Alibaba is gaining a new set of revenue-generating assets from its past investments.
Strategic investments made years ago in AI startups are now starting to pay dividends for Alibaba in the era of token consumption.
Among these assets, the most prominent are the recently listed MiniMax and Zhipu AI, as well as Moonshot AI, which is preparing for an IPO. Alibaba benefits in two ways: it gains net investment income from its equity stakes, and these companies have become significant consumers of Alibaba's computing power amid the token consumption boom.
This represents a classic case of killing two birds with one stone.
On May 13, Alibaba released its quarterly earnings report for the period ending March 31, 2026, along with its annual report for fiscal year 2026. For the full fiscal year, Alibaba's interest income and net investment gains surged by 322% year-over-year to 87.512 billion yuan. This marks the highest income from this segment since 2022.
The substantial increase in Alibaba's interest income and net investment gains is primarily attributed to the public listings of MiniMax and Zhipu AI in January 2026, in which Alibaba holds stakes. Fueled by the token consumption trend, both companies reported significant revenue and profit growth in 2025 and the first quarter of 2026.
* MiniMax: Prior to its IPO, Alibaba held approximately a 13.66% stake, making it the largest external institutional shareholder. MiniMax's revenue grew by 158.9% year-over-year in 2025. * Zhipu AI: Alibaba-affiliated entities (including Ant Group) hold about a 6% stake.
Moonshot AI, which has yet to go public, has even stronger ties to Alibaba. According to Alibaba's fiscal year 2024 annual report disclosure, it holds approximately 36% of Moonshot AI's preferred shares. This suggests that Alibaba's net investment income is poised for further growth once Moonshot AI completes its IPO.
It is noteworthy that these emerging AI leaders are generally key clients of Alibaba Cloud.
* MiniMax: Alibaba Cloud is one of MiniMax's significant computing power suppliers, with procurement amounting to approximately $58.4 million in the first three quarters of 2025. * Zhipu AI: Alibaba Cloud is one of its major suppliers, and its models are commercialized through platforms like Alibaba Cloud's Bailian. * Moonshot AI: It is a flagship model customer for Alibaba Cloud, frequently showcased as a case study.
As these AI companies rode the wave of token consumption at the beginning of 2026, the massive token usage translated into tangible benefits for Alibaba Cloud. The financial report shows that Alibaba Cloud's revenue increased by 38% year-over-year to 41.626 billion yuan, with external commercial revenue growth accelerating to 40%—the fastest pace in nine quarters.
For the token consumption trend and Alibaba's decision-makers who firmly invested in these AI companies, Alibaba owes a debt of gratitude.
During the quarter, Alibaba's total revenue grew by 3% year-over-year to 243.38 billion yuan, while adjusted EBITA decreased by 84% to 5.102 billion yuan. Against this backdrop, Alibaba Cloud significantly outperformed the group average, with revenue up 38% to 41.626 billion yuan and adjusted EBITA rising 57% to 3.796 billion yuan, driven by token consumption and AI transformation benefits.
Clearly, "AI and Token" have become Alibaba's new growth engines.
The factors pressuring Alibaba's profits remain the same three persistent challenges from recent quarters:
* Cloud and AI Infrastructure Build-out: Capital expenditures reached 26.9 billion yuan for the quarter, primarily directed towards data centers, server capacity expansion, and foundational AI infrastructure. * Flash Sales Business and Qwen Consumer User Acquisition: The group's sales and marketing expenses increased by 17.2 billion yuan year-over-year, mainly allocated to instant retail operations and marketing for Qwen AI. * Overall R&D Investment in AI Products and Large Models: Product development expenses, including algorithm iteration and R&D personnel costs, amounted to 18.957 billion yuan.
These expenditures broadly align with strategic directions in AI and e-commerce. In AI, Alibaba continues heavy investment in infrastructure, enhancing its token-centric AI industry chain, and increasing efforts in the consumer market. In e-commerce, alongside acquiring customers through flash sales, Alibaba is also strengthening the AI transformation of its overall e-commerce business.
However, challenges and uncertainties persist.
Across the broader AI landscape, independent AI model and agent companies, both domestically and internationally, are generally attempting to avoid dependency on a single chip supplier or cloud provider. As portfolio companies, MiniMax, Zhipu AI, and Moonshot AI cannot be considered monolithic parts of an "Alibaba system." If Alibaba aims to maximize gains, it must compete with formidable rivals.
In the consumer AI market, Qwen experienced a dip in user numbers after a marketing push during the Lunar New Year period. It currently holds a "runner-up" position. To challenge the top players, it urgently requires sustained national-level attention and enhanced product capabilities.
