Targeting $100 Billion: Alibaba Defines the Token Economy

$100 billion is a figure mentioned by Alibaba CEO Wu Yongming during an earnings meeting. In his plan, annual commercial revenue from cloud and AI, including MaaS, is expected to surpass $100 billion.This implies a five-year CAGR of over 40%.。

It's important to note that Alibaba Cloud's revenue for the 2025 fiscal year is approximately USD 21 billion, and the target of USD 100 billion is about five times its current level. Today, AI-related businesses have become one of Alibaba’s main growth engines. Alibaba Cloud’s revenue has increased by 36% year-over-year, with AI-related product revenue achieving triple-digit growth for the tenth consecutive quarter. Meanwhile, the company’s MaaS platform business is also experiencing rapid growth, establishing a new engine for the entire group.

Alibaba Cloud is transitioning from selling server computing power to a token-based billing model, shifting from a one-time transaction to a recurring revenue stream. Organizationally, the newly established ATH (Alibaba Token Hub) division is led directly by the CEO, building a new framework centered around the creation, delivery, and application activation of tokens.

On the specific business front, the mass production of self-developed chips, a six-fold surge in token consumption over three months, and price hikes of up to 34% for core products indicate that supply, circulation, and consumption layers are simultaneously taking shape. CFO Toby Xu put it plainly: 'The rapid growth of AI + Cloud over the past few quarters has given us more confidence to increase investment and solidify our full-stack AI technology advantage.'

However, looking back at Alibaba’s current valuation framework, the market is still pricing it using the traditional e-commerce model—around a dozen times PE, anchored to CMR growth and e-commerce profit margins. This pricing method is reasonable when CMR grows by 10%,but when cloud revenue growth reaches as high as 36%, it warrants reconsideration.。

Even if only half of this target is achieved, the scale of the cloud business alone would be sufficient to shift the entire group’s valuation paradigm. At that point, the question the market needs to answer will no longer be 'Can Alibaba’s e-commerce continue to grow?' but rather what multiple should be used to value a cloud company with annual revenue in the tens of billions and profit margins comparable to AWS.

This is why Wu Yongming repeatedly emphasized the importance of full-stack AI capabilities during the earnings call—the synergy between foundational infrastructure and the application layer, once operationalized, means Alibaba’s narrative in the capital markets will no longer be an extension of the e-commerce story but a complete transformation of identity.

01 Prioritize ensuring long-term value realization

If we only look at the surface-level numbers, this earnings report is not flawless; however, upon closer examination,this is a normal part of Alibaba’s process of shifting gears and accelerating.。

First, look at revenue. On the surface, it only grew by 2%, but after excluding the divested Sun Art Retail and Intime Department Store, the like-for-like growth reached 9%. These divestitures are already part of Alibaba's past, with the heavy offline assets of hypermarkets and department stores remaining on the books, dragging down overall growth every quarter and diluting return on capital. After the divestitures, the quality of the group's revenue is actually cleaner, and a 9% like-for-like growth rate is not bad in the current consumer environment.

Next, examine the sales and marketing expenses as a percentage of revenue, which rose from 15.2% in Q4 2024 (FY2025 Q3) to a peak of 26.8% in Q3 2025, slightly retreating to 25.3% in Q4 2025, indicating that the expense ratio has peaked and started to decline.

This represents a clear strategic choice:trading short-term investment for long-term valueOver the past few quarters, Alibaba’s capital expenditures have surged significantly, which will be amortized over multiple future quarters in the form of depreciation. Current profits are being eroded, while corresponding revenues will be gradually released during the depreciation period. When Meta heavily invested in AI infrastructure in 2022, the same pattern occurred: the market cut its valuation based on the income statement, but in hindsight, that was precisely a value trough conducive to entry.

The free cash flow trend clearly signals this shift: -18.8 billion and -21.8 billion respectively in Q2 and Q3 2025, with significant outflows for two consecutive quarters; Q4 2025 turned positive at 11.35 billion, reversing over 30 billion within one quarter. This reversal occurred despite continued high levels of AI investment, demonstrating that the core e-commerce business's cash generation is sufficient to cover the consumption of new ventures. Meanwhile, the number of 88VIP members increased from approximately 49 million to over 59 million, with double-digit growth for several consecutive quarters; overseas commercial losses narrowed by 59% year-on-year.This shows that not only is cash flow improving, but the operational quality across all business lines of the group is also on the rise.。

At this point, the numerical explanation is sufficient. However, what truly deserves careful thought in this earnings report is what will 'carry the main burden' in the future.

The revenue growth of the Cloud Intelligence Group has followed a curve in the opposite direction: +13%, +18%, +26%, +34%, +36%. Not a single quarter slowed down over five quarters, accelerating each quarter. In Q4 2025, quarterly revenue reached 43.3 billion, with AI-related product revenue maintaining triple-digit growth for the tenth consecutive quarter. More importantly, the adjusted EBITA of the cloud division has also seen continuous positive growth over these five quarters, with profit margins stabilizing around 9%. High growth combined with stable profit margins indicates that this is not growth bought by burning cash, but rather high-quality revenue.

For investors, a fundamental question arises: What should be the valuation anchor for Alibaba? Previously, the market priced Alibaba using a PE ratio typical for traditional e-commerce businesses, based on CMR-driven profits. But as cloud revenue grows at a 36% pace and gradually approaches 20% of the group's total revenue,Continuing to use the e-commerce valuation framework to price a company with an annual cloud revenue of 146.6 billion yuan and triple-digit growth in AI income may require a reevaluation of this framework itself.。

This is also why, during the earnings call, Wu Yongming spent a significant amount of time discussing AI and cloud services — there’s a substantial expectation gap between the business progress seen by management and the market pricing.

02 The Alibaba Model for Token Economics

The confidence to break through $100 billion in revenue within five years does not lie in the scale of computing power itself, but in the shift of the business model.

In the IaaS era, Alibaba Cloud was like a real estate developer selling plots of land — customers buy server resources and build their own systems to run programs, and once the plots are sold, the transaction is over. The MaaS era is completely different, transforming into a utility provider charging by usage: enterprises call upon model inference, and every computation is billed per token. Pay-as-you-go, always generating recurring revenue; the former is a one-time deal, while the latter offers continuous income, making their economic value fundamentally different.

Observing Alibaba’s pace of switching business models reveals a set of simultaneous changes:Token consumption has grown sixfold in three months, and the core computing product has just announced a maximum price increase of 34%, a rare move in China's cloud market in recent years.。

According to Morgan Stanley’s assessment, this marks the beginning of the first pricing cycle in 20 years. For every 10% increase in contract pricing, EBITA margins expand by four percentage points. Currently, