Chinese LLMs + Weight-loss drugs + Chinese telcos build AI models

Science and technology as the linchpin of Xi Jinping’s vision for China featured squarely in July at the Third Plenum of the Central Committee of the Chinese Communist Party (CCP), a major meeting on long-term social and economic plans. The president and CCP leader emphasized the need to boost “new-quality productive forces” – meaning growth led by technology innovation – to strengthen China’s geopolitical clout. While self-reliance and self-empowerment did not feature as prominently in the post-Plenum document, the call to “accelerate the build-out of independent and controllable industrial supply chains” suggests that self-reliance remains a principal objective. 

Securing important supply chains was also a key topic at the meeting, and the Plenum document calls for security assessments in that regard. The widespread, worldwide computer outages after an update by security software provider Crowdstrike in July further reinforced Beijing’s determination to make supply chains “secure and controllable.” 

In line with the push from the top leadership, the home-grown operating system of telecoms behemoth Huawei, HarmonyOS, now uses its own Kernel, cutting ties with the Android operating system, which Harmony had previously used. Software developed for Android will no longer automatically work on Huawei’s OS.

Antonia Hmaidi, project lead for the CTO, says: “China is preparing for a fortress economy, with export-driven growth fueled by high-tech. Science and technology are especially important for achieving self-reliance, now a mainstay and central pillar of China’s efforts across the board.“

China’s large language models (LLMs) for generative artificial intelligence were off to a late start catching up to US-based Open AI’s ChatGPT, due to heavy state-imposed regulatory requirements. The first batch of eight approved LLMs were only released to the public in August 2023, led by Baidu’s Ernie Bot.

Since then, however, China’s generative AI landscape has become crowded with new entrants. The list of approved LLMs has grown to 117. IT giant Alibaba alone has invested billions of dollars in at least five of the leading generative AI startups, in addition to developing its own open-source LLM, Qwen. Tencent, ByteDance and Huawei are among other large tech firms working on their own generative AI offerings. 

Despite strict regulations on training data and output, many Chinese LLMs are competitive in performance on global leaderboards. They perform especially well in Chinese-language tasks. 

But great power competition with the US may yet throw a wrench in China’s progress. Many Chinese models rely on open-source development frameworks as the basis for their design, especially on Meta’s Llama series. The US Commerce Department is mulling restrictions on open-source models, although this may be difficult to implement. Chinese companies would find it much harder to develop LLMs using only Chinese technology at this point. Export restrictions from the US on advanced semiconductors will also continue to negatively affect innovation at the highest level. 

However, geopolitics can be a boon to Chinese platforms: OpenAI has blocked programmatic access to their services by Chinese users, and other US companies might follow with similar steps in response to security concerns in Washington. Some Chinese companies have since rushed to offer discounts to developers to switch to their services. Restrictions on tech exports will contribute to a growing split in the tech ecosystem inside and outside China.

MERICS Analyst Wendy Chang: “China was off to a slow start in building large language models but has made much progress since. What happens next is as much about tech as about geopolitics: the US could hobble Chinese developments if it manages to cut off hardware and software access. China’s own progress in building AI chips and indigenizing software architecture will also be key to its staying competitive in the AI race.”

Artificial Intelligence – notable developments:

Tech progress

• The 2024 World AI Conference in July in Shanghai focused on robotics. A humanoid general-purpose robot, Qinglong, was unveiled by Humanoid Robot, a lab and state-designated innovation center. Tesla also debuted its latest humanoid robot. (Source (EN): Tech Node), Source (CN): Guangming Daily) 

Domestic dynamics

• US firm OpenAI has ended services in China and cut off access to its generative AI models, furthering a growing split between Chinese and Western AI ecosystems. (Source (EN): Time)
• The Ministry of Industry and Information Technology (MIIT) released a draft AI Industry Standardization Framework Guidance, setting goals for industry standards in six key areas. This is in line with other standardization framework guidance in China. (Source (CN): S&T Daily) 

Foreign involvement

• US company Nvidia, the world's leading producer of AI chips, is reportedly working on a less powerful version of its new Blackwell chip in order to stay in the Chinese market despite US export restrictions. (Source (CN): EE Times China) 
• The first US-China AI dialogue on risks took place in May behind closed doors, focusing on AI risks and the importance of governance measures. China reiterated its complaint against US export controls hampering its AI development. (Source (EN): AP News)
• Meta’s latest open-source LLM, Llama 3.1, performs at the level of the best proprietary models. This upends previous assumptions about the strength of commercial vs freely available LLMs and will likely quicken Washington’s efforts to restrict what foreign adversaries can access. (Source (EN): The Verge)

The weight-loss drug market is a golden opportunity for China to become a dominant player in global pharma. Aided by state support and domestic procurement, Chinese weight-loss products will first take over the domestic market (estimated at USD 10 billion by 2033), so that by the 2030s, they will compete in a global market expected to exceed USD 100 billion. 

