Home / Transcripts / Alibaba Group Holding Limited (BABA) · March 19, 2026

Alibaba Group Holding Limited (BABA) Earnings Call Transcript

March 19, 2026

New York Stock Exchange US Consumer Discretionary Broadline Retail earnings 75 min

Earnings Call Speaker Segments

Lydia Lu executive
#1

Thanks. Hello, everyone, and thanks for joining us for Alibaba Group's December Quarter 2025 Earnings Conference Call. With us on the call today are Chairman Joe Tsai; Eddie Wu, our Chief Executive Officer; Toby Xu, our Chief Financial Officer; and Jiang Fan, who serves as Chief Executive Officer of Alibaba E-commerce Business Group. I want to note that a webcast of this call is available on our corporate website. Later today, a replay will also be accessible on our website. Let me now briefly go over the safe harbor. Statements made during today's discussion may include forward-looking statements that reflect current expectations and assumptions and are subject to risks and uncertainties. Actual results could differ materially. Please see the safe harbor statements in our press release and investor presentation issued today. Keep in mind that some financial measures are presented on a non-GAAP basis. Today's earnings press release and investor presentation include our GAAP results as well as reconciliations between GAAP and non-GAAP measures. Unless otherwise stated, our remarks will be on a year-over-year basis; and with that, I'll hand the call over to Eddie.

Yongming Wu executive
#2

Thanks for joining us for the earnings call covering this quarter. During the last three months, we kept up robust investment momentum behind our two strategic priorities: AI plus cloud, and consumption. Revenue growth at Cloud Intelligence Group sped up to 36%, and our Quick Commerce business kept scaling up while unit economics improved steadily. As the AI agent era begins, the addressable market open to AI infrastructure providers such as Alibaba is poised to expand exponentially. AI models and our capabilities are quickly becoming embedded in mainstream work environments across every industry, with token consumption climbing sharply in all sectors. Cloud and software spending within enterprise IT budgets has traditionally represented only around 5% of company revenue. With model-driven agents beginning to handle everyday work tasks in many sectors, the market we can address is set to expand several times over. Alibaba has put together a comprehensive, full-stack AI capability set—spanning AI infrastructure all the way to the application layer—to serve the explosive rise in AI demand. Facing an industry shift and strategic opening of this magnitude, Alibaba Group is itself moving into a fresh chapter of entrepreneurial reinvention and future-focused critical investment. Let me now take you through the strategic road map for AI at Alibaba. Our capabilities in this area span the entire full stack. Chips, along with cloud computing, form the infrastructure layer for AI, while the application layer is led by Alibaba Token Hub and includes foundation models as well as MaaS across enterprise and consumer use cases. Put together, these give us end-to-end coverage of the whole stack, reaching from AI infrastructure all the way to applications. Considering how massive and enduring the growth momentum in the AI market is, and given that Alibaba is positioned across the full stack of the AI value chain, the business objective behind our AI strategy could not be clearer. Across the coming 5 years, we aim to exceed USD 100 billion in combined external revenue from cloud and AI, MaaS included. Now, on the infrastructure side of things. Because AI demand has stayed really strong, external customer revenue at Cloud Intelligence Group sped up to 35% in this quarter, and revenue from AI-related products posted triple-digit growth compared with a year earlier — that's now 10 quarters in a row. Cloud