Alibaba Group Holding Limited (BABA) Earnings Call Transcript
August 29, 2025
Earnings Call Speaker Segments
Hello, everybody, and thanks for joining us on Alibaba Group's earnings call for the June Quarter of 2025. Joining me are Chairman Joe Tsai; our Chief Executive Officer, Eddie Wu; Toby Xu, who serves as Chief Financial Officer; and Jiang Fan, Chief Executive Officer of the Alibaba E-commerce Business Group. This session is being webcast as well, accessible through the IR section on our corporate site. Later today, a replay will be made available on our website. I'll now briefly go over the safe harbor. Remarks made today could include forward-looking statements, especially those regarding our business and financial results, which carry risks and uncertainties that might lead actual outcomes to differ materially from what those forward-looking statements suggest. We ask that you review the safe harbor statements included in the press release and investor presentation we provided today. Be aware that some financial measures discussed on this call are presented on a non-GAAP basis. Our GAAP results, along with reconciliations between GAAP and non-GAAP measures, appear in our earnings press release. I'll now hand things over to Eddie.
Hello, everyone, and thanks for joining us on today's quarterly earnings call. During this period, we achieved healthy expansion. If you set aside the revenue contributed by Sun Art and Intime, our overall top line increased 10% compared to the same period last year on a like-for-like basis. The growth in revenue from our core operations stayed robust. Customer management revenue generated by our China e-commerce operations climbed 10% versus a year earlier. For Cloud Intelligence Group, revenue growth sped up to 26% compared with the prior-year period, while sales from AI-related products kept up triple-digit expansion for a eighth quarter in a row. AIDC's revenue expanded 19% from a year ago. Within AI + Cloud, faster development of AI applications alongside rising customer uptake of AI products fueled a 26% increase in revenue from external customers versus the same period last year. Over this quarter, AI-related sales represented more than 20% of the revenue coming from external customers, as demand for AI kept expanding quickly. We're additionally observing how AI applications are generating excellent growth momentum for traditional offerings, among them compute and storage. SAP and Alibaba formed a strategic partnership centered on cloud and AI. Serving as SAP's worldwide cloud computing partner, Alibaba Cloud is going to help SAP's customers operate and administer their core software systems on Alibaba's platform. Building on the models we introduced in Q1, SAP is going to provide AI transformation offerings to its enterprise customers as well. This partnership shows that major companies in the SAP ecosystem — among the biggest in the world — recognize what our cloud infrastructure and AI capabilities can do. We've continued advancing what our AI foundation model is capable of. Come July, Alibaba introduced an improved Qwen3, which includes a nonthinking version, a reasoning version, and a version for AI coding — each considered a leading performer globally in its own category. Notably, adoption of Qwen among users overseas got a fast lift from our Qwen3 coder model. On top of that, we released several models as open source, including Wan2.2, which generates video, and Qwen-Image, which turns text into images. We keep making our open source models better, which lets customers create their own AI apps. Meanwhile, the AI native applications that Alibaba develops are also advancing. Amap completed a comprehensive AI overhaul, launching Amap 2025 — the first location-based app built natively on AI anywhere globally. With this update, spatial intelligence reaches dynamic real-world settings, and Amap is positioned well to become a fresh entry point for lifestyle services in the future. DingTalk, in the same way, has wrapped up its latest AI enhancement, creating the first agent-powered work feeds worldwide to explore workplace application models of the next generation. Over on Taobao, we see huge opportunities driven by AI emerging, such as AI search and an AI advertising platform. On the consumer side, we brought Taobao and Tmall Group together with Ele.me and Fliggy under one umbrella — Alibaba China E-commerce Group. Through this reorganization, we've built a broad platform for consumption and lifted the quality of our consumer experience; supply chains, user bases and membership perks have all been unified across our businesses, and we rolled out a tiered loyalty program that connects Ele.me, Fliggy and AMAP. With these newly combined benefits, members enjoy a better experience spanning every kind of consumption scenario. Starting in May, the money we put into quick commerce blew past major milestones at a fast clip and started producing synergies. In August, monthly active consumers for our quick commerce business approached 300 million, and that helped push monthly active consumers for the Taobao app up 25%. Daily order records kept falling in our China E-commerce Group. Looking ahead, two once-in-a-generation openings stand before Alibaba Group: one is a tech foundation centered on AI + Cloud, the other is a full-spectrum platform spanning shopping along with everyday life services consumption. Capturing these openings demands hefty, wide-ranging spending from us. For a company now 26 years old, this marks the start of a new entrepreneurial era; so in February we told the market we'd commit RMB 380 billion across the coming 3 years toward cloud and AI infrastructure. Then in July, we laid