Wall Street Sentiment on Chinese ADRs Shifts from "Hopeless" to Tactical Long
Miles Bennett
China's internet sector rallied hard in July, flipping offshore investor sentiment from June's 'helplessness' to tactical long. Bank of America flags three now-clearer themes: irreversible AI commoditization, easing ByteDance competition, and still-weak consumer spending heading into summer travel.
AI models are commoditizing fast — who actually makes money?
BofA's view: leading on benchmarks for a few weeks no longer justifies buying a single-model stock. Consensus has shifted to three winning vectors — multi-model ecosystems, AI-agent applications (AI programs that execute tasks autonomously), and on-device AI (models running locally on phones or PCs).
This means → capital is no longer betting on "whose model scores highest" but on who can turn a model into a product users will pay for.
Key upcoming releases: DeepSeek V4 Pro, Alibaba Qwen 4.0, Zhipu GLM 5.3, MiniMax M3.1. Among unlisted names, Zhipu is seen as high-quality but its hold on top-tier coding capability is questioned; MiniMax trails on model strength, but its $11 billion valuation appeals to some funds.
Why is Tencent seen as the core AI beneficiary?
After the formal July launch of Hunyuan HY3.0, market doubts about Tencent's foundation-model progress faded sharply.
Tencent's edge sits at both ends: its Work Assistant leads on AI-agent user adoption, while on-device AI adoption improves consumer-side unit economics. In plain terms = it owns the highest-traffic gateway (WeChat) and stands to cut costs as AI runs on users' own devices.
The risk is equally real: a mass rollout of WeChat AI agents would bring significant compute costs plus depreciation pressure, and some funds already project Tencent's 2027 earnings declining year-on-year.
ByteDance is pulling back — why do Meituan and JD win?
Channel checks show ByteDance's competitive pressure in local services and e-commerce is materially easing. That marginal shift is the direct driver of tactical longs on listed transaction platforms.
Meituan: with instant-retail competition cooling, unit economics (profit per order) are expected to improve steadily from H2 2026. JD: an improving EPS outlook plus stepped-up buybacks are drawing renewed interest.
The shared concern: will the cash saved from food-delivery loss reduction get re-deployed into AI and overseas expansion — ventures whose endgame remains unclear?
Alibaba is a consensus buy — so why isn't it a core holding anymore?
The decision to stop losses in instant retail is now consensus and should drive positive EPS revisions over the next few quarters.
Cloud expectations are sky-high near-term: the street broadly forecasts the June-quarter's year-on-year growth at 45% or above, with some funds projecting cloud revenue growth exceeding 50% through 2027.
This reflects a fundamental shift in how capital treats Alibaba — willing to buy, but as a tactical trade, not a long-term core position. Once the cloud and cost-cut catalysts are priced in, profit-taking is likely.
Summer spending is the next test — who gets hit first?
Trip.com: a brief bounce after its regulatory probe ended, but summer travel activity is running weak on soft consumer sentiment and adverse weather. The market fears another round of EPS cuts, and hotel take-rate pressure is mounting.
NetEase: a long-consensus pick during the June sell-off, but an early-July game launch underperformed expectations, raising near-term EPS concerns. Updates to *Justice Online* and a newly opened unlimited server in July are expected to provide near-term support.
In plain terms = the consumer sector is "bouncing without conviction" — funds won't add exposure until summer data offers a clearer signal.
What catalysts are the remaining names waiting for?
Baidu: fund attention has shifted entirely to the design capability, revenue scale, and potential spin-off timeline of its in-house AI chip effort.
Bilibili: low on the radar; funds are waiting for Tencent to complete its equity disposal.
Kuaishou: its "Kling" product faces sustained competition from ByteDance's equivalent, and after a recent fundraise there is no near-term catalyst — market interest is minimal. This means → all three currently lack a single event strong enough to reset fund expectations. Whether the tactical-long window extends ultimately depends on macro consumption delivering a positive signal in summer data.
Content is for reference only, not financial advice.