Inside the End of China’s AI 'Hundred-Model War' and the Shift to Enterprise Value

That dynamic has now fundamentally reversed. JPMorgan notes the market has moved from a focus on "who could develop a working model" to a three-part survival test: continuous frontier performance, commercial viability, and the ability to compete globally . The "hundred-model battle" is giving way to an era of "agentic AI" — autonomous systems that execute real-world tasks in environments like logistics, manufacturing, and enterprise SaaS, focusing on cost efficiency and deep integration rather than headline-grabbing parameter counts .

Perhaps the most telling sign of this shift is where the profit is flowing. JPMorgan’s analysis states that the largest profit pool is not accruing to model-centric startups but to major internet conglomerates like Tencent and Alibaba that own the distribution platforms, cloud gateways, and enterprise channels .

Alex Yao's five competitive factors for consolidation

Based on a series of JPMorgan research reports and Yao's public commentary, five critical factors are separating survivors from casualties in China's AI consolidation :

1. Sustaining frontier-level model performance

Maintaining a globally leading model is no longer optional; it is a trust mechanism for enterprise clients. A model that slips from the frontier will struggle to retain customers in a market where global competitors are also racing forward. As of early 2026, U.S. and Chinese models have traded the lead multiple times, with the performance gap narrowing to just a few percentage points on key benchmarks .

2. Prioritizing task completion rate over token unit price

The brutal price war that defined the earlier phase is subsiding because customers have learned that a cheap token is worthless if it doesn't accomplish a job. In the new intelligent agent paradigm, JPMorgan argues the "task completion rate" — whether the AI reliably finishes a complex, multi-step assignment — is far more decisive for customer retention than the cost per token .

3. The profitability litmus test: gross profit growth vs. R&D spending

Forget vanity metrics like user registrations. The core test of sustainable monetization is whether a company's gross profit growth consistently outpaces its research and development expenditure. JPMorgan has identified this discipline as the fundamental measure of a healthy AI business model in the current environment .

4. Owning distribution and enterprise channels

Distribution is destiny in the new landscape. Companies that control the customer relationship — whether through dominant cloud platforms, mobile super-apps, or deeply embedded enterprise software — capture the bulk of AI revenue. This is why JPMorgan sees companies that own those channels as the ultimate winners of the consolidation, rather than model builders that rely on third-party distribution .

5. Chip supply resilience

Despite well-documented U.S. export restrictions on advanced semiconductors, Yao has stated that near-term AI chip shortages are not expected to create substantial barriers for China's leading internet firms. Domestic hyperscalers are increasingly procuring domestic alternatives, such as Baidu's Kunlun AI chips, to sustain the robust demand for AI computing and cloud services .

The new phase: agentic AI and industrial execution

The industry's pivot extends beyond software monetization. AI is now moving from consumer screens onto factory floors, a shift that is beginning to reshape industrial value chains. Manufacturers are deploying real-time algorithms that coordinate production and optimize assembly lines, capturing more value beyond traditional low-margin assembly . Chinese startups, operating under tighter hardware constraints than their U.S. counterparts, are optimizing for efficiency and targeted deployment — finding momentum not in raw scale, but in reliably delivering useful work in specific industrial scenarios .

China's AI ecosystem in 2026 is a fully integrated stack of foundation models, multimodal systems, and enterprise platforms. The question for investors, as Yao frames it, is no longer "Is China competitive in AI?" It's "who has the structural advantages, and how are they converting them into durable commercial gains?" . The hundred-model war was a gold rush; the consolidation phase is a disciplined, enterprise-driven industry focused on delivering measurable business value.