China's e-commerce sector just failed a very public stress test, and the sell side did not respond as a bloc. In the same 24-hour window in late June, Daiwa Capital Markets cut PDD Holdings and JD.com to Hold while leaving Alibaba at Buy. Three weeks later, Alibaba and Baidu shares jumped twice in eight days on completely different news: an AI product win, not a retail metric. The split is no longer about who sells more parcels during a promotional festival. It is about which of these companies investors now trust to monetize AI, and which are still fighting a subsidy war Beijing has explicitly told them to stop.

A downgrade that wasn't uniform

The trigger was China's 6.18 mid-year shopping festival, the second-largest retail event on the mainland calendar after Singles Day. Gross merchandise value across the event rose just 0.9% year-on-year, versus roughly 15% growth in the 2025 edition 1. Daito Research moved its sector call on Chinese e-commerce from positive to neutral on the back of that number, citing a difficult macro backdrop, tightening regulation, a scaled-back national trade-in subsidy program, and a high prior-year base 4.

Daiwa's response, delivered the following day through analyst John Choi, was not to mark the whole sector down evenly. PDD Holdings was cut to Hold from Buy, with the price target slashed to 80 dollars from 145 1. JD.com received the same rating cut, Hold from Buy, with the target reduced to 27 dollars from 47 and full-year earnings-per-share estimates trimmed 10 to 18% through fiscal 2028 on top of separate concerns about rising memory-chip costs squeezing its electronics business 2. Alibaba, by contrast, kept its Buy rating. Daiwa trimmed the price target modestly, to 175 dollars from 200, acknowledging the same weak 618 print but stopping well short of a downgrade 3. That asymmetry is the story: three companies exposed to the same disappointing data point, two of them cut and one of them spared.

Beijing's other message

The downgrades landed against a regulatory backdrop that helps explain the divergence. Two weeks before the 618 numbers came in, China's State Administration for Market Regulation summoned representatives from Alibaba's Taobao and Tmall units, JD.com, Pinduoduo, Douyin and RedNote over promotional practices during the festival run-up 5. Regulators flagged so-called 10-billion-yuan subsidy campaigns where actual spending fell short of what was advertised, along with non-transparent seller disclosures and business rules that let platforms sidestep liability in disputes 6. The companies were ordered to review and rectify their promotional rules, and officials used the session to push platforms explicitly toward innovation and service quality and away from price-war subsidies as the primary competitive lever.

That is a structural headwind for PDD in particular, whose entire growth model has run through aggressive subsidy spending, and a milder one for JD, whose electronics and appliance categories are now also facing margin pressure from elevated memory-chip input costs. Alibaba is not immune to the same scrutiny, but its growth narrative has already been shifting away from subsidy-funded gross merchandise value and toward cloud and AI, which gives analysts a different variable to underwrite.

The Apple catalyst

That shift showed up in the share prices almost immediately. On July 8, Alibaba surged 9% to around 106 dollars and Baidu gained 5%, with JD and PDD moving only in sympathy rather than on company-specific news 7. The catalyst was a combination of slowing losses in Alibaba's instant-commerce unit and continued strength in its Cloud Intelligence Group, where external revenue grew 38% year-on-year and AI-related products accounted for 30% of cloud revenue for an eleventh consecutive quarter of triple-digit AI growth 7. Separate chatter that DeepSeek and Zhipu are each developing proprietary AI chips added to the sense that China's AI infrastructure buildout has momentum independent of the consumer downturn.

Then, on July 16, both stocks moved again after Beijing cleared Apple's generative AI service for the Chinese market, with Alibaba's Qwen model and Baidu's Ernie model set to power Apple Intelligence features on iPhones sold in China 8. Alibaba's Hong Kong-listed shares rose roughly 4.8% and Baidu's gained close to 3.9% on the news. Unlike the July 8 move, this was not sentiment on a macro data point. It was confirmation that a major global technology company had chosen Chinese large language models for a mass-market consumer product, a validation neither Daiwa nor Daito had been pricing when they made their calls three weeks earlier.

What the next print has to prove

Alibaba's next earnings report, due August 28, is where this thesis gets tested against the numbers. Its fiscal fourth-quarter results already showed the trade-off analysts are underwriting: revenue of 35.28 billion dollars, slightly ahead of the 35.23 billion dollar consensus, but adjusted earnings per American depositary share of just 9 cents against an estimate of 1.12 dollars, as adjusted EBITA fell 84% to 740 million dollars on heavier AI and Taobao Instant Commerce spending. Investors have so far accepted that margin compression as the price of the cloud and AI growth Daiwa is underwriting. August 28 will show whether that growth is still accelerating, or whether the Apple deal was the peak of the good news cycle.

The investor takeaway

Treating Chinese e-commerce as one basket is now the wrong frame. PDD and JD remain exposed to a subsidy model Beijing is actively discouraging and to a consumer that grew merchandise spending under 1% in the year's biggest sales event. Alibaba and Baidu have decoupled from that story and are increasingly being priced as AI infrastructure plays that happen to also run a retail business. The risk for the second group is that August 28 forces a reckoning between the multiple the market is now paying for AI optionality and the profit numbers actually showing up in the filing.