China’s innovation and industrial policy is designed for this. Now that the key innovation in weight-loss drugs has been made (targeting the GLP-1 receptor), China can concentrate on incremental innovation and economies of scale. This puts weight-loss drugs on a similar path to solar panels, digital hardware and electric vehicles. With global demand expected to increase by 20 percent annually over the next decade, weight-loss drugs are a growth market that foreign suppliers are already struggling to supply.

By 2026, Novo Nordisk’s patent in China for semaglutide, the diabetic drug also used for weight loss, will expire. Typically, this would prompt pharma companies to launch new products that are slightly different but significantly better. Chinese companies have already joined in this competition. At least 11 Chinese firms are in the final stages of clinical trials with generics and biosimilars. Of these Chinese firms, Hangzhou Jiuyuan already applied for market approval in April. Suzhou-based firm Innovent announced in July that the next-generation drug mazdutide, which it licensed from US drug company Eli Lily, had cleared the third of five planned Phase III studies.

Biotech might differ from hardware sectors. Clinical trials require more patience and deeper pockets. However, Novo Nordisk, Europe’s most valuable firm, should stop underestimating China. 

MERICS Head of Program Jeroen Groenewegen-Lau says: “Chinese firms will become major players in the global weight-loss drug market by 2030, which will be a harbinger of China’s rising stock in global pharma more generally.”

Biotechnology – notable developments:

Tech progress 

• Chinese scientists have developed the first open-source brain-computer interface chip. The MetaBOC system uses stem cells to cultivate an artificial brain in vitro, which is then linked to an external chip. (Source (CN and EN): S&T Daily, Yicai)

Domestic dynamics

• The Ministry of Science and Technology published ethics guidelines for research on human genome editing. They prohibit research on germline genome editing – i.e. when the genome of an individual is edited to be heritable. (Source (CN): MOST)
• The construction of China's first national-level bio-manufacturing innovation center began in Shenzhen. The center is led by the Shenzhen Institute of Advanced Technology (SIAT) of the Chinese Academy of Sciences and will feature six major technical platforms. (Source (CN): S&T Daily)
• The National Medical Products Administration released the "Management Requirements for the Temporary Import and Use of Urgently Needed Medical Devices by Medical Institutions," aiming to enhance the availability of essential medical devices not yet approved domestically. (Source (CN): NMPA)

Foreign involvement

• The US House Committee on Oversight and Accountability wants to restrict business with Chinese biotech firms like BGI and WuXi AppTec. The Biosecure Act, which still requires approval from Congress, seeks to reduce US reliance on Chinese research and manufacturing. (Source (EN): Reuters)
• Western pharmaceutical companies are increasing licensing deals with Chinese biotech firms as patents expire. Chinese drugmakers need international funding due to domestic price cuts. In 2023, biotech licensing deals hit a record USD 44.1 billion, and have continued strongly into 2024. (Source (EN): Financial Times)

In their effort to optimize China’s digital infrastructure for artificial intelligence, China’s state-owned telecommunications companies (telcos) are venturing beyond their traditional remit to develop AI models, even if private firms have traditionally been much better at developing cloud-based services. 

These changes are driven by a saturated 5G base station market and skyrocketing demand for smart computing power centers – big data infrastructure equipped for handling AI model training and use. The Ministry of Industry and Information Technology (MIIT) stipulated last year that by 2025, China’s overall computing power should grow 15 percent over 2024 to 300 exaflops (1 exaflop corresponds to a computing performance of one quintillion floating-point operations per second). Smart computing power should grow even faster, said the ministry, to reach 105 exaflops, 35 percent of the total. 