On the e-commerce front, Alibaba and Meituan are approaching a decisive moment in the instant retail battle—the summer of 2026 is viewed by both sides' leadership as a critical window. Given the performance pressure on the flash sales leadership after multiple quarters of high investment, the period from June to August is likely to be a watershed: securing more support to decisively suppress Meituan, or maintaining the status quo and gradually scaling back.
The most fundamental challenge may still be the AI revolution of the core business: How exactly should Alibaba execute its AI-powered e-commerce strategy?
Following Alibaba's earnings call, at 10:30 PM Beijing time on May 13, one hour after the U.S. stock market opened, Alibaba's stock price rose 6.8% to $143.95 per share.
**Token Dividends Arrive, Alibaba Transforms**
At Alibaba's Xixi campus in Hangzhou on May 10, "Alibaba Day," over a dozen posters for pet adoption were displayed in the lobby of Building C5 in Zone C. This is an annual event called the "Alibaba Day Stray Animal Adoption Fair."
Unlike previous years, this year's posters fittingly adopted an "AI style." The cat and dog posters were filled with AI-related puns and descriptions, reflecting the company's current focus.
Not far from the lobby, in a presentation hall, mascots representing several of Alibaba's AI products performed enthusiastically for the audience. Meanwhile, presenters enthusiastically explained how to use Qwen to order flowers for family members...
This is a microcosm of the current "company-wide AI transformation" within Alibaba.
After establishing the Alibaba Token Hub (ATH) business group on March 16 and the "Group Technology Committee" on April 8, Alibaba is significantly advancing business AI upgrades internally. It is also restructuring its business landscape and employee management models according to token workflows.
* All employees receive a monthly allocation of free tokens (usable for Qwen, Wukong, Qoder, Qianwen, etc., depending on the department). * For most technical teams, comprehensive evaluations now incorporate dual metrics: token consumption and output quality (assessed via AI quality checks and manual review). * AI tool utilization rates have become a key assessment focus for business units, which can apply to develop targeted agent tools for specific scenarios.
An informed source stated that initially, some departments attempted to make pure token consumption a key target, but quickly shifted their approach after practical experience.
"Currently, the entire group does not use pure token consumption volume as a metric. Instead, it places greater emphasis on effective token consumption and the actual ROI based on AI," the source said.
Regarding the business landscape, ATH, which encompasses the Tongyi Lab, Qianwen, Wukong, MaaS, and the AI Innovation Division, has focused on improving the following aspects over the past two months:
* Using the ATH framework and the Technology Committee system to unify planning for core technical directions. * Reducing "duplicate R&D" across different business segments and increasing the "reuse of technical and product capabilities." * Establishing more fixed and clear cross-departmental communication mechanisms.
For the broader AI market, Alibaba has deconstructed demand based on the token consumption workflow as the main thread, aligning different products with specific demand scenarios. Macroscopically, the core directions going forward include three major areas:
* In foundational cloud and chips: Beyond meeting internal AI business needs, strengthen external market penetration to expand the lead. * In the AI-to-Consumer market: Focus on token consumption, user volume, and ecosystem synergy, pursuing not only consumer user numbers and activity but also the integration and linkage of products like Qwen with other business segments within the Alibaba ecosystem. * For emerging technological directions and scenarios, employ agile, rapid trial models: including new forms of AI hardware, quantum computing, cloud computers, etc.
Notably, Alibaba will maintain its focus on external AI-related investments. During the May 13 earnings conference call, Alibaba's CFO stated, "Over the past year, we have been very firm in making (AI) investments, and we intend to maintain this firm attitude over the next two years, continuing to carry out such investments."
After the model frenzy, Alibaba's current main investment thesis for AI is reportedly:
* Investing in projects that can organically integrate with Alibaba's current token strategy ecosystem, such as some emerging agent or AI application projects. * Shifting the investment perspective from "soft" to "hard," maintaining an open attitude and interest in sectors like AI hardware, embodied intelligence, and drones. * Experimenting with various investment models, including independent venture models for internally incubated projects, potentially exploring a more flexible internal incubation ecosystem.
An informed source clarified that previously circulated rumors about "Alibaba attempting to invest in DeepSeek" are inaccurate. The real situation is that Alibaba representatives did have contact with the DeepSeek team in 2025, but there has been no new contact in 2026. Around the time of the 2025 contact, other companies like ByteDance also engaged with DeepSeek. For Alibaba's current AI ecosystem, DeepSeek is not considered an indispensable target.
However, another industry insider suggested there might be a misunderstanding. "Alibaba's T-Head chip business has indeed been in contact with DeepSeek, related to DeepSeek's domestic chip substitution strategy. But this is a business cooperation and product procurement relationship, not an investment."
