Alibaba Group (BABA) announced their Q2 FY2026 earnings yesterday (26 November).
The 2 major highlights of this quarter’s earnings call were Alibaba Cloud’s AI business and Taobao Instashopping’s on-demand retail operations.
On one hand, AI and cloud performance exceeded expectations, with management noting that demand for cloud services remains very strong and even stating that “there is no AI bubble over the next three years”. On the other hand, the subsidy war in on-demand retail led to a 53% year-on-year decline in Alibaba’s net profit, and management cautioned that “profit fluctuations will continue in the short term”.
With these 2 forces offsetting each other, Alibaba’s stock opened slightly higher but trended down later, with no major overall movement.
Also, analyst Alex Yao from JP Morgan did not seem to ask Deepseek for answer this quarter.
We will share our views shortly, but first, here are the minutes we translated from the executives’ Chinese responses in the Q&A part of the earnings call. The text here might differ from the translation provided by officially appointed 3rd party interpreters during the earnings call itself. You can find the full earnings report on Alibaba Group’s investor relations web site.
Q&A Session
Q: Gary Yu – Morgan Stanley:
My question is about the cloud business. I’d like to understand the future growth outlook — is there still room for acceleration? Also, from a demand perspective, since China doesn’t have AI companies on the scale of those in the U.S., what are the main drivers of revenue from external customers?
A: Eddie Wu, Group CEO:
Based on what we’re seeing now, customer demand remains very strong. The pace at which we can bring Alibaba Cloud’s AI servers online is significantly lagging behind the growth in customer orders, and our backlog continues to expand. So, from the perspective of current data demand, the future growth potential is still in an accelerating phase.
On the demand side, we’re seeing rapid growth across all aspects of enterprise applications. Among the many companies we support — whether in product R&D, the manufacturing process, or in daily customer usage after a product goes live — AI demand continues to penetrate deeply. As a result, whether enterprises are training models, running inference, or enabling their end users to access AI-driven products, all of these rely on cloud computing. We’re therefore seeing substantial, real paid demand from enterprise customers, with significant potential that is still growing. This gives us strong confidence in future AI-related demand.
Q: Kenneth Fong – UBS:
Congratulations on the company’s performance in on-demand retail. Could management share key recent developments in on-demand retail and how it is creating synergies with the core e-commerce business? Based on these developments, what are your expectations for CMR and EBITDA in the December quarter? Thank you.
A: Jiang Fan, Ecommerce CEO:
In the past few months, we’ve focused on improving the unit economics while maintaining market share, and we believe we’ve made very meaningful progress. On one hand, our order mix has improved; on the other, scale effects have driven a notable reduction in logistics costs. Since October, Taobao Instashopping’s unit-economics losses have been cut in half compared with July and August. On top of this, its order share has remained stable, its GMV share has continued to rise, and it has meaningfully boosted related physical e-commerce categories. Let me elaborate.
First, order-mix optimisation. Over the past two months, the proportion of high-ticket orders on the platform has increased. According to the latest data, non–tea & beverage orders now account for over 75%. Average order value has grown by double digits compared with August. This rise in order value has also contributed to Instashopping’s overall GMV share expansion.
Second, as order share has grown, Taobao Instashopping is benefiting from logistics scale effects. We are seeing faster delivery times compared with the same period last year, and logistics cost per order has fallen significantly. Our per-order logistics cost is now notably lower than before Instashopping entered its heavy-investment phase. Driven by these two factors, we have achieved our short-term goal of halving per-order losses versus July and August. At the same time, during the UE-improvement process, user retention and purchase frequency have both been better than expected.
Beyond food delivery, we are also seeing rapid growth in retail categories. Instashopping has meaningfully driven growth in related categories and businesses, especially in physical retail categories such as grocery, health, and supermarket goods. For example, Taobao Instashopping orders for Freshippo and Tmall Supermarket have risen 30% month-on-month versus August. In recent months, we have also actively encouraged brands to join Taobao Instashopping, and going forward we will accelerate the synergy and integration between these categories and Instashopping.
Overall, we firmly believe that Instashopping has enormous synergy potential with the Alibaba ecosystem. We have already completed the first stage of rapid scale expansion. In the second stage, our UE improvements are tracking as expected, laying the foundation for long-term sustainable development of the on-demand retail business and strengthening our conviction to invest for the long term. Next, we will continue to refine the user experience, focus on high-value user operations, and prioritise retail-category development. Taobao Instashopping is one of the core strategic upgrades of the Taotian platform. Our goal is to drive RMB-trillion-level GMV within three years and lift overall category market share.