On the supply side, tightening US export controls limits China’s access to the most advanced graphic-processing units (GPUs), which are needed to train and deploy large language models (LLMs). Chinese firms are increasingly forced to turn to domestic alternatives that are not as good. 

Chinese telcos’ pivot to the “model-as-a-service” business carries a risk of further stretching these limited resources. China Mobile, China Telecom and China Unicom are each spending over CNY 10 billion to purchase AI servers and to deliver LLMs as services to their users. China Mobile, for instance, launched a full-service computing power center, equipped with nearly 4,000 AI accelerator cards and 33 percent domestic AI chips, offering an exaflop of performance. This is only one of 12 such centers China Mobile operates. 

Moreover, it is unclear whether telcos can be competitive in this space, unless they partner with private tech firms. Treating AI models as infrastructure mirrors earlier conceptions of “public clouds.” The cloud services that the telcos have developed over the years primarily serve government and state-owned enterprise customers. Already, Xingchen, the flagship LLM of China Telecom, follows this pattern. 

MERICS Lead Analyst Rebecca Arcesati says: “Scarcity in computing power is likely to affect China more in the future. Having telecommunications operators develop their own LLMs could hamper the efficient use of the limited computing power available.” 

Digital connectivity – notable developments: 

Tech progress

• A team from Beijing University of Posts and Telecommunications claims to have built the world's first 6G field network, integrating AI and communication. China aims to commercialize 6G technology by 2030, with standards anticipated in 2025. (Source (CN): S&T Daily) 
• At the 2024 World Artificial Intelligence Conference in Shanghai, Ant Group launched its YinYu Cloud platform for privacy-preserving computing of large models. Data owners are so far hesitant to share their data with large models for fear of data leakage. (Source (CN): Leiphone) 
• Alibaba Cloud won first prize in the 2023 China Association for Automation Science and Technology Progress Award for its ultra-large-scale cloud network technology. (Source (CN): Leiphone) 

Domestic dynamics

• China is set to launch its G60 Starlink satellite constellation in early August, aiming to deploy 1,296 satellites initially and expand to over 14,000. This state-backed initiative seeks to provide global internet coverage including remote areas and compete with SpaceX's Starlink. (Source (CN): 36kr) 
• China's first nationwide survey on data resources, was released on May 24 at a digital economy summit in Fuzhou, Fujian province. The survey suggests that China is underutilizing its data, with less than 3% of data generated last year stored and processed. (Source (EN): SCMP) 

Foreign involvement 

• Alibaba Cloud announced investments in new, AI-focused data centers in South Korea, Malaysia, the Philippines, Thailand, and Mexico, expanding its global network to 31 regions and 95 availability zones. Alibaba hopes to take a significant market share in the emerging AI cloud market in the Global South. (Source (CN): S&T Daily) 
• At the World Mobile Communications Conference in Shanghai in 2024, Huawei announced 5G-A, an evolution of 5G, to enhance capabilities in speed, bandwidth, and latency, particularly in high-demand scenarios. (Source (CN): 36kr) 

Semiconductors – notable developments:

Tech progress

• A team at Beida published a new way to grow crystals for semiconductors in the prestigious international magazine “Science.” (Source (CN): S&T Daily)

Domestic dynamics

• China’s government is urging automakers to purchase 25% of their chips locally by 2025. These non-mandatory guidelines follow a line of similar announcements, including Xi Jinping asking a company CEO whether he already sources his chips domestically (Source (CN): EETimes)
• The third tranche of the Big Fund is starting to invest. First investments went to Hua Hong Semiconductors, while strategically, the fund will invest more in AI-relevant chips. (Source (CN): 36kr)

Foreign involvement

• After Nvidia again revamped its China offerings to follow US export controls in 2023, Chinese companies are now choosing domestic offerings more frequently. Huawei, Baidu Kunlun and Moore Thread are all offering chips that are cheaper and good enough. (Source (CN): Caijing, Caijing)
• More than 50% of Japanese semiconductor manufacturing equipment exports go to China. Surrounding US export restrictions, Chinese companies have ramped up buying of semiconductor manufacturing companies to prepare for possible stricter controls in the future. (Source (CN): 36kr)

Green tech – notable developments:

Tech progress

• China overtook Denmark as the world leader in wind turbine patents in 2023. Chinese firms are looking to expand their 60 percent global market share as the US and Europe move to protect their industries. (Source (EN): Nikkei Asia)

