China’s artificial intelligence industry has entered what analysts are calling a “Warring States Period,” a phase of intense, multi-front competition among a well-defined set of powerful players, each with distinct advantages, strategies, and capital structures. A new analysis by Park Jimin, CEO of PWS GROUP and a specialist in Chinese AI, identifies eight major factions organized into three distinct capital models, offering a framework for understanding the complex competitive dynamics that will shape the industry’s future.
The “Warring States” metaphor is apt. Just as China’s historical period saw powerful kingdoms competing for supremacy before eventual unification, China’s AI landscape is now characterized by well-resourced competitors fighting across multiple dimensions simultaneously, model performance, user acquisition, enterprise penetration, and hardware self-sufficiency. The outcome of this competition will determine not only which companies survive, but also the shape of China’s AI industry for the next decade.
The Three Capital Models
Park’s analysis organizes the eight factions into three distinct models based on their capital structure and strategic orientation.
The Large Enterprise Model encompasses the self-funded technology giants that are building AI as an extension of their existing businesses. ByteDance’s Doubao leads this group in terms of consumer reach, with approximately 345 million monthly active users as of March 2026, according to QuestMobile data. ByteDance invested roughly 150 billion yuan ($22 billion ) in AI infrastructure in 2025 alone. Alibaba’s Qwen platform has approximately 166 million MAU and is backed by a commitment of 380 billion yuan ($55.7 billion) over the next three years across its cloud, e-commerce, payments, and logistics operations. Tencent’s Hunyuan and Yuanbao products benefit from the company’s 79 billion yuan ($11.6 billion) in capital expenditure in 2025 and deep integration into the WeChat ecosystem. Baidu’s ERNIE (Wenxin) is pivoting toward AI cloud services and autonomous driving, increasingly open-sourcing its models. Huawei’s Pangu model focuses on industrial AI for manufacturing, finance, and government, backed by the company’s extraordinary annual R&D spending of 192.3 billion yuan ($28.2 billion).
(Related: China’s Agentic AI Boom: How OpenClaw Drove Token Consumption to 140 Trillion Per Day)
The Research Lab Model is represented primarily by DeepSeek, which was spun off from the quantitative hedge fund High-Flyer Capital Management and is controlled by founder Liang Wenfeng. DeepSeek’s approach is defined by a relentless focus on efficiency over scale, its V3 model was trained for approximately $5.576 million, a fraction of what US competitors spend. This capital efficiency has allowed it to achieve frontier-level performance while operating with far fewer resources than its rivals, making it a uniquely disruptive force in the industry.
The Policy-Industry Linked Model includes companies that have been built through a combination of academic research, private investment, and state capital. Zhipu AI, which emerged from Tsinghua University’s technology transfer program, raised approximately 8.363 billion yuan before its Hong Kong listing and benefits from local government investment from entities like Hangzhou’s state capital arm. iFlytek’s Spark model focuses on industry-specific AI applications in education, healthcare, translation, and public services, leveraging its deep relationships with government clients.
(Related: China’s Token Economy Mints New AI Billionaires as MiniMax and Zhipu Surpass Baidu in Market Value)
The Key Battlegrounds
Park’s analysis identifies several key dimensions on which these factions are competing. Model performance, while important, is increasingly becoming a commodity—the gap between the top models is narrowing, and differentiation on benchmarks alone is insufficient to build a sustainable competitive advantage.
The more decisive battlegrounds are inference cost and industrial deployment. As Park notes, “Competition in China’s AI will not be decided by a single model performance leaderboard. The winners will be determined by who processes inference at lower cost and secures industrial sites and local government projects.” This observation reflects the maturation of the Chinese AI market, where the focus is shifting from building impressive models to deploying them profitably at scale.
Inference cost is particularly critical because it determines the economics of AI services. Companies that can deliver equivalent performance at lower cost will be able to offer more competitive pricing, attract more users, and generate more data for further model improvement. The efficiency innovations pioneered by DeepSeek have set a new standard, and all players are now racing to match or exceed them.
The Role of Ecosystems
A critical differentiator in the Warring States competition is the depth of each faction’s ecosystem. The large enterprise players, ByteDance, Alibaba, Tencent, and Baidu, have significant advantages here, as they can embed AI into existing products with hundreds of millions of users, generating data and revenue at scale.
ByteDance’s integration of AI into Douyin (TikTok’s Chinese counterpart), its news aggregation platform Toutiao, and its productivity tools gives it unparalleled reach. Alibaba can embed Qwen into its e-commerce, cloud, and enterprise software platforms. Tencent’s WeChat integration creates a distribution channel that is virtually unmatched in the consumer market.
(Related: ByteDance’s Doubao Tests Paid Subscriptions, Ending China’s Free-AI Era)
What Comes After the Warring States?
The historical Warring States period ended with the unification of China under the Qin dynasty. Whether China’s AI industry will similarly consolidate around a single dominant player or evolve into a stable oligopoly of several major factions remains an open question.
What is clear is that the current period of intense competition is accelerating innovation and driving down costs, benefiting Chinese AI users and enterprises. The companies that emerge from this period will be battle-hardened, efficient, and deeply embedded in China’s economic infrastructure. They will be formidable competitors not just domestically, but on the global stage.