A: Toby Xu, CFO:
Let me address your second question regarding CMR and EBITDA fluctuations. As Jiang Fan just mentioned, Taobao Instashopping has had a clear positive impact on user engagement and on driving GMV in related categories, so it will also have a positive contribution to CMR. Going forward, our main focus will be strengthening collaboration between near-field and longer-distance retail and ensuring this synergy is fully realised.
Regarding the EBITDA of the China e-commerce group, Instashopping investment reached a peak this quarter (September quarter). As efficiency improves — including significant UE improvements and stabilisation of scale — we expect overall investment to contract meaningfully next quarter. Of course, we will dynamically adjust our investment strategy depending on the competitive landscape.
Customer-management revenue (CMR) in e-commerce is primarily affected by payment-processing fees and the base of site-wide ad-tech spending. Since we began charging payment-processing fees in September last year, we expect CMR growth to moderate next quarter. As we have consistently emphasised, our primary objective is mid-to-long-term market share. In pursuing this, we will continue to invest firmly in consumers and merchant growth and continue driving the business-model upgrade of our e-commerce platforms. During this process, CMR and EBITDA will naturally experience short-term fluctuations.
Q: Alex Yao – JP Morgan:
As Jiang Fan mentioned earlier, our first-stage investment cycle has ended, and we are now entering the second stage — the efficiency-optimisation stage. My question is: as efficiency improvements generate cost savings, how do we think about allocating these benefits across key stakeholders in the value chain? For example, will consumer subsidies remain at their current intensity, which would imply a more gradual financial improvement? Similarly, how will we distribute cost savings among consumers, merchants, and the platform in terms of subsidy levels?
A related question: If we do not reduce consumer subsidies and continue down our current path — improving user mix, increasing order share, and raising the proportion of high-value orders — how much further room is there for UE improvement?
A: Jiang Fan:
Let me take this question. I actually touched on part of this earlier. Our UE improvement in recent months has come from two areas. First, higher average order value, which increases our per-order revenue because our revenue is correlated with AOV. Second, as I mentioned, our logistics efficiency has improved significantly with scale.
I believe there is still substantial room ahead. On the consumer side, our growth over the past few months has been driven mainly by a large number of new users. As these users convert into higher-loyalty users on our platform, we will be able to lift AOV further and also adjust how we deploy subsidies. At the same time, traffic inside the Taobao app — including the Instashopping channel — has grown rapidly over the past few months. Instashopping has already become a channel with over 100 million DAUs, and there is considerable commercialisation potential there. This also represents a meaningful opportunity for future UE improvement.
Of course, the broader market remains highly competitive, so we will continue to assess opportunities and dynamically adjust our strategies based on competitive conditions.
Q: Ronald Keung – Goldman Sachs:
I’d like to ask about capital expenditures over the next three years. How should we interpret the previously mentioned RMB 380 billion? In particular, given that RMB 120 billion has already been spent over the past four quarters, how should we think about the incremental revenue generated from this CapEx? And how should we consider the ratio between CapEx and incremental revenue? Any outlook on incremental revenue contribution from these investments? Thank you.
A: Toby Xu:
The RMB 380 billion CapEx figure we mentioned earlier is essentially a three-year planning number. But based on what we’re seeing now — as I mentioned just moments ago — the pace at which we can bring new servers online is still far behind the growth in customer orders. So at the moment, whether it’s supply-chain constraints, data center build-out schedules, or the overall deployment pace, we are moving as fast as we can to meet customer demand. Under these circumstances, if we still cannot adequately meet demand, we do not rule out further increases in investment.
Of course, the broader market’s supply-chain dynamics are being affected by many factors, but overall, we continue to take a proactive investment stance toward AI infrastructure in order to meet customer needs. From a high-level perspective, the RMB 380 billion we previously referenced may actually prove to be on the low side — this is based on what we are seeing in current customer demand.
Regarding how CapEx translates into incremental revenue, and how to estimate that ratio, I think it is difficult to quantify at this stage because overall progress in the AI industry is still at a relatively early stage. Our AI infrastructure is being utilised in multiple ways, and several of these usage patterns are still evolving. For example, some of our infrastructure is directly leased to customers for training; some is leased for inference. We also use our own servers for inference on Bailian, as well as for internal applications such as Amap, Taobao, Qwen, and Quark — where these AI capabilities are being converted into membership services or membership products.
As a result, the revenue contribution and gross margin profile of our AI infrastructure varies depending on the usage scenario. So at this stage, the ratio between CapEx and incremental revenue is not very stable. Over the long term, what we focus on is the overall quality of the Tokens generated by our infrastructure and the cost-performance ratio of those Tokens.
