china’s top three telecommunications companies—china mobile, china unicom, and china telecom—have integrated deepseek’s open-source models into their systems. they have also provided exclusive computing power solutions and optimized environments for the deepseek-r1 model to enhance its performance, the country’s ministry of industry and information technology announced on february 9.
other major chinese tech giants, including alibaba, baidu, and tencent, have also incorporated deepseek’s models into their cloud platforms, accelerating the advancement of intelligent applications.
meanwhile, byd, china’s largest electric vehicle (ev) manufacturer, which sold approximately 4.2 million evs last year, announced on february 11 that it will integrate deepseek’s software into 21 of its ev models. this will enable advanced autonomous driving features across its 18 models priced above 100,000 yuan ($13,686).
deepseek’s ai models stand out for their cost-efficiency, with training expenses reportedly much lower than other large language models, while delivering performance comparable to openai’s gpt-4o.
the widespread adoption of deepseek’s technology is expected to provide cost-effective, high-performance ai services for small and medium-sized enterprises (smes) and individual users, lowering the barriers to ai adoption and accelerating the integration of ai across industries.
impact on metal demand
the expansion of ai is poised to boost long-term demand for key metals, including copper, aluminum, tungsten, molybdenum, gallium, germanium, battery metals, and rare earth elements.
ai operations rely heavily on data centers and servers, where copper plays a crucial role in power distribution, grounding, and interconnectivity. industry projections estimate that global copper demand from data centers will surpass 1 million tons by 2026. the rapid growth of ai is also driving increased copper demand in electrical grid systems. additionally, aluminum is used in cooling and structural components of data centers.
demand for indium phosphide (inp) photonic integrated circuit (pic) technology is rising due to the high computational workloads of ai-driven data centers. the expansion of ai infrastructure is also expected to increase the use of compound semiconductors such as gallium nitride and gallium arsenide. molybdenum and tungsten are used in high-temperature components and electrode materials for advanced ai hardware.
rare earth metals play a key role in ai-related magnetic and optical materials. the rapid development of ai products is expected to drive demand for neodymium iron boron (ndfeb) magnets used in precision motors, rare earth polishing powders for wafer processing, and magnets in audio devices.
furthermore, major chinese smartphone brands—including huawei, honor, and oppo—have integrated deepseek’s ai into their products, accelerating the development and adoption of ai-powered smartphones. industry forecasts suggest that ai smartphone shipments will reach 550 million units globally by 2027, accounting for more than 40% of total phone shipments. the consumer electronics sector currently consumes about 30% of the world's cobalt and 7% of global lithium production.
challenges and global response
despite its growing presence in china, deepseek faces challenges abroad. in january, italian regulators removed the deepseek app from the country’s app store due to data security concerns. australia has banned the use of deepseek technology on all government devices, while major japanese corporations such as toyota, mitsubishi, and softbank have prohibited its use over information security risks. in february, texas became the first u.s. state to ban deepseek on government equipment.
however, concerns about deepseek in western countries may accelerate the development of their own ai industries. former u.s. president donald trump called deepseek a "wake-up call" for the american tech sector. meanwhile, south korea’s acting president, choi sang-moo, described deepseek as a “new impact” and announced plans to invest 34 trillion won ($23.5 billion) into ai and semiconductor development.