Alibaba's (BABA) Qwen model has passed 3 billion downloads, beating some well-known US models. This is Alibaba's biggest AI development.
Open weight models and Qwen's reach
A couple of weeks ago, Jensen Huang of Nvidia (NVDA) put out an open letter backing open weight models and asking regulators not to rule them out. Many people signed on, and Qwen sits at the center of this push. Apple (AAPL) is now letting Qwen into its Siri system in the Chinese domestic market. Tesla (TSLA) is in talks to test Qwen tools.
Pricing pressure on Google and Meta
Qwen's demand means heavy pricing pressure for Meta (META) and Google (GOOGL). A recent study found a 600x gap between the cheapest and most expensive ways to run generative AI - the priciest frontier models versus the cheapest open weight models, some of them versions of Qwen. Not every model can do every task. As companies watch AI costs closely, more of them will move toward low cost, and this is where Alibaba and Qwen can lead. They are already taking market share.
Baidu: held to a tougher standard
Baidu (BIDU) gets punished even when it beats estimates because it is judged against a different, tougher standard. It keeps investing heavily in AI, which is good, and should post solid numbers. But it is measured like a traditional business, so the focus is margins versus AI spending. Alibaba gets more room here because investors are making a longer-term bet on it than on Baidu.
One point on Baidu that is being missed: its Apollo Go robotaxi service will roll out in London soon. The negative mood around self-driving cars means this is not getting the coverage it deserves. Investors should look closely at what Baidu is doing in autonomy. Do not underestimate the mid-term value that autonomy and robotaxis can add to these companies.
Ad revenue falling as AI spending rises
At the big-picture level, the global story has been that heavy AI investment is fine because ad revenue funds it. Now the reverse is showing up: as ad revenues fall, AI investment rises. That makes investors more nervous, since the spending looks riskier.
Alibaba: misses have not stopped the stock
Alibaba missed estimates four quarters in a row, yet the stock kept climbing. Usually a miss resets expectations, but a miss now is unlikely to reset much given the Qwen story. The real story is Qwen's ability to break into global markets with its models. As long as investment in AI infrastructure is still seen as the right move, Alibaba should keep getting a pass in the near term.
JD and the wider group
JD (JD) posted a retail operating margin of 4.6%, a record for a peak promotional quarter. Among the three - Baidu, Alibaba, and JD - JD is the most traditional business, focused on logistics, and has already reported. Alibaba is the long-term AI play and is well placed to take market share from North American leaders. Baidu is the wild card because of its tougher standard, but is a solid investment going forward on the strength of autonomy. PDD (PDD), the parent of Temu, sits in a separate category.
