China AI trading: Uncertain "who will be the ultimate winner," betting on beneficiaries of the "AI application battle"
```
Author: Long Yue
Source: Hard AI
As the "AI application battle" unfolds, JPMorgan Chase chooses to avoid betting on a single winner and instead focuses on earlier realization and higher visibility in second-order beneficiary fields.
On January 30, the JPMorgan China Securities Research team released the China Internet Industry Research Report: "2026 Outlook: Trading AI Second-order Beneficiaries." The report covers several directions, including Chinese consumer AI, enterprise AI, agent business, and gaming. Its core judgment is not "who will win" but "which sectors will benefit first and are easier to be traded by the market."
No rush to judge "application winners," the market is not yet ready for pricing
JPMorgan makes it clear in the report:
“It is still too early to trade on ‘AI application winners and losers’ in the Chinese chatbot market at this stage, as trends are still brewing. Short-term changes in market share reflect distribution and product iteration pace, not a solid moat.”
In its view, early user metrics and market share shifts are insufficient to support judgments about structural outcomes. The report emphasizes:
“In our view, a feasible investment approach at this stage is to avoid making black-or-white bets on winners at the application layer. Instead, target beneficiaries in the industry’s advancement of AI interaction, whose upside depends more on adoption breadth than on the market share of a single application.”
2026 may be the "activation year" as chatbots enter high-frequency usage
On consumer AI, the report defines 2026 as a key milestone:
“We believe 2026 could become China’s consumer AI ‘activation year,’ as chatbots progress from trial use to entrenched habits and begin reshaping how users discover information, evaluate options, and initiate actions.”
JPMorgan references specific data, noting the significance of ByteDance’s “Doubao” user scale:
“ByteDance’s Doubao daily active users reached about 100 million, marking a milestone as it indicates that general chat interaction has crossed the ‘good enough’ threshold for mass-market applications.”
The report places this trend in a global context, citing Reuters:
“ChatGPT has 800 million weekly active users, which underscores just how fast consumer AI can compound its growth once the product offers utility and distribution scales.””
JPMorgan believes chatbots are evolving into “high-frequency traffic gateways at the upstream of numerous AI use cases,” which will directly change the way platforms compete.
Second-order trading logic: token, marketing, and productivity
While avoiding bets on the final winners, JPMorgan clearly anchors its trading logic to second-order exposure:
“Instead, we anchor our operational recommendations on second-order exposure: 1) AI infrastructure and cloud providers, who can capture the rising inference workloads; 2) advertising platforms, which benefit from the increase in industry sales and marketing intensity.”
Specifically:
- In AI infrastructure (overweight: Alibaba, Baidu): The report argues that “the growth of daily multi-turn chatbot conversations will directly benefit inference demand and Token throughput.” As agents expand from dialogue to execution, token consumption will compound, rewarding scalable infrastructure and ecosystem integrators along the value chain.
- In online advertising platforms (overweight: Kuaishou Technology): Major platforms will vastly increase customer acquisition expenditure to capture AI mindshare. JPMorgan expects “performance-oriented channels to command a significant share of the budget.” Even if Kuaishou is not ultimately the chatbot winner, it stands to benefit from the overall rise in industry marketing intensity.
“We expect large platforms to increase spending on customer acquisition and repeat engagement as they compete for chatbot mindshare.”
- In online gaming (overweight: Tencent, NetEase): AI compresses the "creativity to market" cycle, improving R&D ROI. Tencent has released an open-source 3D generation model that quickly converts text/images to high-quality 3D visuals. This structural advantage in content supply is underestimated by the market.
- JPMorgan also notes that performance-oriented channels will be main beneficiaries.
Enterprise AI: From "experimentation" to "measurable returns"
Regarding enterprise AI, the report stresses the tipping point is measurable returns in high-stakes workflows:
“We believe that the inflection point for enterprise AI adoption occurs when the value proposition becomes measurable within high-stakes workflows.”
JPMorgan cites several public disclosures as examples, including:
“Over a quarter of Google’s new code is first generated by AI, then reviewed and approved by engineers.”
“Developers using GitHub Copilot complete standardized coding tasks 55.8% faster than control groups.”
In the Chinese market, the report references Tencent’s disclosure:
“Its AI coding tool CodeBuddy reduced coding time by 40% and increased development efficiency by 16%.”
JPMorgan urges caution regarding enterprise-disclosed data but sees this as an important indicator:
“It shows that vendors are now actively marketing AI value propositions in operational language (savings, efficiency)—often a precursor to broader procurement and adoption.”
Agent business: the ‘conversation-to-execution’ transition layer
JPMorgan sees agents as the core next-step transition layer. Alibaba’s "Qianwen" application now enables ordering food, booking travel, and making payments directly via chat interfaces.
The report states: “Chat interfaces can reduce friction between discovery, comparison, and checkout.”
However, JPMorgan remains cautious:
“Given the early stage and limited disclosures, we avoid premature conclusions about ‘winners and losers.’”
The report notes short-term opportunities are more reflected in:
“Funnel efficiency, rather than immediate market share disruption.”
While short-term market share disruption is unlikely, first-order advantages lie in increasing the value of each incremental user session. For vertical leaders like Ctrip, the risk lies in traffic entry migrating upstream, but they can safeguard high-frequency use cases by embedding agent layers into their own apps.
Valuation reshaping: Discounts compared to global peers
Looking back at 2025, China Internet stocks were mainly driven by "narrative shifts" rather than earnings forecast adjustments.
The report points out that although there has been some rebound recently, dynamic P/E ratios for China’s leaders are still significantly discounted compared to global peers.
- Valuation comparison: Alibaba (2026 calendar year expected PE 20x) vs. Google (29x) and Amazon (26x); Pinduoduo (8x) vs. eBay (17x).
- Key view: JPMorgan believes that as AI narratives converge, “narrowing valuation gaps on a case-by-case basis appears achievable.” This favors companies with strong long-term narrative credibility and real execution milestones.
This article is from WeChat Official Account “Hard AI”. For more cutting-edge AI information, please visit here
Risk Reminder and DisclaimerThe market contains risk, so investment requires caution. This article does not constitute personal investment advice, nor does it take into account the unique investment objectives, financial circumstances, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this article is at the user's own risk.
```