_________

One of China's hottest semiconductor startups has teamed up with one of the country's top robotaxi companies. Moore Threads, the Chinese GPU manufacturer and "Nvidia chaser", announced a partnership with Pony.ai, a leading developer of autonomous driving solutions.

The latest generation of Moore Threads' AI chips, the MTT S5000, and its KUAE intelligent computing cluster, including integrated software, will be used to train Pony.ai's world model and accelerate the scaling of its robotaxi fleet with L4 autonomous driving, the two companies announced.

This cooperation marks an important milestone for China's evolving automotive ecosystem, where cutting-edge manufacturers and mobility providers can now rely on domestically produced computing power.

For details about the technology involved and what tie-ups like this mean for US attempts to stall China's progress in artificial intelligence, please click here for this week's full Must Read.

_________

The weeks leading up to the Chinese New Year saw an intense cluster of AI model launches.

ByteDance released Seedance 2.0, a text-to-video model capable of producing high-quality multi-shot sequences in 60 seconds. Simultaneously, Zhipu AI revealed GLM-5, an open-source model prioritising coding and agentic engineering.
GLM-5 integrates “sparse attention” to lower costs and reached the top spot among open-source models on benchmarking sites shortly after its reveal, causing Zhipu's shares to surge 29 per cent.

Alibaba contributed to the momentum with Qwen3.5-Plus, an open-source foundation model that outperformed Gemini 3 Pro in reasoning tasks while costing only one-eighteenth as much.

These launches coincide with massive growth in China's generative AI user base, which reached 602 million by late 2025.

Chinese frontier AI companies are now competing with US rivals in real time. The shrinking performance gap and superior cost efficiency of Chinese computing are increasingly defining the global AI landscape.

For details and more analysis, click here to read the full, free version of this week’s Highlights:

Beijing-based Galaxea AI completed a 1 billion yuan Series B round in February 2026, lifting its valuation above 10 billion yuan, or close to US$1.5 billion.

Founded only in September 2023, it is the youngest of China's four embodied AI unicorns.

Positioned as an "embodied AI foundation company," Galaxea focuses on "brain, body, and toolchain" capabilities rather than traditional manufacturing. Its G0 Plus model and R1 Pro robot are already seeing demand from research and corporate clients.

Galaxea’s rapid ascent shows that investors are prioritising platform providers over hardware sellers. The market is pricing in extraordinary growth for companies defining the software-driven future of robotics.

Han’s CNC Technology raised HK$4.83 billion (US$619–620 million) in its Hong Kong IPO, drawing major cornerstone investors like GIC and Schroders.

As China’s leading manufacturer of printed circuit board (PCB) production equipment, the company is leveraging the AI boom. Increased demand for AI servers and GPUs has driven a surge in orders for the high-layer-count PCBs Han’s CNC specialises in.

This IPO illustrates how the explosive growth of large AI models is reshaping the global electronics industry. The market is now pricing AI exposure into the entire manufacturing supply chain as automation cascades through Chinese factories.

Li Auto CEO Li Xiang has called the forthcoming L9 SUV a "pioneering work of embodied intelligent robots."

Priced at 559,800 yuan, or around US$81,000, the vehicle features in-house M100 chips and advanced AI designed to recognise and anticipate occupant needs.

This „all in with AI” strategy comes as Li Auto seeks to rebound from a sales contraction in 2025, targeting 40 per cent growth for 2026.

To see this partial rebranding from a carmaker to a robotics company purely as a scramble onto the AI bandwagon to impress investors would be wrong, although there is a marketing aspect.

There is, however, a genuine convergence between the automotive and robotics industries. The "world models" enabling autonomous navigation in cars are fundamentally the same as those teaching humanoids to navigate factories.

China Southern Power Grid, one of the country’s two main state-owned electricity networks, has released a dedicated “AI+” strategy and begun implementing it.

Working with Huawei, Baidu, and SenseTime, the company has built an intelligent computing cluster with a total capacity exceeding 3,000P in total scale, Chinese media reported.

The grid operator has developed a proprietary model system called "MegaWatt," comprising three foundational large models, 31 domain-specific models and more than 3,000 scenario models. Its natural language processing model is called MegaWatt NLP 2.0.

Southern Power Grid’s strategy illustrates how the “AI+” mandate handed down by Beijing to central SOEs is translating into large-scale infrastructure investment and operational deployment.

