Despite restrictions around access to advanced technology, Huawei’s new smartphone – powered by SMIC chips – sees it back at the fore of 5G enabled devices. We explore the implications of China’s efforts to secure semiconductor independence.
- In early September, the Huawei Mate 60 Pro emerged as the most controversial smartphone worldwide.
- The timing, coming ahead of the launch of Apple iPhone 15, was a moment of techno-nationalist pride in China. Despite punitive US sanctions to curtail Huawei’s access to advanced technology since 2019, Huawei seems to have engineered its way back to the forefront of 5G smartphones.
- Indeed, the geopolitics of semiconductor remain alive, with knock-on impact on other Chinese technology and internet platform companies.
A surprise to the financial market, less so for industry players
US sanctions to restrict Huawei’s access to advanced technology have been in place since 2019. The latest Mate 60 Pro smartphone highlights Huawei’s comeback in producing 5G-capable devices. The domestic foundry company, Semiconductor Manufacturing International Corp (SMIC) provides the critical helping hand. As revealed by TechInsights, the smartphone Huawei Mate 60 Pro features the Kirin 9000s processor that uses the SMIC’s 7nm (N+2) foundry process. While Huawei’s smartphone advancement surprised the financial market, this technological milestone was anticipated by industry players in the semiconductor space. In 2022, TechInsights reported that SMIC had used the 7nm (N+1) to produce the MINERVA7 Bitcoin Miner ASIC, heralding the pathway for the SMIC-Huawei partnership to produce advanced smartphones. While SMIC is restricted from obtaining advanced lithography machines (especially EUV) from ASML, it has been using the DUVs (deep-ultraviolet lithography) with multi-patterning to produce the sub-10nm chipsets.
The risk of more export controls into China could impact other Chinese corporates, beyond SMIC and Huawei
As background, the level of multi-lateral cooperation between US, Netherlands and Japan has already increased in 2023. Netherlands has imposed additional new export controls on its advanced immersion DUV (NXT: 2000i and subsequent immersion systems), effective 1 September 2023. Meanwhile, Japan implemented the export controls on 23 types of equipment in June 2023, which comprise advanced manufacturing equipment, photomask coating, etching, deposition etc. As it stands, China does not yet have a homegrown company to rival ASML at the leading edge of lithography. In terms of domestic capability, according to Global Times (Aug 2023), Shanghai Micro Electronics Equipment (SMEE) will launch its 28-nm lithography machine by end-2023. The scope for more export restrictions on equipment and input materials to China could likely increase. Certain US legislators are calling for a complete ban on technological exports to both Huawei and SMIC. A broadening of export restrictions, if any, could disrupt a swathe of companies which has been investing in their chip design capabilities.
Embracing the trend towards more homegrown semiconductor capability but tighter export controls by the US could add more speed bumps
Major Chinese companies (Alibaba, Tencent, Baidu) have been investing in self-designed chips to support their cloud computing and artificial intelligence strategy. Much of the fabrication is outsourced to non-domestic foundries such as TSMC and Samsung. Baidu, for example, holds a c.71% stake in Kunlun Chip Technology Company, which designs the Kunlun chips for search engine, cloud and deep learning computing. For now, these Chinese companies continue to depend on Nvidia’s GPUs for much of their AI training needs. In fact, Baidu, Bytedance, Tencent and Alibaba were reportedly rushing to purchase the Nvidia 800 chips for around USD5bn, on concerns that the US may impose more export controls (FT, Aug 2023). Baidu itself has previously mentioned that its own Kunlun chips are more suitable for AI inference, while it uses the Nvidia GPUs for AI training due to higher computational demands. Indeed, in mid-October, the US Department of Commerce announced additional export controls with a broader scope of AI semiconductor chips that falls under the restriction list. The latest list includes the Nvidia A800 and H800 chips. Furthermore, the licensing requirements are expanded to a number of countries and foreign subsidiaries, in order to stem the circumvention of export rules by Chinese buyers.
The scope and timing of further US response to the Huawei-SMIC development will be keenly watched over the near-term given some diplomatic sensitivity ahead of the upcoming Asia-Pacific Economic Cooperation (APEC) summit in November 2023. A potential face-to-face meeting between US President Biden and China President Xi Jinping could take place during the summit. That said, diplomatic sensitivity alone does not imply a pause in retaliatory actions. For context, in October 2022, the US implemented export control controls on advanced computing and semiconductor manufacturing items, before the President Biden-Xi meeting during the G20 summit (November 2022, Bali).
As our equity colleague (Dave Egan) highlights, the potential addressable market for AI should be larger than the traditional tech industry and we are just scratching the surface. The SMIC-Huawei development underscores that China is serious about building up its semiconductor independence, guided by the aspiration to be a global leader in AI. An escalation of export controls and restrictions will leave more companies, beyond SMIC and Huawei, bruised. Most might find a way to innovate and survive. But the road ahead is bumpy.
The geopolitics of semiconductor