Intelligence Group's share of the market has climbed for 3 straight quarters, reaching 36%, and the gap between us and the rest keeps getting bigger. For fiscal year 2026, Alibaba Cloud's cumulative external revenue through February officially went past RMB 100 billion. Token consumption on the model studio platform rose 6x over the last 3 months, and we think MaaS will turn into the biggest revenue product for Cloud Intelligence Group. T-Head's own GPU chips are now in scaled mass production. By February 2026, T-Head had shipped a cumulative total of 470,000 AI chips. Looking at actual business rollouts via Alibaba Cloud, external customers are served by over 60% of T-Head's ships, and we have finished scaled adoption for those customers' AI workloads. AI workloads belonging to more than 400 enterprise customers spanning sectors such as Internet financial services and autonomous driving are now supported by T-Head. Our confidence is high that T-Head's capacity to supply compute will keep growing, delivering high-quality compute into our cloud infrastructure and MaaS platform, which boosts how competitive our cloud services are overall. Now to the application layer, as we call it, where the entire cycle of generating, distributing, and utilizing tokens sits at the center, we established a new entity named the Alibaba Token Hub Business Group, or ATH. The Tongyi Laboratory, the MaaS business line, the Qwen business unit, the Wukong business unit, and the AI Innovation business unit all sit inside it. Alibaba's AI strategy is carried out through this group, which functions as the organizational backbone, and it also serves as the hub enabling efficient coordination across our various AI businesses. Our newest generation large model, Qwen3.5-Plus, was brought out over Chinese New Year, and on a broad set of benchmarks spanning reasoning, coding, and agentic capabilities, it delivered excellent results. A notable gain in inference efficiency was shown by Qwen3.5-Plus thanks to foundational architectural innovation. Shortly, following on from Qwen3.5, the next generation of models tuned for coding and agentic use cases will be launched by us. On the consumer side, our models are so capable that Qwen now has more than 300 million monthly active users engaging with it directly. During the Chinese New Year period, we deepened how tightly it plugs into Alibaba's ecosystem by connecting the Qwen app to Taobao Instant Commerce, Alipay, Fliggy, Damai, and Amap. That linkage gives Qwen capabilities no one else has for everyday situations, making it China's first all-in-one personal AI assistant that covers life, work, and learning. More recently, we introduced Wukong, an enterprise AI agent platform. Wukong is the first AI native, enterprise grade agent platform anywhere in the world, and it enables companies to bring AI into their workflows while remaining consistent with each organization's data permissions and management processes. Regarding Alibaba's AI offerings designed for workplace environments, Wukong serves as the single, unified point of interaction, and business-to-business tools from companies across Alibaba's entire ecosystem will be integrated step by step so that Wukong can evolve into the top AI system for work. Turning to the other strategic area of focus for Alibaba — the consumption space — we continue to advance our key initiatives. During this quarter, our quick commerce operation expanded its scale even more, with share still climbing, customers staying highly loyal, and improvements quarter over quarter in both unit economics and average order value. At the same time, clear synergies emerged between quick commerce and e-commerce, and those pushed monthly active consumers on the Taobao app to double-digit growth compared with a year earlier. That concludes my comments. I'll hand it over to Toby now for the financial portion of the update.