out a plan to channel RMB 50 billion into consumption. When you combine the way AI is reshaping every sector with tight coupling between AI and the cloud, the result will be the technology industry's single greatest opportunity in the next ten years. As for Alibaba, we hold the fourth biggest cloud infrastructure globally—number one in Asia—and we also have technology capabilities across the full stack, spanning AI compute power, AI cloud platforms, AI models, an open source ecosystem, and AI applications. Our CapEx spending on AI and cloud infrastructure during this quarter reached RMB 38.6 billion. Over the last 4 quarters combined, the total we've put into AI infrastructure along with AI product R&D exceeds RMB 100 billion. The money we've put into AI is now delivering concrete outcomes. You can see that in how Alibaba Cloud has swung back to fast expansion, driven by demand for AI, and in the AI-powered features we offer across consumer and enterprise use cases. As a result, a clearer route is taking shape for AI to fuel strong growth at Alibaba. On top of that, we hold a strong position in China, which is the biggest e-commerce market globally and the most promising one for service consumption. China has mature e-commerce infrastructure, dense population concentrations, and robust appetite for service consumption, giving us a firm base to merge our quick commerce operation with the Taobao app. We think this coming together will serve consumers' need for a single-stop consumption journey while also fulfilling merchants' wish to reach consumers across many scenarios. This will boost how efficiently commerce operates and set the stage for a single AI assistant handling all consumption needs. When it comes to quick commerce, Alibaba's strategic placement carries ambitions that go past merely fighting it out within one category. Our intention is to serve the one-stop consumption requirements of our 1 billion consumers and to shape business models for a comprehensive consumption platform in the AI era. Regarding consumption, our long-term objective is building a comprehensive consumption platform that serves our 1 billion consumers across the full spectrum of shopping and daily life needs; we aim to deliver the best experience to the largest consumer base with the highest purchase frequency, ultimately leading in a RMB 30 trillion addressable market. In the coming three years, Alibaba is setting out on a fresh path, approaching it like a startup, aiming to fuel strong business expansion by making ongoing commitments focused on the key domains of consumption along with AI + Cloud. We firmly hold the view that directing resources into the core of our operations will sharpen our competitive edge and propel growth well into the future.
Appreciate that, Eddie. As Eddie was saying, we're entering a new chapter of entrepreneurial activity, directing capital toward 2 strategic pillars of consumption in AI + Cloud. Those pillars represent the 2 biggest long-term possibilities we're going after in a disciplined manner. To reflect this tighter focus, we've also changed the way we report our finances. Starting this quarter, we completed a strategic merger of Taobao and Tmall Group, Ele.me and Fliggy under Alibaba China E-commerce Group, transforming our value proposition into a full consumption platform. This is not just an organizational adjustment; it's a major strategic bet designed to reshape the consumer experience and unlock long-term value across our ecosystem. During this quarter, we made substantial headway on that objective by increasing our commitment to quick commerce, which is becoming a vital application for seizing fresh demand and molding tomorrow's consumer journey. While our quick commerce operation achieved important benchmarks, it also contributed to a 25% increase compared with the same period last year in Taobao app monthly active consumers throughout the initial 3 weeks of August. Simultaneously, we are building out the AI + Cloud infrastructure meant to support the coming phase of technological change, establishing Alibaba as a pivotal facilitator of enterprise AI uptake across sectors. Growth in our cloud unit sped up, with segment revenue and revenue from external customers each climbing 26%, driven by a surge in AI demand and more customers embracing public cloud services for their AI workloads. At the same time, we remain committed to improving operating efficiency and profitability; AIDC moved meaningfully closer to breakeven this quarter while maintaining robust growth momentum. Turning to our consolidated financial results, total revenue reached RMB 247.7 billion. Had we excluded revenue from Sun Art and Intime, like-for-like revenue would have increased 10% year-over-year. Adjusted EBITDA dropped 14%. The main reason is that we've been strategically investing in scaling quick commerce — we want to capture new consumption patterns and open up future monetization opportunities. That decline was partly offset by better margins in a number of businesses, including AIDC and other units, which kept making progress on operating efficiency. GAAP net income came in 76% higher, largely driven by mark-to-market changes on our equity investments, plus a gain from divesting Trendyol's local consumer service business. Operating cash flow reached RMB 20.7 billion. Free cash flow was negative RMB 18.8 billion, an outflow. The main driver here was that we moved faster on building out AI and Cloud infrastructure, which pushed CapEx up to roughly RMB 39 billion, along with what we're putting into Taobao Instant Commerce. With close to USD 50 billion in net cash behind us, a balance sheet that's both healthy and lightly leveraged, and