Researchers in Tianjin are integrating AI, brain-computer interfaces (BCI) and wearables to assist stroke recovery. By using brain waves to guide a device that stimulates pressure points, they are developing a "Traditional Chinese Medicine (TCM) brain-computer interface".

While clinical trials are under way, the government has already supported testing earlier versions of this method on the Shenzhou spacecraft.

There are around 700,000 TCM practitioners treating over 1 billion patients annually in China.

Large models and embodied AI offer an "intelligence upgrade" for a huge industry, helping to organise scattered historical knowledge and augment Western medical practices. There are also opportunities for Western medical device makers.

For more details and analysis of all of these developments, please click here for the long, free version of the China AI2X Briefing:

_________

_________

Number of tokens processed per minute at its peak by ByteDance for its AI app Doubao on the day of China’s Spring Festival Gala.

Source: Xin Jingbao

Number of times the phrase "Qwen help me" was sent to Alibaba's AI app in the 11 days leading up to the Chinese New Year.

Source: Xinlang Keji

Total marketing spend by Chinese tech firms Alibaba, Baidu, ByteDance, and Tencent to promote their AI applications during the 2026 Chinese New Year holiday.

Source: Zhongguo Xinwen Wang

Percentage increase in the stock price of Zhipu.ai, developer of the Chinese AI model GLM-5, between its IPO in Hong Kong on January 8 and February 13, 2026.

Source: Bloomberg

_________

On February 10, 2026, eight central government departments led by the National Development and Reform Commission jointly released the "Implementation Opinions on Accelerating the Promotion and Application of Artificial Intelligence in the Bidding and Tendering Field."

Beijing is deploying AI as a tool against bid-rigging and corruption in public procurement. The document outlines 20 priority scenarios in which AI will flag irregularities and supervise tender review committees.

Full coverage of key detection scenarios in selected provinces and cities is targeted for the end of 2026, with nationwide rollout to follow by the end of 2027.

As the corruption problem is systemic, deeply embedded into the system, no independent observes expect a fundamental change, however. AI cannot replace a free press and an independent, capable judiciary. The risk for corrupt officials has just gone up a little bit, however.

On February 11, 2026, Premier Li Qiang presided over a State Council study session on AI development, calling for a comprehensive push across innovation, industrial growth and real-economy application.

The session signals that AI promotion has moved to the direct agenda of China's cabinet. Li Qiang called for breakthroughs across the full technology chain, large-scale commercial deployment, stronger coordination of data, computing power and electricity supply, and improvements to AI governance through updated laws, standards and ethical guidelines. In short, China signals that it aims to speed up the deployment of AI in its industries even further this year.

On February 10, 2026, the State-owned Assets Supervision and Administration Commission (SASAC) convened a meeting to deepen the "AI+" special action plan for China's central state-owned enterprises.

China's centrally managed state enterprises have been directed to significantly expand investment in computing power and to pursue coordinated development of "computing power plus electricity." State-owned enterprises (SOEs) are also required to intensify efforts in large-model technology and to identify high-value AI use cases aligned with their core business operations.

This essentially means that the central leadership of the Communist party and the government try to leverage their direct control over SOEs to push for an acceleration of AI deployment across all sectors.

On February 12, 2026, the Shenzhen Municipal Bureau of Industry and Information Technology issued the "Shenzhen AI+ Advanced Manufacturing Action Plan 2026 to 2027."

By 2027, Shenzhen aims to establish a national pilot base for AI applications in advanced manufacturing, develop one hundred application scenarios and launch one hundred vertical industry models. The same plan calls for pilot programmes on integrated "vehicle-road-cloud" systems and AI empowerment across the full automotive industry chain.

On February 3, 2026, Shanghai Mayor Gong Zheng's Government Work Report committed the city to adding more than 50 advanced smart factories in 2026 under its ongoing "AI Plus" initiative. The city will also strengthen computing infrastructure and promote the broad deployment of AI agents and next-generation intelligent terminals across key industries.

_________

Chinese tech giants are spending the equivalent of US$650 million on bubble tea and other promotional gifts during the "2026 Spring Festival AI War".

Alibaba, ByteDance, Tencent, and Baidu battle over more than traffic, trying to make users adopt the habit of ordering everything from drinks and food to airline tickets and hotels through agentic workflows.

The companies hope to establish themselves as "super gateways" of internet traffic in this new era, effectively shifting the competition from parameter contests over who has the best large model to who can best monetise AI.

For an analysis of this user acquisition strategy and the emergence of AI as a business model beyond subscriptions, click here to read this week's Deep Dive.