Toby Xu executive
#3

Eddie, thanks. Our strategic priorities haven't changed: we're staying focused on AI plus cloud, along with consumption businesses. Momentum looks strong, with gains across technology, customer adoption, market share and user engagement. When it comes to AI plus cloud, we hold full-stack AI capabilities spanning all 3 core elements — model, cloud infrastructure and chips — and we lead in each through Qwen, Alibaba Cloud and T-Head; we also run China's most comprehensive consumer ecosystem, one that can monetize via AI. Qwen App's launch marked a major milestone, and it can unite our consumer applications. Turning to consumption, our quick commerce business kept gaining GMV market share in the December quarter, while unit economics and AOV kept improving as well. Let's now review the financial results. At the consolidated level, total revenue came to RMB 284.8 billion; stripping out revenue from Sun Art and Intime, revenue on a like-for-like basis rose by 9%. Total adjusted EBITA fell by 57%, mainly because of our strategic investments in technology-related innovation initiatives and on the consumption front, quick commerce included, partly offset by better operating results in cloud and stronger operating efficiencies across various businesses. GAAP net income reached RMB 15.6 billion, down 66%. Operating cash flow was an inflow of RMB 36 billion. Free cash flow stood at RMB 11.3 billion, down RMB 27.7 billion versus the same quarter last year. We're reinvesting our cash flow to lead in AI and quick commerce. As of December 31, 2025, we held USD 42.5 billion in net cash. Setting aside those with maturities beyond 5 years, our net position stands beyond the USD 60 billion mark. That balance sheet strength gives us the confidence to reinvest for long-term growth. Let's now turn to our consumption businesses. Revenue from China e-commerce group was RMB 159.3 billion, up 6%. Customer management revenue rose by 1%. The slowdown in revenue growth stemmed primarily from weaker transaction activities and the phase out of the impact of software service fee implementation. The Taobao App saw a double-digit increase in MAC during the quarter, driven by growing mind share and the increasing scale of our quick commerce business. Revenue from our quick commerce business climbed 56% to RMB 20.8 billion. During the quarter, we carried out our plan to further grow the scale of our quick commerce business, improved user experience, improved UE and increased AOV month-over-month during the quarter. Alibaba China E-commerce Group adjusted EBITA was RMB 34.6 billion, down 43%, primarily due to the investment in quick commerce, user experiences and technology. Looking ahead, this adjusted EBITA will keep fluctuating quarter-over-quarter due to intense competition and significant investment in user experience. Revenue from AIDC grew 4% this quarter. AIDC's adjusted EBITA loss narrowed significantly year-over-year, driven by a combination of logistics optimization and investment efficiency enhancement. The UE of the AliExpress Choice business also improved on a sequential basis. Moving on, I'd like to walk through the business updates and results for Cloud Intelligence Group. Our Cloud Business turned in yet another quarter of accelerating growth. Revenue from external customers was up 35%, compared with 29% in the prior quarter. AI-related products remain the key driver behind this momentum. We achieved our tenth quarter in a row of triple-digit growth in AI revenue. Its portion of external cloud revenue keeps climbing. That clearly mirrors both the scale and the acceleration of our AI business. The adjusted EBITA margin held roughly steady at 9%. We'll keep investing in customer growth and technology innovation so as to boost adoption of AI cloud infrastructure and reinforce our market leadership. Revenue for all other segments fell 25% to RMB 67.3 billion, largely because of the disposal of the Sun Art and Intime businesses, along with a decline in revenue from Cainiao, partially offset by higher revenue from Freshippo and Alibaba Health. Adjusted EBITA for all others was a loss of RMB 9.8 billion, mainly driven by greater investment in technology businesses, including Qwen models and consumer-facing Qwen, partly offset by better results at Cainiao, Hujing DME and other businesses. The Qwen Model has grown into one of the most widely adopted open source model families worldwide, passing 1 billion cumulative downloads on Hugging Face by the end of this January, while consumer-facing Qwen has exceeded 300 million MAU across platforms, which strengthens user engagement and expands long-term monetization potential. We have been ramping up investments on these technology fronts, including the Spring Festival campaign. Building on the strong momentum and results we've achieved, as Eddie mentioned earlier, we will keep investing substantially in Qwen models and the Qwen App. Our unallocated adjusted EBITA was a loss of RMB 2.7 billion, versus a loss of RMB 0.2 billion in the same quarter last year, reflecting costs tied to talent retention incentives from the one-off replacement awards plan of Ele.me. Thank you. We'll now open it up for Q&A.

Lydia Lu executive
#4

Hello, everybody. Feel free to pose your questions in either Chinese or English. Consecutive interpretation will be handled by a translator from a third party. Should any inconsistency arise, what our management said in the original language takes precedence. If the Chinese rendering isn't audible to you, a bilingual transcript of this call will be posted to our website no later than 1 week following the meeting's conclusion. <<foreign_lang>> Operator, kindly proceed with the first question. Thanks.

Robin Zhu analyst
#5

Can you walk us through some concrete instances of how Token Hub is going to alter the way the various cloud and AI businesses collaborate moving forward, from an organizational perspective? And on the strategic side, which changes or objectives do you expect this new structure to deliver that would be an improvement over how things were arranged before? Also, could management lay out how it ranks priorities across cloud and AI — is it share of market and top-line expansion, like the goal you just announced, or is it having the strongest first-party model capabilities, or is it consumer-side traction with customers adopting agentic AI, or something different?