solid access to capital markets, we've got plenty of room to keep funding long-term strategic bets without weakening our financial footing. During this quarter, under our share repurchase program, we repurchased about 7 million ADSs, spending USD 815 million in total. Our commitment to returning value to shareholders remains intact, and we do that through buybacks, dividends, and funding growth. How quickly we hand back that capital will continue to flex based on market conditions and where our strategic priorities sit. Moving on to segment performance, starting with Alibaba China E-commerce Group. Revenue for that group came to RMB 140.1 billion, marking a 10% increase. Within our e-commerce business, customer management revenue also climbed 10%, driven mostly by an improved take rate. The June 18 shopping festival performed well, delivering healthy consumer growth on the Taobao app, thanks to our launch of simple promotional tools and greater support for merchants providing high-quality products and service. The count of 88VIP members — our big-spending consumer cohort — kept climbing at a double-digit rate compared to a year earlier, going past 53 million. Our quick commerce business saw revenue climb 12%, mostly on higher order volume stemming from the Taobao Instant Commerce rollout in late April. From the time it went live, we've witnessed promising business traction, showing robust user uptake and building order momentum. At the same time, we broadened our product selection and expanded front warehouse coverage into nonfood categories as part of our push to better the user experience and lift operating efficiency. We carried out our strategy to create synergies linking quick commerce with the broader Alibaba ecosystem by tapping supply chain, users, and membership perks across our operations; in August, the Taobao app introduced a tiered loyalty program tying together Alibaba Group's China e-commerce, quick commerce, and travel platforms. Adjusted EBITDA for the Alibaba China e-commerce group fell 21%. If you set aside the investments in our quick commerce business, EBITDA for our Alibaba China e-commerce group has risen compared to a year ago. AIDC revenue jumped 19%, mainly on the back of strong cross-border business performance. AIDC's adjusted EBITDA loss shrank considerably, getting close to breakeven. We continue to improve how efficiently we run things, and both our chosen UE and the Trendyol International operation showed clear quarter-over-quarter gains. Looking ahead, we're committed to making our operations and investments more efficient. As a result, we expect our profitability to continue improving. Revenue in the Cloud segment rose 26%, driven largely by gains in public cloud revenue. AI revenue maintained its triple-digit growth as demand for AI keeps rising fast, and we're also seeing greater demand for compute, storage, and other public cloud offerings that enable AI adoption. On an adjusted EBITA basis, our margin was essentially unchanged from the prior year, coming in at 8.8%. We intend to continue directing capital toward expanding our customer base and advancing technology, which includes AI-related offerings and services, so that we can accelerate cloud uptake tied to AI and maintain our position as market leader. As previously communicated, we have changed how we report our segments to more accurately reflect our strategic priorities. Our financial reporting structure has been simplified by placing Cainiao, Amap, and Hujing DME within the all other category. The all other segment saw a 28% decline in revenue, driven mainly by the divestment of Sun Art and Intime. On an adjusted EBITDA basis, all remaining operations posted a loss of RMB 1.4 billion. The primary driver here was increased spending on our technology ventures, while improved performance at places like Freshippo helped counterbalance that to some degree. Our "all other" category consists of a collection of innovative efforts, including a number of strategic AI-focused technology infrastructure projects and businesses. We continue to pursue efficiency improvements across all of our business lines, while simultaneously directing funds toward AI opportunities to maintain our competitive position and drive future expansion. In summary, this quarter marked substantial headway in our strategic investments across 2 pillars that will propel Alibaba's next stage of growth: our comprehensive consumption platform and AI + Cloud infrastructure. Regarding commerce, by uniting several operations within the Alibaba China E-commerce Group, we're achieving greater synergy in supply chains, user networks, and membership offerings, which enables us to address shifting consumer requirements more effectively and seize long-term expansion prospects. Turning to cloud, revenue acceleration occurred because of strong demand powered by AI. As we expand infrastructure capacity, we're assisting additional customers in deploying and scaling their AI workloads. With clear focus and firm commitment, we're putting resources into the two historic opportunities before us—AI and domestic consumption. The RMB 380 billion we've earmarked for technology reflects a long-term ambition: creating the infrastructure that AI proliferation depends on. At the same time, our targeted quick commerce build-out is designed to unlock fresh demand and China's long-term consumption potential. Thanks to a solid balance sheet, operating cash flow and business momentum, we're positioned well to fund these bets, deliver sustainable growth and reinforce the core capabilities that will shape what Alibaba becomes. Thank you. That wraps up our prepared remarks, and we can now take your questions.
