On January 4, global consultancy Frost & Sullivan unveiled its 2025 China GPU Cloud Market Research Report, highlighting significant shifts in market dynamics.

The report reveals that in the highly specialized field of self-developed GPU cloud services, Baidu Intelligent Cloud has claimed the top spot with a 40.4% market share, showcasing its leadership in product innovation, computational capacity, and industry-wide adoption.

The term “self-developed GPU cloud” refers to a market model where vendors construct and operate intelligent computing clusters comprising tens of thousands of cards, built on proprietary AI chips. These clusters deliver comprehensive AI computational power through cloud services, emphasizing a closed-loop ecosystem that integrates chips, clusters, and cloud platforms. This model serves as a rigorous test of a vendor’s full-stack technological capabilities.

According to Frost & Sullivan, China’s GPU cloud market has evolved from a “functional” phase to an “efficient and sustainable” one, driven by the rapid deployment of large AI models. Competition now hinges on full-stack capabilities, including software-hardware synergy, large-scale cluster management, and commercial delivery efficiency.

Baidu Intelligent Cloud’s rise to the forefront of the self-developed GPU cloud sector is part of a broader trend. Previous IDC reports have consistently ranked it as China’s leading AI public cloud provider for six consecutive years. In the AI infrastructure race, vendors with full-stack self-development and vertical integration capabilities are gaining a competitive edge.

Baidu Intelligent Cloud’s market dominance stems from its long-term investment in a four-layer full-stack AI architecture, spanning chips to applications. This unique framework forms its core competitive advantage.

At the foundational computational layer, proprietary chips and ultra-large-scale clusters serve as the bedrock. AI acceleration chips centered around the Kunlun Core have been deployed at scale across hundreds of enterprises in finance, energy, and manufacturing.

In 2025, Baidu Intelligent Cloud launched China’s first fully self-developed 30,000-card Kunlun Core cluster, capable of training multiple large models with hundreds of billions of parameters simultaneously. This milestone marks a significant leap in domestic computational power.

At the recent Baidu World 2025 conference, the company announced that the Kunlun Core will adhere to an annual generation upgrade cycle and introduced the new “Tianchi” super-node, which substantially enhances interconnection bandwidth and overall performance.

At the platform services layer, the Baige and Qianfan platforms act as an intelligent engine. The Baige AI Computing Platform addresses computational efficiency bottlenecks, providing high-performance support for complex AI tasks.

The Qianfan Platform focuses on democratizing AI application development. The upgraded Qianfan 4.0 offers an agent-centric, one-stop enterprise service platform. Currently, it serves over 460,000 enterprise users and hosts over 1.3 million agents, becoming a pivotal platform for large-scale agent deployment.

This end-to-end self-development, from hardware to platforms, creates a tightly integrated technological ecosystem. It enables Baidu Intelligent Cloud to optimize deeply for large model training and inference, delivering end-to-end performance improvements and cost efficiencies to meet enterprises’ full-lifecycle needs, from model development to business deployment.

The true value of technological prowess lies in industrial adoption. Baidu Intelligent Cloud’s full-stack capabilities are driving AI penetration across key sectors, including finance, transportation, gaming, and energy.

Currently, over 65% of central enterprises, all systemically important banks, 95% of mainstream automakers, and more than half of gaming firms rely on Baidu Intelligent Cloud for large model deployment.

In the financial sector, Baidu Intelligent Cloud’s collaboration with China Merchants Bank, based on the Kunlun Core, enabled efficient training of hundred-billion-parameter models using minimal servers.

At the business level, more profound transformations are underway. The off-exchange trading agent co-developed by Galaxy Securities and Baidu Intelligent Cloud revolutionized traditional manual quotation processes. Capable of sub-second responses without omissions, it increased customer conversion rates from inquiry to order placement from 10% to 30%.

In enhancing user experience and service intelligence, China Eastern Airlines’ digital employee “Dongdong,” developed with Baidu Intelligent Cloud, is seamlessly integrated into its app. It covers end-to-end services, from ticketing, seat selection, and check-in to complex pet shipping inquiries.

In gaming and cutting-edge technology, 37 Interactive Entertainment’s AI customer service system, built on the Wenxin Large Model, achieved a response accuracy rate of 99.3%.

In the first half of 2025, Baidu’s AI-driven new business revenue, including intelligent cloud services, surpassed 19.4 billion yuan, up 36% year-on-year, maintaining leading growth among major cloud providers.

The evolving market landscape clearly indicates the next phase of AI competition: isolated technological strengths no longer suffice as a durable competitive advantage. The ability to deeply integrate software and hardware while translating technology into scalable industrial value will be decisive.

References:

https://finance.sina.com.cn/jjxw/2026-01-04/doc-inhfcpht8932004.shtml

https://tech.cnr.cn/techph/20260104/t20260104_527482716.shtml

https://www.sohu.com/a/958566841_120103973?scm=10001.325_13-325_13.0.0.5_32&spm=smpc.channel_248.block3_308_NDdFbm_1_fd.1.1764128759745wm7GOXb_324