Yongming Wu executive
#6

<<foreign_lang>> <<interpreted>> Thanks a lot for that question. My view is that the aim and the purpose behind creating the ATH Business Group ties very closely to the era we find ourselves in as 2025 wraps up and as we move through the opening months of 2026. When it comes to how AI is developing, we have now entered the agent-driven phase of AI development. That differs from the earlier stage of AI development — during the agentic AI era, what we must accomplish is an extremely tight fusion of the model with the application. During the earlier AI era, much of the data used to train models was static, yet in the genic era, we have to strengthen the integration linking models and applications, achieving tight integration, and a great deal of the data is now arriving from the customer side. So when you examine the various layers that AI deployment involves, spanning application, model, the AI infrastructure, and down through chips, what I believe is most distinctive and most essential regarding the agentic AI era is the requirement to accomplish this tight integration linking application and model. That stands as the critical priority. Now let me turn to the interconnection and the synergies linking the various businesses with respect to ATH. If you examine where this industry's trends are heading, we believe we perceive those trends quite clearly. AI agents are going to be tightly integrated alongside the application layer. And there is going to be a great multitude of highly diverse applications. On the consumer or 2C side, we are aggressively building out the Qwen app as a personal assistant for individuals. And on the 2B side, we are positioning Wukong as a 2B assistant. At the AI application layer, there is going to be a great multitude of distinct industry and vertically specialized applications serving different industry use cases. And every bit of that must be underpinned by a highly robust model as a service, MaaS layer. So MaaS, naturally, supports our own internal applications along with a great multitude of external and industry-specific use cases that leverage AI. So within this context, we see enormous value we are able to deliver and a massive total addressable market, or TAM. So looking ahead, we view the AI application layer as the primary channel through which tokens will be distributed. And the stronger the model capabilities you are able to offer at the MaaS layer, the more attractive and compelling all these different offerings become for customers. That is the business logic we have set out within this new business unit. So from the standpoint of both the model and the application layers, our number one priority is unquestionably to develop the most intelligent models. And I truly must stress that only once you possess the most powerful models are you genuinely able to drive the deployment of AI applications across every kind of different industry. Only with the strongest models are you able to attract applications from across diverse industries to adopt our MaaS offering. Still, to build the most robust models, you must collaborate very closely with various industries and with our own 2C and 2B applications in order to connect with our MaaS two applications across every kind of different industry and use case. So we need to get more users to leverage and make use of our models so that we can gradually benefit from the data flywheel effect. Only that way can we continuously enhance our models' capabilities. So that is among the reasons we established the ATH business unit at this moment. So in summary, what I would say is that our number one priority is unquestionably enhancing model capabilities. Still, enhancing model capabilities demands concerted efforts across the whole model pipeline as well as on the application and infrastructure side so that sustained improvements are achieved over the long term.

Joyce Ju analyst
#7

Congratulations on the strong headway you've achieved in cloud and AI. What I'd like to ask is this: we're observing CMR expansion decelerating markedly during the December quarter, in light of macroeconomic headwinds. China's online retail sales, as we've seen, rose just 2% year-over-year in Q4 '25. Yet in more recent times, NBS figures suggest a pickup again across January and February. Might you offer your most current perspective regarding CMR patterns as we move toward the March quarter? Also, have you begun noticing any uptick in how consumers feel?

Yongming Wu executive
#8

Thank you for that question. It's true that during the December quarter, soft macro consumption, a mild winter, and the Chinese New Year arriving later all weighed on growth for that period. Additionally, because the promotional window ran longer, our spending on consumer benefits rose relative to earlier years. Consequently, both the CMR and EBITA trajectories weakened. As we enter the March quarter, with consumer sentiment improving as we've seen and momentum stemming from our Quick Commerce strategy, our physical goods GMV along with CMR performance have bounced back substantially versus the December quarter, and EBITA should strengthen in line with that.

Lydia Lu executive
#9

All right. Let's move on to the next question.

Gary Yu analyst
#10

My question is about quick commerce. As I see it, over the last few months, we've hit some milestones when it comes to GDV market share, and we've also seen improvement there. Looking ahead, how should we think about what takes priority? Is market share what we're chasing, or do we want to use this window to make unit economics better and bring losses down? Also, how should we think about the synergy between quick commerce and traditional e-commerce? And how do we see those synergies feeding into stronger CMR growth down the road?

