraviolet lithography machines or EUV tools made mainly by ASML. China hasn’t been able to freely access these due to export curbs by the US and its allies. Huawei attempts to work around it. Instead of depending on smaller transistors, it focuses on raising chip efficiency using stacking, interconnects and software-hardware optimisation.
What is chip stacking, why does it matter?
Chip stacking refers to placing semiconductor components together in layers or tightly connected packages instead of relying on one monolithic chip. It allows firms to improve computing performance by increasing memory bandwidth and reducing the distance data must travel inside a computing system. The approach isn’t new. Nvidia, AMD, and Apple increasingly rely on advanced packaging and chiplet designs to improve performance and energy efficiency. Huawei’s effort suggests China may try to compete more aggressively in these areas where manufacturing drawbacks can be partly offset through engineering and system design.
Has China caught up with the US?
Not yet. China still trails global leaders in advanced semicon manufacturing. TSMC and Samsung lead fabrication processes, while Nvidia dominates AI chips due to a combination of hardware capability, software ecosystems, and developer adoption. Chinese firms also continue to face hurdles in manufacturing yields, advanced memory access, and high-end lithography. Analysts note that architecture improvements can narrow performance gaps in some applications, particularly AI workloads, but they don’t eliminate the merits of access to frontier manufacturing technology.
Why is this critical for AI computing?
AI workloads differ from traditional computing tasks because they require large amounts of parallel processing and rapid data movement between processors and memory. As a result, performance increasingly depends not only on transistor density but also on memory bandwidth, networking, and software optimisation. This is one reason why Nvidia’s dominance extends beyond chips alone. Huawei appears to be attempting a similar strategy by building a broader computing architecture around its Ascend AI chips.
Could this weaken the impact of US sanctions?
Potentially, but only partly. The US strategy has largely been built around limiting China’s access to advanced manufacturing equipment and high-end AI chips. If Chinese firms can improve performance through packaging, interconnects, and system architecture, some of those restrictions may become less effective over time. However, sanctions still create major constraints. China remains dependent on imported tools, specialised semiconductor materials, and several parts of the global chip supply chain. Analysts therefore view Huawei’s work less as a technological breakthrough that changes the industry immediately and more as an attempt to build an alternative path for semiconductor progress under restrictions.
What does this mean for the global chip industry?
The development reflects a broader shift underway in semiconductors. For decades, progress was measured mainly through transistor shrinkage and Moore’s Law. Increasingly, firms are also focusing on packaging, specialised AI accelerators, and system-level optimisation to improve computing performance. Huawei’s announcement suggests geopolitical restrictions may accelerate this shift by forcing companies to find alternative methods of improving chip capability. The broader question is no longer only whether China can match the most advanced manufacturing processes, but whether alternative approaches to chip design can reduce the effectiveness of technology restrictions over time.