Hello, everybody. On this call today, feel free to pose your questions in either Chinese or English. During the Q&A portion, consecutive interpretation will be supplied by a translator from a third party. Kindly be aware that this translation exists purely for convenience. Should any inconsistency arise, what our management stated in the original language shall take precedence. Should you find that the Chinese translation isn't audible, a bilingual transcript of today's call will be posted to our website no later than 1 week following the meeting's conclusion. <<foreign_lang>> Operator, kindly begin the Q&A session. Thanks.
Nice work on the strong cloud revenue increase. My question concerns the step-up investment you made recently in quick commerce, plus the food delivery operation. Could management tell us what vision you hold for the quick commerce growth chance in China? Also, what is your plan for investing in quick commerce? For how long will this heavy spending continue? And in what way will that spending deliver long-term value for Taobao overall and for your China E-commerce platform? Could management also share some of the newest progress of < >, the quick commerce business? Which synergies have you achieved up to now? And what should we anticipate regarding how the investment affects our GMV as well as CMR growth across upcoming quarters?
<<foreign_lang>> < > Thank you for asking. I'll start by going over certain progress achieved to date in our Instant Commerce operation, after which I'll give our expectations for Instant Commerce going forward. It has been 4 months since we introduced Taobao Instant Commerce, and we consider our engagement of users and merchants, our building of logistics capabilities, and our marketing to have been very successful. Especially starting in July, order volume growth, user scale, merchant supplies, and delivery capacity have all surpassed what we expected. Actually, looking only at the food delivery to home category, we already hold the leading market position where orders are concerned. Allow me to give several figures. To start, our peak daily order volume hit 120 million, while weekly average daily orders came to 80 million during August. Second, user scale. Quick commerce monthly active consumers, MAC, hit 300 million during August, which is 200% growth versus before April. Third, merchant supply. This fast expansion in business scale has likewise drawn new merchants to join Taobao Instant Commerce. High-quality merchant supply, above all, has climbed to industry-leading levels. Fourth, fulfillment capacity. Daily active riders have gone beyond 2 million; that represents a 3x rise from April. And this further indicates that we have generated over 1 million new jobs. As I mentioned last quarter, the first-stage goals for our quick commerce business involved scaling up user growth and building consumer mind share; given these developments across the past few months, we have already met our first stage goals beyond our own expectations. Now let me discuss the synergies linking quick commerce and our e-commerce business. To begin, quick commerce is a major driver of overall user scale and engagement. During August, quick commerce generated 20% growth in the Taobao app's DAUs. The high-frequency nature of quick commerce has additionally led to a significant rise in average purchase days per user. As user engagement rises, we observe a clear trend whereby quick commerce drives incremental income for our e-commerce business. In particular, first, growth in traffic drives advertising and CMR growth. And then the second point is that because users are engaging more, we're bringing in new users step by step and winning back existing ones, we're able to cut down on sales and marketing spending. We anticipate this pattern will persist and grow across our current operations, and we're sure it will generate meaningful additional profit on the e-commerce front. Now I'd like to offer my perspective on operating efficiency and unit economics, UE, for quick commerce, since I'm aware these are areas you focus on in particular. To begin with, when discussing operating efficiency, scale is something you simply can't ignore. Previously, we had only 1/3 the scale of our competitors. And across numerous cities and provinces, our market share came in below 20%. When the market share gap is that enormous, discussing efficiency makes no sense. Today, though, our quick commerce operation holds the scale lead, and we're able to boost our operating efficiency swiftly. Our competitors have delivered outstanding results in food delivery, particularly when it comes to efficiency, and we're now working hard to close that distance. The room we have to enhance efficiency is considerable. Let's examine this over the short term first, then over the long term. Over the short term, our losses will shrink, mainly because of the following: the first is optimizing our customer mix. We ramped up marketing spend for user growth across the country during the last 4 months and brought in a great many new customers. Bringing in new users demands a big upfront outlay, yet we're excellent at keeping users around; and as repeat customers make up a larger share, UE will get better. The second is optimizing our order mix; at the next stage, we're aiming to raise the share of high-value orders, which includes high-value meal orders, and to grow the share of orders in nonfood categories. UE will thus get better because average order value, AOV, will be higher. And the third is optimizing fulfillment efficiency along with costs. Early on, when order volume grew 4x, our top priority as a platform was making sure the user experience held up. So in July and August, we put in a further sizable investment to fix the serious near-term shortage of delivery capacity. Later on, once order scale levels off, our logistics costs will fall considerably, which will help UE improve further. So over the short term, we anticipate that while we keep investing in consumer benefit through better logistics and subsidy efficiency plus order structure optimization, our UE losses can be cut in half. Over the long term, as order density rises, there's a great deal of room to optimize our logistics costs relative to 4 months ago. Beyond that, there's also a great deal of room to sharpen how we target engagement with off-line merchants. From where I sit, scale is the leading factor behind reaching efficiency. Given the scale and market share we've just attained, we're confident we can reach efficiency that leads the industry over the long term. Simultaneously, we won't focus on the stand-alone profitability of quick commerce delivery. When you add in the growing advantages for our e-commerce operations, our view is that quick commerce is going to deliver durable, positive economic value across the whole platform, and we can do that without losing our edge on pricing. Now I'd like to turn to how the nonfood category is developing; within quick commerce, nonfood deliveries fall into 2 segments. The first is the established hyperlocal quick commerce e-commerce model, while the second is e-commerce — a hybrid approach combining quick commerce with e-commerce. On the hyperlocal front, we've drawn on our ample supplies and robust supply chain to build out a lightening warehouse model. Supply in these warehouses has grown quickly across the last few months. Today we operate more than 50,000 lightning warehouses, with orders rising over 360% year-on-year, and 25% of the supply held in those lightning warehouses originates from supply chains inside the Alibaba ecosystem. Second, we gain fulfillment capacity from front warehouses that Freshippo built for its fresh groceries category. Once Freshippo's supply was connected into Taobao Instant Commerce, order volume surpassed 2 million, an increase of 70% year-on-year. Turning to the hybrid model, Tmall Supermarket is moving from a conventional B2C fulfillment approach toward a quick commerce one. That keeps its pricing competitive while delivering far quicker shipping. We're also bringing Tmall brands' off-line stores onto Taobao Instant Commerce at a steady pace, which lets those brands run unified online-offline O2O operations. Our expectation is that as many as 1 million branded offline stores will eventually join Taobao Instant Commerce. Bringing Tmall together with Taobao Instant Commerce will unlock fresh growth drivers for brands and give shoppers a new retail experience. Broadly speaking, we anticipate that Taobao Instant Commerce and quick commerce will contribute RMB 1 trillion in annualized incremental GMV to the platform over the coming 3 years; we likewise think the food delivery market's move away from a single dominant player toward multi-platform competition hands merchants and consumers greater choice, a positive for the industry over the long term. As this transition unfolds, we as a platform are putting real financial resources behind it and investing so as to create more than 1 million direct jobs, push industry transformation forward, and spur consumption along with the wider economy. Thank you.
Thomas, thanks for the kind words about the quarter. Let's talk about Alibaba Cloud. That segment had a great showing, with growth speeding up to 26% compared to a year ago in this period. So first, what's the right way to read that pickup? Do you expect it to keep going in future quarters, and how does that shape what we should expect for FY '26? If I look at this faster growth next to what international competitors are doing, how do we weigh the speed of monetization against what's happening in the U.S.? On a different note, cloud margin came in at 8.8%. What's the right way to think about where margins go from here? And since this acceleration is so striking, can you walk through how the different industry verticals performed in this quarter? I remember we talked about a number of industries last quarter, traditional ones included, that were also starting to take advantage of cloud opportunities. Is there anything new we're noticing now, or meaningfully more advancement, during this quarter? And on the CapEx side, what's the right way to frame our CapEx guidance, considering the CapEx we actually realized over this quarter.