Yongming Wu executive
#11

Certainly, as our market share has grown, we've kept making substantial gains in UE, and that's been powered by better efficiency in fulfillment logistics, by better monetization, and also by optimizing order mix. Given those drivers, we anticipate further UE optimization over the quarters ahead. As for the beneficial effect quick commerce is having on our conventional e-commerce operation and on our whole ecosystem, we observed a very sizable rise in AACs on the platform over the last year. Our AAC figure grew by 150 million in 2025, and that includes 100 million conventional e-commerce physical goods AAC, a number larger than what the prior 3 years produced together. Right now, new consumers' ARPU and purchase frequency sit below those of existing users. So our goal is to keep lifting their ARPU and purchase frequency, and that will act as a fresh growth engine for our platform over the years to come. Quick commerce is plainly fueling sales across a range of categories, including food and fresh produce and health care, and it's adding to accelerated growth at Freshippo and Tmall supermarkets. On the outlook, we're holding to our target of reaching over RMB 1 trillion in Quick Commerce GMV by FY '28. We expect to produce positive cash flow once that GMV target is met, and we expect the quick commerce business to turn profitable in FY '29. Quick Commerce has turned into a cornerstone of our e-commerce business, and it plays a strategically vital role in the AI era by bringing in customers, boosting user engagement, meeting diverse consumer demand, raising transactions and improving monetization, and backing logistics infrastructure. We're dedicated to putting investment into quick commerce across the next two years as we pursue the RMB 1 trillion GMV target as a market leader.

Lydia Lu executive
#12

Operator, let's move on to the next question.

Alicis Yap analyst
#13

I'd like to ask a few things about your chip operation, T-Head. Reports have come out suggesting Alibaba intends to spin off the T-Head division through a separate listing. Could management give us any details on that? If that's the case, when do you expect it to happen? While we're on the subject, could you provide additional operating metrics? Beyond the 470,000 chips you said were shipped to outside customers, how does that figure tie back to the revenue line? What growth rate do you anticipate for the chip business over the next year? I believe you said 60% of these currently come from external customers, so perhaps you could also tell us whether those chips sold to outside buyers are mostly for inferencing. Then on the internal side, are they used for model training as well as inferencing? Finally, how do the < > chips, or T-Head chips, stack up against other chips made domestically? It would be great if management could offer some specifics.

Yongming Wu executive
#14

<<foreign_lang>> < > All right. Many thanks for that question. I'd like to use this chance to go into a bit more depth, since T-Head represents a highly important piece of Alibaba's AI strategy across the whole company. Within China's homegrown AI chip landscape, we strongly hold the view that when it comes to technological and product capabilities, T-Head sits among the top tier of that domestic ecosystem. Our offerings span the full AI workflow, starting at model training and fine-tuning and extending all the way to inference. What's more, our T-Head AI chips already see broad, large-scale deployment through Alibaba Cloud, serving both training workloads and inference use cases. Meanwhile, more than 60% of T-Head chips go to external commercial customers spanning Alibaba Cloud's public cloud and hybrid cloud offerings. Those outside commercial clients come from a range of industries, such as Internet finance, autonomous driving and intelligent manufacturing. These external commercial customers are putting T-Head chips to work in both their training and inferencing workloads. On top of that, with the T-Head software stack, we offer outstanding compatibility with the Linux ecosystem. That means customers can move their systems over with ease, without devoting lots of time to migration. One more thing I'd add is that, as I see it, what T-Head means to Alibaba isn't just our ambition to narrow the distance between chips made domestically and their foreign counterparts, foreign-produced chips, when it comes to manufacturing processes and overall performance along many dimensions. Yet since our chips continue to trail foreign counterparts and performance across various respects, we aim to pursue deeper co-design with Alibaba's cloud infrastructure and the Qwen model so as to deliver better cost effectiveness. That's a key differentiator, and it's how our approach to chip design at T-Head distinguishes us from other chip companies. The main thing we're aiming for is building AI that gives you more bang for your buck; that's what will turn it into a crucial offering for the < > platform, and it lets us bring inference expenses down as we move forward. On top of making our AI more efficient in general and cutting costs, there's a second element in the mix, which is the particular situation the AI sector in China finds itself in right now. Given that, a major advantage on our side is having a guaranteed source of AI compute. My view is that across the coming 3 to 5 years, AI computing capacity worldwide is going to be in extremely tight supply, particularly within the Chinese market, and we are the sole cloud compute player there with in-house chip design know-how. That makes T-Head absolutely vital to the Alibaba Group, and boosting the availability of AI compute will help our cloud and AI operations, MaaS included, build up stronger momentum for growth. At T-Head, across the last 2 years, we've managed to commercialize and roll out chips successfully, with cumulative volume above 470,000 units and yearly revenue hitting the RMB 10 billion mark. If we look toward '27, well, all the way through 2026, this year and into '27. In the coming year, we anticipate that T-Head's manufacturing capacity for top-tier AI chips will keep on growing. That's going to deliver solid compute support for the group's AI operations and act as a strong engine of growth for our AI efforts as a whole. We likewise think that gains in profitability down the road will come from pushing profit levels even higher, and that too will be quite advantageous. All things considered, what T-Head brings to Alibaba isn't just about trimming costs. Its main role is making sure our supply chain stays resilient during a period when compute is hard to come by. To my mind, that's essential to Alibaba's AI strategy. So yes, it's conceivable, and we aren't shutting the door on T-Head weighing an IPO down the road, even though right now we don't have any set time line.