Q: Ellie Jiang – Macquarie:
A follow-up question: As a full-stack AI service provider, we are clearly in a major investment cycle, with investments spanning multiple layers of the value chain. Given ongoing supply-chain volatility, how are we thinking about resource allocation? Specifically, at the model MaaS layer, how do we continue investing in foundational capabilities, while at the application layer we see very aggressive iteration in products like the Qwen, Amap, and others? And given the current environment, how do we assess the industry’s return on capital for AI investments related to training and inference? Thank you.
A: Eddie Wu:
Indeed, as a full-stack AI service provider, our products and infrastructure can be applied across many areas in this broader AI investment cycle. But internally, we do have a set of priorities, and I can share a few of them.
Our most important priority is ensuring the training of our base models. For AI infrastructure to attract more customers or unlock more high-value scenarios, we must continually improve the capabilities of our models. Only with continuous iteration can we unlock more users and more use cases. Once higher-value scenarios are unlocked, both the volume and the quality of Tokens consumed — as well as customers’ willingness to pay for these Tokens — will naturally increase. This is therefore a top priority.
Another priority is inference usage. We view inference services on our Bailian platform (百炼平台) as highly important. Because Bailian can serve global customers, the key is ensuring our AI resources are used efficiently around the clock — smoothing peaks and filling valleys — so that each AI server can run at near full load 24 hours a day to generate more Tokens. Scaling up the Bailian resource pool is, for us, also a high-priority area.
Beyond that, we consider the inference services for our own internal AI applications — and those of external customers — but these are lower in priority. And even among external customers, we apply a prioritisation framework. If a customer uses our services broadly — storage, big data, CPU resources — in addition to GPUs, then that customer will have a higher priority. If a customer only wants to lease GPUs for basic inference, then that demand will be relatively lower priority.
Your second question is also an important one. We look at it from two angles. From the demand side, we see two clear drivers. First, foundational models — whether base LLMs, video-generation models, or future all-modal models — are all improving rapidly, and the scaling-law trend has not plateaued. The industry has not hit a ceiling yet. We’ve also seen many breakthroughs from models such as Gemini 3. As model capabilities improve, AI can do more tasks and cover more scenarios. As task capabilities expand, AI penetration across industries will continue to increase.
With these two demand drivers, we see AI demand over the next three years as highly certain. And under such strong demand growth, on the supply side, as many of you analysts have likely observed, starting from the second half of this year, virtually every segment — fabs, DRAM, storage, CPUs — has been short of supply, with AI servers constrained across multiple components. This entire wave of shortages is driven by AI demand.
Looking ahead, the expansion cycles for these supply-chain players have significant bottlenecks. These expansion cycles will likely take two to three years. During that period, we expect demand growth to outpace supply increases. So for the next three years, we believe AI resources will remain in a structural undersupply situation.
In fact, what we’re seeing — both within our own operations and across major U.S. cloud providers — is that not only are the latest GPUs running at full capacity, but even GPUs from one, three, or five generations ago are fully utilised. So, in our view, for at least the next three years, the idea of an “AI bubble” is unlikely.
Q: Jialong Shi – Nomura:
In last quarter’s earnings call, management mentioned that Alibaba aims to gain greater market share in China’s consumer market. We’ve also seen that the company has increased investment in on-demand retail in recent months and gained share. Beyond on-demand retail, which other consumer-sector subsegments does management view as having strong potential and as areas where the company may increase investment in the future?
A: Jiang Fan:
Over the past many years, Alibaba has already entered a wide range of consumer-sector verticals. In addition to the significant investment we made this year in on-demand retail, we also have existing layouts in areas such as Freshippo, offline supermarket O2O models, Fliggy, and Amap’s local services. Our focus now is to better integrate these businesses and build stronger cross-segment synergies, so that we can enhance our overall market share in the broader consumer sector going forward.
[THE END]
Disclaimer:
This earnings call minutes was compiled by Momentum Works from the publicly available earnings call of Alibaba Group held on 26 November 2025. It is intended solely for informational and analytical purposes. The minutes may contain unintentional errors or omissions of some details due to audio quality, accents, and real-time interpretation during the call.
All spoken content remains the copyright and intellectual property of Alibaba Group. Please refer to the company’s official recording or transcript for complete accuracy and authoritative reference.
Any analysis, commentary, or opinions provided by Momentum Works are independent, based on our own research and perspectives, and do not represent the views of Alibaba Group or any other organisations mentioned.
You can also make reference to the following Momentum Works reports for more:
