SOEs actively deploying DeepSeek AI models

In a move toward large-scale implementation of artificial intelligence technologies, several State-owned enterprises have recently announced an accelerated integration with DeepSeek, China's homegrown AI reasoning model that has taken the world by storm recently.

To date, major energy giants, such as Sinopec, PetroChina, China National Offshore Oil Corp, Sinochem Group, CHN Energy and China Southern Power Grid, have all completed private deployments of DeepSeek's AI models into their proprietary large language models.

Sinopec plans to leverage DeepSeek's high-quality programming capabilities to optimize seismic data processing, reservoir development, chemical product research, and customer service, further propelling the petrochemical industry toward intelligent and digital transformation.

Similarly, Sinochem highlighted DeepSeek's advanced capabilities in complex information processing, logical reasoning and open-domain knowledge comprehension, which are expected to boost its core business operations in a wider range of application scenarios, the group said.

This round of AI deployment by energy SOEs follows a recent such move by the telecommunications sector, wherein the country's three major operators — China Mobile, China Telecom and China Unicom — first used the country's hottest tech startup's AI models in their cloud services.

Beyond telecommunications and energy players, more SOEs are expected to collaborate with tech companies like DeepSeek to advance AI applications in communication technologies, cybersecurity, and smart office solutions, with more cooperation opportunities anticipated in sectors such as transportation, logistics and finance, said Zhou Lisha, a researcher at the Institute for State-owned Enterprises of Tsinghua University.

"DeepSeek's emergence has realized equal AI access," said Zhou. "In the future, competition in the AI area will be driven by the scale and quality of data. With the vast data resources at the disposal of SOEs, their integration with DeepSeek will help accelerate data circulation and commercialization."

The automotive, financial, technology and construction sectors are also actively advancing their own LLMs through further integration.

China Energy Engineering Group, a central SOE engaged in power and infrastructure, has completed the full-platform integration of DeepSeek models, embedding them into its self-developed AI assistant for comprehensive AI empowerment. Dongfeng Motor Corp, a key State-owned automaker, has also integrated DeepSeek models into its autonomous brands such as Voyah, Mengshi and Yipai.

Such upgraded LLMs in the automotive industry can facilitate enterprise data analysis and intelligent decision-making, as well as boost intelligent driving and in-car interaction capabilities, while in transportation and logistics, they could optimize infrastructure construction, delivery route planning and smart warehouse management, Zhou said.

The recent AI integration deployments are part of a broader "AI+ initiative" spearheaded by State-owned assets regulators to accelerate the deep integration of AI into SOEs' diverse production scenarios.

At a meeting held by the State-owned Assets Supervision and Administration Commission of the State Council last week, central enterprises were urged to prioritize AI development in their 15th Five-Year Plan (2026-30) and increase funding to bolster AI research and development.

The meeting also underscored the strategic window for AI development, calling on SOEs to focus on mastering core technologies and fostering original innovations.

Three high-value application scenarios were highlighted — strategic significance (such as energy and communication infrastructure), economic benefits (such as financial risk control and intelligent operations), and close ties to people's livelihoods (such as smart governance and environmental management).

Li Jin, chief researcher at the China Enterprise Research Institute in Beijing, said the emphasis on high-value scenarios sets clear boundaries for SOEs' AI development, preventing chaotic allocation of R&D resources at the source. Looking forward, cross-industry and cross-ownership collaborations among SOEs will become one of the key drivers for AI breakthroughs and LLM technologies, Li said.

"For central SOEs, the focus (of AI development) should not be on competing for talent, but on strengthening internal capabilities. It is through reforming institutional mechanisms that they can fully unleash the potential of existing talent, tolerate innovation failures and ignite bold experimentation," Li said.