China Internet sector: Has bad news instead become good news?

Overview

This report is Bernstein's periodic view update on the China internet industry, focusing on valuation levels after the sector's pullback in 1H 2026, the debate over AI capex and inference costs, AI progress at Tencent and Alibaba, competition in consumption and instant retail, and tactical allocation opportunities among core covered companies. The tone is more constructive than before: macro consumption and regulatory concerns have not disappeared, but investor expectations and valuations have been significantly cut, improving the sector's risk-reward.

Core views

The core views are: first, sector valuations have fallen back close to the lows of 2022-2023, and much of the negative expectation has already been priced in; second, the long-term ROIC of AI capex will remain debated, but the view that platforms need to subsidize AI inference costs without limit is seen as overly pessimistic; third, Tencent's AI path across Hy3, Xiaowei, and the WeChat ecosystem is becoming clearer, while Alibaba Cloud growth and segment-level restructuring have improved the Alibaba narrative; fourth, NetEase, JD, and Boss Zhipin are attractive on the back of low valuations, buybacks, earnings growth, or business data support; fifth, although Meituan and PDD may see tactical rebounds, competition, capital returns, and management communication still constrain medium- to long-term re-rating.

Analysis framework

The report combines company engagement, feedback from North America investor roadshows, high-frequency industry data, an AI token consumption framework, capex relative to cash flow capacity, valuation multiples, FCF/EV yields, net cash as a percentage of market cap, earnings forecast revisions, and event catalyst paths to conduct both top-down and stock-level tactical assessment of the China internet sector.

Methodology notes

• Valuation methodsForward P/E and FCF/EV yield

  Forward P/E and free cash flow yield

  The report uses 2026E and 2027E P/E as well as FCF/EV yield to assess whether the sector is close to historical bottoms, and notes that FCF/EV helped identify bottoms in 2021-2023.

• AI economicsInference cost sensitivity

  Sensitivity of inference cost to transaction volume and GMV

  The report distinguishes between ordinary chat and agentic transactions in terms of token consumption, arguing that Tencent's token consumption will rise parabolically only if agentic transaction volume and GMV expand significantly.

• FundamentalsEstimate revision analysis

  Earnings forecast revisions

  The report updates forecasts for Tencent, Alibaba, and JD, using revenue, margin, Non-GAAP operating profit or EBITA, and EPS versus market consensus to judge changes in fundamentals.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Tencent / 700.HK

  Core long idea, rated O, target price HKD780

  Strengths

  Valuation is seen as low; the release of Hy3, Xiaowei testing, and AI scenarios in the WeChat ecosystem are improving market perception; cash flow has strong capacity to absorb capex; Delta Force and PC gaming growth provide support.

  Weaknesses

  Short-term operating profit revisions may not have fully bottomed yet, gaming lacks an obvious next blockbuster, and AI investment may still create expense pressure.

  Comparison

  The report argues Tencent should not remain at 11-12x forward PE over the long term, and could return to the 17-18x forward PE range if concerns ease.

  Risks

  Hy4 or Xiaowei feedback falls short of expectations, AI inference costs rise, gaming growth slows, or profit forecasts continue to be revised down.

• Alibaba / BABA / 9988.HK

  Core long idea, rated O, BABA target price USD180, 9988.HK target price HKD176

  Strengths

  Alibaba Cloud growth remains on track, and rising compute pricing plus the Apsara Conference may continue to improve AI sentiment; segment adjustments help reported growth in China e-commerce and cloud; CXMT and Kimi holdings have potential value.

  Weaknesses

  Overall forecasts were slightly cut, quick commerce and instant retail competition remains intense, and some business reclassification may be more of a narrative improvement than a genuine operating improvement.

  Comparison

  The report believes Alibaba should also return to a mid-teens PE range, but its valuation and earnings visibility are still affected by e-commerce competition and AI investment.

  Risks

  Cloud growth slows, AI capex returns disappoint, quick commerce losses re-emerge, or weak consumption drags on e-commerce.

• NetEase / NTES / 9999.HK

  Preferred name, rated O, NTES target price USD150, 9999.HK target price HKD235

  Strengths

  A low second-half revenue base, accelerating buybacks, expected Southbound Stock Connect inclusion, and follow-up news on Ananta form the catalyst path; historical backtests show a high correlation between accelerating buybacks and subsequent stock performance.

  Weaknesses

  After the launch of Sea of Remnants, the stock price may see a temporary pullback.

  Comparison

  Compared with other internet names, NetEase combines a higher FCF/EV yield with clearer support from buybacks.

  Risks

  New game performance disappoints, buyback pace slows, or competition in the gaming industry or changes in AI content production create pressure.

• JD / JD / 9618.HK

  Low-valuation long idea, rated O, JD target price USD40, 9618.HK target price HKD155

  Strengths

  A contrast between mid-single-digit forward PE and double-digit profit growth; narrowing food delivery losses, a profit-first strategy, and net cash levels provide support.

  Weaknesses

  Revenue growth forecasts were cut, and macro uncertainty still weighs on discretionary consumption and retail growth.

  Comparison

  The report groups JD together with Boss Zhipin as names that are too cheaply valued and whose earnings growth is underestimated.

  Risks

  Weak consumption recovery, food delivery or instant retail investment exceeds expectations, or revenue growth continues to slow.

• Boss Zhipin / BZ

  Low-valuation growth idea, rated O, target price USD18

  Strengths

  High-frequency data still show job postings growing by a mid-teens percentage; AI-improved matching efficiency helps lift ARPJ and enterprise ARPU; valuation is around 4x after excluding cash and investments.

  Weaknesses

  The business remains affected by the job market and corporate hiring demand.

  Comparison

  FCF/EV yield is close to 20%, standing out among covered companies for cash flow valuation attractiveness.

  Risks

  Weak macro employment, shrinking corporate hiring budgets, or AI matching improvements failing to translate into commercial uplift.

• Meituan / 3690.HK

  Neutral name, rated M, target price HKD85

  Strengths

  Food delivery losses are narrowing faster than expected, and 6-7x forward PE may already be close to e-commerce valuation lows.

  Weaknesses

  After the rebound, upside leaves the report authors with a mixed view; instant retail, food delivery, and e-commerce are seen as the same highly competitive and highly substitutable market.

  Comparison

  Compared with JD and Boss Zhipin, the report is more cautious on Meituan's risk-reward after the rebound.

  Risks

  AI investment continues to expand, overseas or Middle East business uncertainty, or lower order volume reduces losses but also weighs on growth.

• PDD / PDD

  Neutral name, rated M, target price USD110

  Strengths

  News of Shein's Hong Kong IPO may prompt investors to refocus on Temu valuation, and some investors are starting to probe positioning again.

  Weaknesses

  Management's continued lack of communication with investors and unclear willingness to return capital affect how long-term investors discount the value of cash flow.

  Comparison

  Although net cash exceeds 40% of market cap, compared with JD and Boss Zhipin, PDD has not yet demonstrated an equally clear buyback or capital return path.

  Risks

  Temu valuation controversy, intensifying competition, lack of capital return, and insufficient management communication causing the valuation discount to persist.

Key data

• Year-to-date performance of covered stocksAverage decline of 16%As of the recent rebound discussed in the report, China internet covered stocks were still down 16% on average versus the start of the year.
• Capex intensity of Tencent and AlibabaTencent capex is 10.9% of LTM revenue; Alibaba is 12.3%The report believes Tencent has greater room to increase capex relative to gross profit and operating cash flow.
• Alibaba Cloud growthJune-quarter revenue growth accelerated to the mid-40% rangeA low-teens margin implies 35%-40% incremental cash margin, and Alibaba indicated that compute pricing may continue to rise in the second half.
• AI token consumptionChat typically uses hundreds of tokens; agentic transactions use tens of thousands of tokensBased on this, the report believes inference costs will only rise materially when agentic transactions and GMV grow at scale.
• Catering consumption shareIn Q2 2026, catering revenue accounted for 12.9% of retail consumptionThis is close to the long-term average of 13%, indicating that competition in instant retail and catering continues in a high-penetration environment.
• Sector valuationAverage 2026E P/E of covered stocks is 14.3x, and 2027E P/E is 12.1x2027E P/E is only slightly above the lows of 2022-2023.
• FCF/EV yieldTencent and JD about 5%, NetEase about 10%, Boss Zhipin close to 20%The report treats this as an important indicator for assessing valuation bottoms and the attractiveness of cash returns.
• Net cash as a percentage of market capMore than 40% for companies such as JD, Boss Zhipin, and PDDJD and Boss Zhipin management have already used part of the cash through buybacks, while PDD's capital return remains an unresolved issue.

Impact & implications

The report's investment implication is relatively positive: with low valuations, low expectations, and the AI narrative shifting from cost concerns toward product and commercialization possibilities, the sector has opportunities for valuation recovery and event-driven rebounds over the coming months. It is more suitable to focus on leading names and undervalued targets with clear catalysts, buybacks, or support from earnings revisions, while still watching for weak macro consumption, intensifying competition, and uncertainty over AI investment returns.

Risks

• China consumption remains weak; although the report maintains a 2.5% retail consumption SAAR growth trend line, the recovery in demand is not yet firm.
• Regulatory and macro policy concerns still exist, and whether policy easing will emerge remains to be seen.
• The long-term ROIC of AI capex is uncertain; if agentic transactions and GMV grow rapidly, inference costs could rise significantly.
• E-commerce, food delivery, and instant retail have increasingly converged into the same competitive market, and price subsidies plus fulfillment competition may compress profits.
• There is downside risk to forecasts for Tencent gaming growth, Alibaba Cloud growth, NetEase new game performance, and JD revenue growth.
• Unclear capital return at PDD and insufficient management communication may continue to weigh on long-term valuation.

What to watch

• User feedback and the commercialization path of Tencent Xiaowei rollout.
• The timing of Tencent Hy4 preview and improvements in model capabilities, especially whether it can further unify the AI path of the Hy and Weixin teams.
• Alibaba Apsara Conference, cloud revenue growth, rising compute pricing, and T-head-related arrangements.
• Progress of CXMT IPO and revaluation of Alibaba's stake value.
• The extent of the post-launch stock pullback for NetEase Sea of Remnants, follow-up Ananta news, and progress on Southbound Stock Connect inclusion.
• The impact of JD food delivery losses, R&D investment, and the profit-first strategy on earnings forecasts.
• Boss Zhipin hiring high-frequency data, ARPJ, and enterprise ARPU improvement.
• Whether NBS macro data imply a less restrictive macro policy environment.