Domestic dynamics

• Construction began on China's largest offshore photovoltaic power station to date. The 2 GW offshore power station is being built by China National Nuclear Power in Jiangsu Province. (Source (CN): S&T Daily)
• China’s Ministry of Industry and Information Technology (MIIT) released draft rules raising, the minimum capital ratio for solar PV manufacturing projects. The measures aim to limit new investments in the government-backed sector as it faces overcapacity. (Source (CN): MIIT)
• In July, both Jinko Solar and Trina Solar terminated plans to raise money for capacity expansion. Heavy investment in the sector has led to a stark reduction in prices for solar modules and eroded profit margins. (Source (CN): Xinhua)

Foreign involvement

• Chinese EV manufacturers are expanding into African and Middle Eastern markets. The emerging "Nezha Automobile (NETA)" entered Kenya, and Xpeng Motors launched new cars in Egypt. (Source (CN): 36kr)
• Jinko Power Technology announced plans to sell its Antequera solar farm project in Spain. The firm will sell the 175 MW solar project to a Hong Kong-based subsidiary of Chinese hydropower titan China Huadian, a state-owned company. (Source (EN): Ycai)

Quantum – notable developments:

Tech progress

• Domestic dilution chiller ez-Q Fridge was delivered to a customer. The fridge is the country's first commercially available mass dilution chiller required to make a superconducting quantum computer. (Source (CN): People’s Daily)

Domestic dynamics

• The China Quantum Computing Industry Summit in Guangzhou brought together over 100 key players to discuss partnerships, ecosystem development, and talent growth. Guangdong Guoteng and China Unicom launched the "Tianji" Quantum Security Service Platform. (Source (CN): S&T Daily)
• Researchers (including Pan Jianwei) at the University of Science and Technology of China have built a cold-atom quantum simulator that confirmed the antiferromagnetic phase transition in the Fermi-Hubbard model, for the first time surpassing classical computing capabilities. (Source (CN): S&T Daily)

International engagement

• The US Commerce Department added China’s major entities engaged in quantum research and development to its Entity List for acquiring or attempting to acquire US-origin items to enhance the PRC's quantum capabilities. (Source (EN): Bureau of Industry & Security)
• On June 17, China’s “Original Wukong” superconducting quantum computer was reported to have surpassed 10 million global visits and to have completed 236,000 quantum tasks. This achievement is seen as an indicator for China’s progress. (Source (CN): S&T Daily)

Huawei has emerged as the leader for semiconductors in China. It designed the first domestically produced 5G-capable smartphone chip, the Kirin 9000S and is invested in 97 companies through its Hubble Investment arm. In the global technology competition between China and the US, Huawei thus now plays a key role in two of the most critical technologies, mobile internet and semiconductors. With its solar inverters, forays into autonomous driving and its own AI framework, PaddlePaddle, Huawei is now active across many of the technology areas the government has identified as critical. 

By some metrics, the company is one of the most innovative, , and its focus on manufacturing and hardware over services and software is closely in line with Beijing’s priorities. It is thus likely to continue to play a very important role in critical technology development in China, and is turning into a manufacturing champion like Samsung, active across many technologies.

Known today for its highly disciplined, militaristic “wolf” culture – Huawei's close ties to the government facilitated its meteoric rise as a telecommunications champion in the 1990s and early 2000s, fueled by the government’s standardization and regulatory power during the 4G era. The company invested in indigenous innovation early, and its successes in 5G ultimately led to the US government taking notice and placing Huawei on its Entity List in 2019 for violating Iran sanctions. 

Huawei’s addition to the Entity List reinforced to China’s government the importance of self-sufficiency in key technologies, a lesson it had learned thirty years earlier when the US instituted embargos after the crackdown on protesters following the Tiananmen uprising. 

The company, having demonstrated its ability to innovate with 5G, is now involved across the semiconductor supply chain, and is seen as responsible for key breakthroughs, including designing the most advanced Chinese competitor to Nvidia GPUs, needed for Artificial Intelligence, today. 

Sources:

• Huawei is quietly dominating China’s semiconductor supply chain (Source (EN): MERICS)
• Huawei has basically realized the localization of EDA tools for 14nm and above, and domestic EDA companies are catching up (Source (CN): Dramexchange)