Lydia Lu executive
#15

Let's take the next question.

Yuan Liao analyst
#16

<<foreign_lang>> < > I'd like to ask about the business goals behind the AI strategy you referred to a moment ago. Over the coming 5 years, revenue should surpass 100 billion. Can you share further specifics about that objective? Say those 5 years run through 2031 — what CAGR would be implied across that stretch? Also, might you unpack the forces behind such expansion and explain how we ought to interpret those forces? With revenue scaling like that, at what point should we anticipate consistent gains in Alibaba Cloud's margins?

Yongming Wu executive
#17

<<foreign_lang>> < > Appreciate the question. Indeed, we do hold the view that in 5 years, revenue tied to our AI and cloud business will be above $100 billion. That strikes us as quite evident. Consider the market expansion we observe right now, plus how strong our product lineup is and the road map that gets us there. In my view, the primary force beneath all of it is really ongoing breakthroughs in what large AI models can do. And over the last couple of months — the two months of 2026 — we have clearly witnessed a trend in which large models have acquired the ability to carry out complicated B2B workflows. A growing number of enterprises are putting agents built on large models to work on end-to-end business tasks. That signals a basic shift in how the market regards IT budgets, meaning IT budgets historically devoted to AI and cloud services. What's changing is that plenty of enterprises, when they consume tokens, no longer view token consumption as belonging to their IT budget. Rather, they regard tokens as a piece of their broader operational or R&D costs. Tokens function as a crucial input to their production, rather than merely a slice of their IT budget. So this stands as the most basic long-term element we identify as propelling AI growth ahead. The biggest growth forces, I believe, will emerge from three areas. The first is the MaaS-driven business, which truly serves as the central growth engine. A range of different use cases will underpin growth in our MaaS business, among them our own applications, along with a wide assortment of AI application scenarios drawn from our entire customer base and spanning many different industries, AI application software included. And we hold that growth propelled by MaaS initiatives will be a crucial revenue driver going forward for both AI and cloud services. Yet second, when it comes to AI and cloud computing, a further highly significant growth opportunity exists. Naturally, we expect public MaaS to become a sizable market down the road. However, among a substantial number of large — medium and large-sized enterprises, there will likewise be appetite for enterprise level, internal inference and training, a new marketplace; and over the long term that market will persist. This isn't something that's going to vanish just because every enterprise bases its choices on its own particular business model along with the security needs of its specific use case or the uniqueness of a given application scenario. Consequently, when it comes to certain application scenarios, enterprises will choose to rely on public MaaS API services, whereas plenty of others will go with solutions deployed privately inside the enterprise. So application scenarios of that kind amount to a substantial incremental growth opportunity for Alibaba Cloud's AI infrastructure. Third, there's yet another significant driver, a significant opportunity that I believe frequently gets overlooked. What I'm referring to is CPU-centric cloud computing, the conventional kind of cloud computing, which still offers considerable room to expand in this era shaped by AI. Conventional cloud computing, then, was built for IT engineers, who in China might add up to a few million, perhaps no greater than 10 million potentially traditional IT engineers. And those people have formed the customer base for conventional cloud computing. In the future, though, there might be billions of agents brought into being by large AI models together with the environment in which they operate. The environment where these agents operate will likewise demand considerable backing from conventional CPU-centric cloud computing; they require those conventional CPUs along with databases, storage and huge quantities of memory to sustain their long-horizon problem solving and continuous operations; so the difficulty is in reshaping the conventional cloud computing market, moving it from a cloud platform built for human users, those IT engineers, toward one tuned for agent-based implication. Therefore, I believe the room to grow there is enormous. So one major challenge facing us this year involves converting conventional cloud computing into a platform better matched to agentic use. And that sits at the heart of Alibaba Cloud's upgrade. As revenue from this business keeps rising, our AI business is going to transform and upgrade, moving away from selling resources and toward selling intelligence, selling intelligent capabilities. And I believe that amounts to an enormous upgrade to the business model. Simultaneously, through integrating our proprietary T-Head chips, we are delivering and will deliver lower costs and greater efficiency, and we believe that as our AI and cloud business keeps expanding in revenue scale, cloud profitability ought to become progressively more visible and we see it sitting on a steady path of improvement; the process of continuing to improve, though, isn't linear. There's a chance of a scale effect breakthrough, reaching an economy of scale or scaling up our T-Head chips, and there could be an enormous leap forward, yet I think that depends on the product as well and on those sorts of economies of scale. It simply won't play out in a linear way. So you inquired about the CAGR, the compound annual growth rate, between 2026 and 2031. I think you can enter that into your calculator and work out what it would be if we assumed it were linear, but my view is that it won't be linear. The way we spend on research and development, along with how we expand in the market, won't follow a straight line, and a portion of what we're funding right now might not produce meaningful growth for as long as 1 or perhaps 2 years down the road; even so, when it comes to that broader 5-year target, our confidence in reaching it is very strong.