<<foreign_lang>> < > All right. I appreciate those questions a great deal. I jotted them down as you asked, and I noticed there were a fair number of them. I'll start with the earliest one, which concerned how we see our growth rate going forward; among our customers, we certainly are seeing an extremely obvious pattern around their adoption of AI products. They are developing AI offerings, and that points to demand that is very robust. First off, AI models are continually getting more capable, which means brand-new AI applications keep emerging and are being put to work across an expanding range of use cases. At the same time, a lot of vendors whose use cases used to involve running workloads on traditional CPUs are now beginning to leverage large models and apply AI to carry out those same original functions. Given all of that, demand is extremely strong and continues to grow at a robust pace; during recent quarters, inference workloads have grown quickly, while training and inferencing have both kept expanding. Beyond that swift growth in inferencing, we've also observed a few fresh industry trends on the training side over the last few quarters. On top of those big foundation models that the market keeps releasing upgrades for, there's a lot of training activity happening too, and that opens up fresh possibilities. Take auto manufacturers, for example, or businesses operating in education, or those working with multimedia applications. Each of them needs to run training. To serve their own needs, they build proprietary models using data they own exclusively. That pushes up how much our AI infrastructure gets used overall. So it gives infrastructure demand a really solid lift. We're also seeing additional openings on the training side, where organizations take advantage of our open source models, which creates substantial training needs for them as well. Consider the education sector as an example, as well as organizations in health care, and also companies creating development tools and platform businesses. With these proprietary models—where they leverage their own data and their ownsystemsto address their own use cases through post training or fine-tuning on our Q1 models—we're observing numerous strong opportunities. Building on that, they're utilizing our Alibaba cloud computing platform; and in the same way, starting from that base, we can gradually develop commercialized offerings—post training services—for our open source models. Your second question was about gross margin and how we compare ourselves to what's observed overseas. Let me share my perspective on China's cloud sector as a whole. In my view, the trend points toward this market becoming more concentrated than those abroad. The reason is that Chinese developers, regardless of what they're building, require complete, full-stack capabilities — and that's exactly what we deliver, with a comprehensive stack that reaches across conventional cloud computing, compute and storage, AI compute, and AI models. On top of that, we've built an open ecosystem that's extremely friendly to developers. So what we're trying to do is stay ahead of the market's typical pace of growth and pick up more share at the same time. On the strategic side, our top focus is bringing in more users and pushing into additional use cases, rather than trying to improve gross margin right away. Then, on your question about CapEx — what we spend on AI — we're sticking with the three-year plan to put RMB 380 billion into cloud and AI. That said, depending on how the supply chain looks in any given quarter, you could see some ups and downs from one quarter to the next. In the same way, because policies affecting AI chips and supply keep changing, we've got contingency arrangements in place that let us collaborate with a range of partners and handle varying supply chain circumstances. Therefore, no matter what kind of industry shifts come up, I have confidence that we'll keep going according to plan and as anticipated, proceeding with that RMB 380 billion CapEx commitment we've laid out.
Considering the cross-selling we've already witnessed in our food delivery segment, are there plans to increase the in-store portion of local services — that is, < > — since we've observed Ele.me stepping up promotions for in-store coupons in certain regions? So how ought we to view any additional investment or expansion in this in-store side of the business over the coming months, on top of food delivery?
<<foreign_lang>> < > Yes, thank you. I'll address this question. First, what I'd note is that this ties very closely to the scale we've reached in quick commerce, which I just mentioned to you. At present we have enormous scale of users on our quick commerce channel each day. Essentially, that means 150 million active users daily; and among those users, some visit stores themselves for self-pickup. That also covers group purchase with coupons. So when it comes to meeting our users' needs, especially those who head offline to stores, plenty of synergy exists there; and therefore we'll also look at offering these users more varied services; and as a matter of fact, we're already running some tests and pilots in certain selected cities.
My first question concerns how fast investments will move ahead. You spoke of your pledge to the 3-year plan of putting RMB 380 billion into cloud and AI. I'd like to hear as well about what you intend for investments ahead on the commerce side. Regarding consumption, how will the tempo of those investments look, and besides putting money into quick commerce, what other investments are you considering, say in supply chains or in users? And how will the tempo of those investments look? Then second, I want to ask about CMR. We saw CMR grow strongly this quarter, rising 10% year-on-year. Yet we know that as we enter September, the beneficial effect of the software service fee that you began applying last year will fade. So going forward, how much benefit will QZT penetration bring to CMR growth? And beyond that, the added traffic and GMV that quick commerce is generating — how much benefit will that bring to CMR?
