Alibaba Group Holding Limited · Q2 2026 call

This quarter, Alibaba Cloud's external revenue growth accelerated to 45%, a 22-quarter high, while adjusted EBITDA margin reached 11.6%. Notably, this 45% growth was broad-based, driven by compute, storage, model as a service (MaaS), and AI applications. We proactively scaled back low-margin business, continuing to improve the quality of our growth.

In quick commerce, we continued to narrow losses substantially while growing business scale by 45%, with unit economics improving quarter over quarter. Having crossed the AI commercialization inflection point last quarter, we're now seeing growth accelerate and margins expand this quarter.

To realize synergies across our commerce platforms and strengthen our full-stack AI capabilities, we have implemented strategic realignment of certain businesses in our financial reporting.

Free cash flow was an outflow of RMB 44.7 billion, compared to an outflow of RMB 18.8 billion in the same quarter last year. The decrease was mainly attributed to the investment in cloud infrastructure.

In addition, AliExpress achieved operating profit this quarter.

Based on average gross margins today, roughly we can break even on AI-related capex in three years and of course average gross margin continues to rise and we expect to be able to shorten that payback period, say to 2.5 years.

an A100 purchased in 2020 or a V100 purchased in 2018 even today are still running at full capacity.

As we ramp up deployment of our own proprietary chips in our data centers, as they account for an increasing proportion of total chips and replace commercially procured chips, we can expect to see substantially higher gross margin as well as profitability.

At our current level of gross margin for AI products and under the assumption of a three-year payback period on capex, theoretically keeping our growth rate below 33% would already enable positive cash flow. However, that is not our strategic choice at this time.

As our product gross margin improves and our proprietary chip substitution rate increases, our payback period will shorten to two and a half years or even less. And so under those circumstances, while pursuing growth of over 40%, we'll also be able to maintain positive cash flow.

We expect the transaction volume of Quick Commerce for non-food categories to surpass that of food categories within the next fiscal year

We've observed that AI-related products generated 12.4 billion RMB in revenue this quarter, and if we convert that into an annualized US dollar figure, that works out to US$7.3 billion in annual revenue. Looking ahead to the next quarter, our own forecast is that that same annualized revenue for AI next quarter will approach US$10 billion.

This exponential growth in demand for inference has marked a fundamental shift in the model whereby compute has now become the core asset driving AI revenue, and today all AI-related revenue models are centered on AI compute.

For the past couple of years a lot of people have asked, what is the super app for AI? The answer is that the real super application is compute—cloud-based AI compute—because all of these different workloads need to run on a full stack of AI cloud compute, including training, inferencing, AI software and agents requiring GPUs, CPUs, storage, databases, virtualization, as well as tools among others.

As a result we're highly confident in our ability to achieve our goal of 100 billion in external cloud revenue by 2030, and we have good visibility into achieving gross margin of 20%.

growth in Bailian's MaaS business is very rapid, and as of August we reached 16 billion RMB—or surpassed 16 billion RMB—in ARR. Given the current growth momentum as well as the pipeline of new models slated for launch, we remain confident that we will achieve our year-end target of 30 billion ARR by the end of the year.

in my personal view, I think that the current monetization model for large language models through APIs is just a short-term approach, a short-term transitional approach, and it's certainly not the ultimate business model.

the last generation of T-Head chips, we've already manufactured over 500,000 of them and shipped, and then the latest generation in August has already been deployed on Alibaba's AI Cloud as super nodes. And I think we're one of the only companies that's able to deploy such proprietary chips, domestic chips, at scale.