Lydia Lu executive
#18

Thank you. Peter, let's take the last question.

Alex Yao analyst
#19

Let me shift to another topic and ask about E-commerce. You said earlier that we were in a 3-year investment cycle for E-commerce. Is that being changed right now, or is it being extended because of new opportunities that have come up in Instant Commerce and in agentic commerce? Or are we still looking at it based on the original 3-year plan, which would put us right now in the middle of that 3-year period, where I assume we would start to see the returns on a stable basis from those investments? Could you walk us through where e-commerce is headed overall, given that three-year investment cycle you mentioned earlier, and also let us know how you're thinking about your positioning and your strategies on this e-commerce track?

Yongming Wu executive
#20

Thanks. Like I said a moment ago, we're putting a very substantial amount of money into the instant retail business this year — the Quick Commerce business this year. And right now, what we're seeing is a really clear-cut opportunity in this area. To restate what I said earlier, over the coming two years we're going to keep putting substantial money in, because our target is to push quick commerce sales past RMB 1 trillion. On top of that, we expect that within those same two years, the money we're pouring into quick commerce will start delivering favorable economic returns across our E-commerce business as a whole.

Executive executive
#21

<<foreign_lang>> < > However, I want to build on that point by introducing the AI angle, since Eddie has discussed AI quite extensively. In my view, e-commerce will also be affected by AI in a very, very major way; that said, when we discuss AI, a 3-year horizon is simply too far out, given that the technology is currently advancing on a timescale of weeks or months; yet this is exactly the reason we are placing substantial bets on the AI side, and we are using AI to launch fresh experiences for consumers and merchants alike, while also enhancing merchants' business models through AI. Our belief is that AI will enable us to achieve enormous improvements in e-commerce throughout various segments of the e-commerce operation. It benefits our B2B business, in which we perceive enormous potential for deploying it, and we will aggressively capture every one of these emerging opportunities.

Lydia Lu executive
#22

Alright. That brings the question-and-answer portion of today's earnings call to a close. Thanks so much for being with us today, and we hope to talk with you again before long.

Operator operator
#23

Appreciate it. That brings today's call to a close. Thanks for taking part. You're free to hang up your lines now.

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.