Thank you. I'll address this question; since there's a once-in-a-generation opportunity to put money into the consumer market today, we are of course seizing it. Yet our investments in the consumption market aren't only beginning now. We have continuously invested on the user side. Absolutely — the Taobao and Tmall Group has already been directing funds toward users through supply chain initiatives, and that applies not just to Taobao and Tmall Group but also to Freshippo and our other operations, where we're likewise investing in supply chains. So when we talk about this substantial outlay we're currently committing to the consumer marketplace — RMB 50 billion aimed at building out quick commerce — that figure represents additional spending. It comes on top of the investments we were already making; we'll set the pace and rhythm of that spending according to market conditions and how the market evolves, making sure it's justified. Thank you. Now, your second question had to do with CMR. I think what really mattered here was how the take rate played in. As we mentioned, CMR growth this quarter came in positive and strong, and the main driver behind that was a significant increase in take rate. There were essentially two big factors behind it. The first was the 0.6% software service fee. The second is QZT, which runs on AI and keeps reaching deeper penetration. And I'd say that going forward, across the coming quarters, we should also see these areas contributing favorably to CMR. Quick commerce is also going to help CMR, since we're growing users and getting them to come back more often, and that pushes the take rate up. So over the next two quarters or so, CMR should grow pretty quickly, much like what you're seeing right now.
We've been tracking your model releases quite closely, and from what we can observe in the Qwen3 block, it signals a shift away from an era centered on training toward one driven by agents — an agent-focused period. So my question: during this transition, what extra capabilities, resources, or investments are required? And could you share a bit about your most recent work on agent products and agent applications?
I'll handle that one. Yes, we genuinely observe this overall evolutionary trajectory for models, moving from basic chatbots in the direction of agents. Several trends stand out to us in the agent era. First, without question, models need bigger context windows so they can handle more complicated tasks involving lengthier chains of thought. They must be capable of leveraging various tools or sequences of tools, and agents must be able to reach various systems — invoking various internal systems inside enterprises. As a result, these shifts open up fresh possibilities for us in our role as an infrastructure provider. A large number of these agents are going to require virtual servers, or they'll need to keep numerous browser windows open at the same time, or they'll need plenty of mobile virtual machines plus sandbox environments. So for a cloud vendor such as Alibaba Cloud, all of this is highly encouraging. We serve many clients across many sectors, and we can deliver all of it through a new offering named AgentBay, which provides a sandbox environment made specifically for agents. So this really is an agent-powered age for AI and cloud development, and from our standpoint, it's a fantastic opportunity. Let me build on that by sharing additional thinking on models as they connect to agent-driven AI. In my view, coding abilities are going to matter enormously, since models with robust coding abilities can link up with a wide range of tools and a wide range of enterprise internal systems; that way, they'll manage to address many tasks inside the enterprise or even tackle some more advanced tasks on the consumer side. One more thing worth adding regarding agent, agentic products: we have a fair number of offerings that plug into that and perform extremely well inside the Alibaba ecosystem. Take, for instance, many enterprises — particularly e-commerce ones — that require a client service product from Alibaba Cloud, beyond just compute and infrastructure being supplied. We can also create synergies with Taobao, with DingTalk, with Amap, and even with Alipay so that we can deliver additional categories of automated solutions at the business level; via these tools, such enterprises can build AI agents that handle these assorted tasks inside the enterprise more effectively.
I want to ask about quick commerce, or Instant Commerce, since this isn't the first attempt by Alibaba to get into this market. Going back to when Ele.me was acquired in 2018, a great deal of time, energy, and resources went into growing our share of that space, yet it appears that across a span of 2 to 3 years, that initial strategic goal was never truly achieved. So I asked DeepSeek why Ele.me, while under Alibaba, still failed to defeat Meituan and become larger than Meituan. What DeepSeek said to me was that the offspring of a wealthy trader cannot overcome the enterprising wolf brought up in a poor household; regardless, whether DeepSeek is correct or not doesn't matter. Still, I'm certain we've done plenty of reflection and thought as we head into this fresh round of battle. So I'm certain you'll adopt some different strategies or approaches to maximize success. Could you just tell us what's different on this occasion? And which different tactics are we using?
