China explores military applications with DeepSeek

Asia Times

28-02-2025

DeepSeek's artificial intelligence (AI) model can be used in various military applications, from controlling unmanned vehicles to giving commands, according to a white paper recently published by Chongqing Landship Information Technology, an autonomous driving solution provider.
Landship said DeepSeek has excellent potential in military use, such as command, communications, and intelligence, surveillance, and reconnaissance (ISR).
'In military strategic planning, DeepSeek can conduct in-depth analysis of massive intelligence information and provide military commanders with accurate decision-making support,' it said. 'DeepSeek has powerful language understanding and generation capabilities, allowing it to adapt to battlefield tasks quickly.'
Landship added that DeepSeek could collaborate with satellites, radars, and drones to improve the efficiency and accuracy of military reconnaissance. It can quickly identify key military targets from satellite images and estimate their sizes and numbers, supporting military decision-making.
Landship, formerly Beijing Landship Information Technology Co Ltd, was co-founded in 2012 by Zhang Dezhao and several other autonomous driving experts from Tsinghua University.
In 2015, the same team founded Beijing Zhixingzhe Technology Co, or IDriverPlus, to develop self-driving solutions, including a product called IDriverBrain.
IDriverPlus and Landship brand their products as Xingji, which means 'star horse' (or Qianlima — a mythical horse that can run very fast) in Chinese.
Landship said on February 27 that it has deployed DeepSeek in a self-driving military vehicle called Xingji P60. It displayed the vehicle at the International Defence Exhibition and Conference (IDEX 2025) in Abu Dhabi from February 17 to 21.
The company said the P60 integrated Landship's civil-use self-driving software and DeepSeek's military-use large language models (LLMs) to achieve efficient information processing and decision-making in complex and changing environments. LLMs refer to AI models like ChatGPT, which can understand human language.
IDriverPlus' Chief Technology Officer and Landship's General Manager Wang Xiao said the company hopes to lead new trends in the defense sector by offering high-performance products at affordable costs.
According to online footage, the vehicle was not called P60 but CS/VP16B at the exhibition. It was shown at the booth of Norinco, or China North Industries Corporation, a state-owned defense and technology company.
Last October, IDriverPlus signed a strategic cooperation agreement with Harbin No.1 Jiqi Manufacture Group Company Ltd, a unit of Norinco, to jointly develop all-terrain unnamed rescue vehicles.
Huawei's support
DeepSeek launched its latest AI model, DeepSeek R1, on January 20. It trained the AI model using only 2,000 Nvidia H800 graphic processing units and a low-cost method called 'knowledge distillation.'
During the distillation, DeepSeek asked a more intelligent chatbot, such as Alibaba's Qwen or OpenAI's ChatGPT, many questions and used the answers to fine-tune its logic to become more intelligent.
DeepSeek R1's debut in late January caused a slump in United States stocks, as investors worried they might have overvalued AI stocks.
Xu Bingjun, a military columnist and a senior researcher with Xinhua Liaowang think tank, praised DeepSeek's low-cost strategy in a recent article titled 'How DeepSeek changes military AI and its impact on the United States and Western countries.'
'DeepSeek can be applied to intelligent combat systems to significantly improve combat effectiveness by analyzing real-time battlefield situations, optimizing combat plans, and predicting enemy actions,' he says in the article.
'With DeepSeek, more autonomous weapons, drones, and unmanned tanks and ships can be developed,' he says. 'These unmanned combat platforms can perform tasks autonomously in complex environments, reducing casualties while improving combat efficiency.'
In early February, US lawmakers introduced a bipartisan congressional bill to ban DeepSeek from government devices due to national security concerns. Some states, like New York and Virginia, the US Navy, and the National Aeronautics and Space Administration (NASA), have stopped their employees from using DeepSeek on official devices.
Landship said it worked with Huawei Technologies' Mobile Data Center (MDC) to draft its latest White Paper about DeepSeek's potential military applications. It said it hopes to promote using DeepSeek in the military field through cooperation.
The White Paper disclosed Huawei's AI goals for 2025:
March – Deploying DeepSeek R1 on Huawei MDC's system to realize semantic understanding and assisted decision-making, and also to improve real-time reasoning;
April – Deploying DeepSeek's Janus-pro 7B on Huawei MDC's system to realize multi-modal video understanding and improve real-time image understanding;
June – Running DeepSeek's natural language processing (NLP) program with Huawei's 310P and Fuzhou Rockchip Electronics' RK3588 chips;
September – Testing DeepSeek-powered drones;
December – Using the drones to identify complex camouflage targets.
On August 17, 2022, IDriverPlus and Huawei signed a comprehensive cooperation agreement to develop 'autonomous driving + AI' solutions using Huawei's Ascend AI chips.
On September 9 of the same year, Huawei's Vice President Deng Taihua visited IDriverPlus' headquarters in Beijing to discuss self-driving vehicle projects. IDriverPlus said it would test its software at Huawei's Ascend AI Computing Center, which commenced operations in February 2023.
Yong Jian is a contributor to the Asia Times. He is a Chinese journalist who specializes in Chinese technology, economy and politics.
Read: DeepSeek is now the brain of Chinese state-owned firms