Chip wars: The uncertain impact of US sanctions on China's tech surge

The US continues to enforce stringent sanctions on China's tech industry. Central to these efforts is the imposition of US export restrictions designed to prevent China from acquiring advanced chips, particularly those that could bolster military capabilities in areas like artificial intelligence.

These restrictions have notably impacted China's access to cutting-edge graphics processors from leading companies such as Nvidia and AMD and have also hindered the import of high-end semiconductor equipment into China.

The effectiveness of these sanctions is receiving mixed assessments. While China appears to have made some progress in the chip sector, it still seems to be lagging in developing Large Language Models (LLMs).

The progress China has made in semiconductor manufacturing is demonstrated by Huawei's latest smartphone, featuring SMIC's 7-nanometer (nm) technology.

Although this technology lags behind the most advanced standard of 3 or 4 nm, it marks a substantial advancement in China's chip production capabilities.

However, the situation seems different regarding LLMs, where the progress is not as notable.

Lutfi Turkcan, a researcher at Koc University, points out that China may be struggling to keep pace with the most advanced technologies. "The fact that China has yet to provide a substantial response in the field of LLMs is telling," Turkcan tells TRT World.

"If you're unable to achieve this technological breakthrough, it suggests things are not going well."

US competition

The US publicly acknowledges its rivalry with China. In May 2022, US Secretary of State Antony Blinken succinctly described the Biden administration's approach towards China as "invest, align and compete".

Technology plays a crucial role in this competition. During a recent talk at the Brookings Institution in Washington, US Ambassador to China Nicholas Burns emphasised this point.

According to Burns, the competition spans beyond commercial interests, touching on strategic sectors like AI, machine learning, biotechnology, and quantum technology. Burns underscores that the US does not "plan to be the number two" in the technology "battle".

To achieve this, the US has been actively working to restrict China's technological advancement, particularly in sectors that overlap with military and national security.

This approach is reflected in a series of executive actions designed to limit American investments in Chinese technology companies.

In November 2020, former president Donald Trump issued a ban that prevented US investors from buying or investing in securities of companies associated with China's military-industrial complex, referring to these entities as "communist Chinese companies".

This ban focused on critical sectors such as aerospace, shipbuilding, construction, information technology, and communications.

President Joe Biden further intensified this policy with an executive order in June 2021. Last month, Biden chose to maintain this policy, citing concerns over the modernisation of "military, intelligence, and other security apparatuses" in China.

Gauging sanction efficacy

Semiconductors have become a focal point in the strategic competition between the US and China, serving as a litmus test for the effectiveness of US policy measures.

These components are not only central to the technology industry but are integral to a wide array of products ranging from everyday items like smartphones and toasters to more complex systems like cars and fighter jets.

Reflecting their ubiquitous role, the semiconductor industry boasts a global value exceeding $600 billion and is projected to surpass a trillion dollars by the end of the decade.

The Biden administration has employed export controls as a key strategy in this arena, primarily to prevent American-made semiconductors from being utilised in China's military apparatus.

Semiconductors, along with sensors, communications, and computing technologies, are vital components in modern defence systems.

Consequently, the US has imposed extensive restrictions designed to restrict China's access to vital semiconductor technologies.

US Commerce Secretary Gina Raimondo has been explicit in this regard, saying the US cannot allow the transfer of "the most sophisticated, highest-processing power AI chips, which would enable China to train their frontier models".

These chips could enable China to advance its AI technology significantly, potentially affecting the balance of power in both military and geopolitical arenas.

The ongoing technology restrictions have certainly impacted China's technological development, but it's important to note that they have not completely halted its advancement.

To circumvent these limitations, one key strategy China has adopted is the procurement of advanced chips initially designed for commercial purposes and subsequently repurposing them for military applications.

The country has also amassed a range of chip-making equipment, some of which were meant to be restricted by the US. This situation is highlighted by the current US criminal investigation into Applied Materials, a leading semiconductor equipment manufacturer.

The investigation is centred on allegations that the company may have illegally bypassed export controls to provide equipment to SMIC, China's foremost chipmaker, through South Korea without the necessary export licences.

Similarly, Lam Research, another company involved in manufacturing semiconductor equipment, continues to earn a significant portion of its revenue from China.

Recent reports suggest that almost half of Lam Research's total revenue is generated from the Chinese market, a notable increase from the previous year.

China is actively enhancing its semiconductor infrastructure to achieve self-reliance.

In 2021-22, China invested a substantial $290 billion in semiconductor-related projects. Most recently, Guangdong province in southern China introduced a new chip industry fund worth $1.5 billion. This initiative is in line with other similar projects across the country.

This effort is partially focused on the AI race, which is currently most prominently represented by LLMs.

LLMs gained worldwide attention with the success of OpenAI's product, ChatGPT. Recognised for their transformative potential across various sectors, LLMs are at the heart of future technological advancements.

The US has particularly focused its restrictions on the types of chips that power LLMs, considering the significant implications these might have for military use and national security.

China, acknowledging the crucial role of LLMs in its ambition to exceed the military capabilities of the US, has given high importance to this technology.

However, China faces a dual challenge in this area: the limitations of its infrastructure and the constraints imposed by its regulatory environment.

First, China is facing challenges in acquiring the necessary hardware for building LLMs. The US has enacted barriers to impede China's procurement of sophisticated GPUs.

Political Data Scientist Selim Yaman from the American University in Washington highlights the significance of GPUs for training cutting-edge AI projects. "GPUs, initially created for gaming, are perfectly suited for the complex mathematical tasks required in AI development," Yaman tells TRT World.

The additional challenge for China concerning LLMs lies in regulating the technology. Due to their black-box nature and limited clarity on the mechanisms of output generation, LLMs pose a threat to China's stringent speech policies.

Yet, imposing excessive controls on content and language can undermine the model's effectiveness. According to Yaman, "imposing constraints on the model leads to a decline in overall performance and user experience."

This leads to a tradeoff for the Chinese government, weighing its censorship priorities against the need for technological development.

In an effort to find a middle ground, China established a specialised group under The China Electronic Standardisation Institute.

The group, comprising leaders from Baidu, Huawei, 360 Security, and Alibaba, is responsible for formulating new standards for LLMs. Their objective is to minimise the impact of this technology on the Chinese regime's control measures, while harnessing its potential to foster innovation and upgrade traditional sectors.

Will sanctions work?

While the US is posing significant challenges for China through its sanctions, the approach of employing export controls and sanctions is not without drawbacks.

There's a danger that these actions might have unintended consequences. China may end up achieving self-sufficiency in the same technologies it is prohibited from importing.

Additionally, export restrictions can undermine the competitive edge of US domestic businesses, as they miss out on lucrative markets.

Turkcan states, "Even if sanctions create micro-benefits and seem beneficial in small areas, they generally backfire. Striving to inflict maximum damage on the adversary while keeping harm to oneself minimal may not work."