Well, I suppose you could go ask DeepSeek once more, but I'll try my hand at answering. It's true that we invested. We bought Ele.me and have kept putting money into it for many years. Still, I'd start by noting that Ele.me has in fact achieved a great deal of progress across these years. That might not show up in market share, since market share ties to our investment, our strategy, and traffic; even so, I'd say that during these years, Ele.me has improved enormously when it comes to infrastructure development and capabilities development. Were that not so, it simply would not have been feasible for Taobao Instant Commerce to reach such swift development within such a brief stretch of time; we couldn't begin at 0 and build this business from nothing and within merely 2 to 4 months be capable of serving 120 million orders — and not merely 120 million orders, but doing that while providing an outstanding user experience. I believe all the credit for that belongs to Ele.me and the experience and capabilities we accumulated after that acquisition. To succeed in this business, you must have sufficient merchants. You must have sufficient fulfillment capacity, or delivery capacity, and you must have sufficient consumers — users on the consumer side. If any one of those elements is missing, then your investment efficiency will end up poor; so as far as where we stand today, after integrating Ele.me into the Taobao app, we possess a huge number of users on Taobao, and those are highly active users. We also possess the merchant base that has been accumulated on Ele.me. Beyond that, we've got the fulfillment infrastructure and the logistics network already in place. That's what this investment is built on. The second point concerns how we approach the investment rationale, since our view isn't limited to quick commerce as a standalone operation; we're focused on its potential, the net positive contribution it could deliver across our entire e-commerce operation over the near term, the mid term, and the long term. So our thinking and rationale here differ from what they were previously. Naturally, plenty of effort remains ahead of us if we're going to nail this business and hit all of those objectives.
I want to ask about return on invested capital. We're about to put nearly RMB 50 billion into quick commerce, likely within a couple of months, so how should we view the ROIC on that spending? My thinking is this: if that same money went toward AI instead — maybe into something with a far larger TAM and much faster growth in cloud — we might see a notably stronger return. So going forward, how should we think about dividing capital internally between retail and AI?
Well, Eddie and I have already both said on this call that 2 enormous historic opportunities sit in front of us right now. One is AI, which we've covered. The other is consumption. The chance there is to shift consumption along the path from e-commerce to local — from hyperlocal e-commerce to instant commerce — and we have to seize both of these opportunities firmly. Each one matters enormously. So regarding what we're putting into both areas, naturally these are very large, very substantial commitments; in our view, they're historic investments. As I noted in my prepared remarks, whether you look at it through our capabilities or through our resources. When I say resources, that can mean our cash reserves, our cash flows, and resources throughout our whole balance sheet. We hold enough resources to carry out these very substantial investments, as Eddie put it, investments at scale across both of these huge opportunities. So what really matters for us is balancing near-term against longer-term returns and deciding where priority belongs; and we can examine that through the lens of the AI investments we've been making. I believe you've also clearly seen that those investments have already pushed up the growth rate of our cloud business. As was mentioned a moment ago, that applies not only to this quarter — we expect the next few quarters to show cloud growth rate increases too. Naturally, we'll also evaluate the business through EBITDA margin. Yet our top priority right now is directing money into these investments rather than generating profits — the profit rate is the lower of those 2 priorities; that doesn't mean we ignore profit rates or don't care about them, obviously. So that covers AI investments along with investments in quick commerce. As Jianfeng just noted, we aren't earning anything yet from those Instant Commerce investments, yet it's already quite clear that folding instant commerce into the Taobao app is bringing more traffic. It's bringing more frequency. That, in turn, is pushing up advertising on the platform too. So as we move from e-commerce into Instant Commerce, every one of these investments will produce solid returns down the line. And I have confidence that our resources are enough to let us do this while keeping our attention on what really matters most. One thing worth pointing out: we shouldn't fixate only on near-term returns and, in the process, miss the chance to capture those longer-term returns. So getting that balance right is essential.
Thanks to all of you for being here with us right now. We value the time you're giving us, and we're eager to talk with you once more before long.
Thanks. That brings today's call to a close. You're free to hang up at this point. Thanks again.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete Alibaba Group Holding Limited transcript - plus 132,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.
Get an API key View API docs →For developers and AI pipelines
Programmatic access to Alibaba Group Holding Limited earnings transcripts and 132,000+ others is available through the
EarningsAPI REST API and the hosted MCP server.
Quarterly plans from $162 - full transcripts, speaker segments, full-text search,
and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.