From an external investment perspective, Alibaba is seeking more "returns on investment": through strategic investments, it backs more emerging companies that can organically integrate with Alibaba Cloud, Qwen, and T-Head. These companies can not only generate investment returns for Alibaba but also become growth clients in terms of token consumption.
From an internal product and business perspective, Alibaba needs its own AI tools and products to serve as an arsenal for its internal AI evolution. Simultaneously, Alibaba urgently requires these products to participate in external market competition and secure positions within the token-driven market landscape.
Within these two dimensions, Alibaba faces both opportunities and challenges. However, one easily overlooked point is the relationship between its strategic resolve and public market sentiment.
Currently, Alibaba's firm commitment to AI is partly because the capital market's high expectations for Alibaba primarily stem from its AI layout and vision. But the risk may also originate here—it, too, needs a new narrative.
Initially, the outside world was captivated by the narrative of "AI infrastructure, the new utilities." Then, expectations grew, leading to narratives about AI for consumers and AI reshaping e-commerce. However, the latter two currently present greater challenges. Essentially, the infrastructure narrative is one of "resource advantage + first-mover advantage," where Alibaba has built sufficient barriers through early action and substantial resources. But the latter two narratives test consumer product capability and operational strength on one hand, and efficiency and determination to reshape existing business and interest systems on the other.
**Qwen and Flash Sales: Two Battlefields Determining the Trajectory**
The day after Alibaba Day, on May 11, Alibaba announced the full integration of Qwen and Taobao.
This provides a clearer observation point for understanding "Alibaba's approach to AI-powered e-commerce."
The overall layout can be likened to a "tree structure":
* Fertile Soil (Technical Foundation Layer): The Qwen large model, and Taotian AIGX. * Trunk (Vertical Line Around Traffic): AI search and recommendation advertising systems, new AI-powered product libraries, etc. * Branches (B2B Merchant AI Tools): AI-powered customer service tools, Alimama AI advertising tools, AIGC tools, etc. * Fruit (Consumer-Facing Products): The Qwen AI shopping assistant, AI try-on, AI universal search, etc.
According to available information, in 2025, Taotian invested significant effort and resources into key areas like the foundation and trunk layers, prioritizing R&D for crucial scenarios within the branches and fruit. These foundational efforts are the reason Qwen and Taobao can now be integrated and why numerous new AI tools for both businesses and consumers have emerged.
Inevitably, intersections occur within this system.
At the entry point, Qwen and Taobao intersect. On the traffic side, search and advertising systems intersect with AI tools for both B and C ends. On the tool side, the output of B-end tools and the exploration of C-end tools intersect...
However, the real "choke points" also appear at these intersections:
* As the consumer entry point and tool for the e-commerce segment, Qwen is not merely a simple "embedding"; it actually impacts the existing business model. * The overall traffic mechanism may require a redesign of "smart" rules alongside this more thorough AI evolution. * Consumers use AI hoping to streamline decisions and reduce ads; merchants use AI tools to produce materials more efficiently. Can AI's power simultaneously play the role of "Neo" in both these worlds?
Among these intersections, the most critical battlefield may be the integration of Qwen and Taobao—as it represents the true "shared stage" for two of Alibaba's strategic directions over the past few years. It is a moment of validation for how much chemical reaction can occur between Alibaba's AI and its core e-commerce business.
However, it's important to note that Taotian faces not only the "new world" war of AI; the story in the traditional world is not over: the battle in instant retail, represented by flash sales, continues.
The war between Flash Sales and Meituan has become: "Success means a business revolution that defeats the giant (Meituan), achieving growth in user numbers, order volume, and GMV; failure means tens of billions in sunk costs."
The key point here lies in how to define success or failure.
The author lists three key statements made by Alibaba's senior management regarding the flash sales business during the May 13 conference call:
* Since April, while maintaining order scale, significant optimization of Unit Economics has been driven by improvements in logistics efficiency and order structure optimization. * "We are confident in achieving overall profitability in instant retail." * Flash sales have shown a clear promotional effect on physical goods e-commerce, especially in new customer acquisition, user activity, driving transactions, and logistics infrastructure.
The author believes the war for scale (order volume, market share ranking) may be nearing or has already reached its end. It is entering a "digestion cycle," focusing on精细化运营 of the existing scale, reducing per-order costs, and truly bringing incremental value to the traditional e-commerce business by cultivating users' "cross-shopping" habits.
The second statement is actually the key point. It is about achieving profitability. This might also be an "intersection" as described earlier: Flash sales need to explore a model that does not rely on "2025-style subsidies" yet can maintain scale and drive cross-shopping.
This, too, is an expectation for "returns on investment."
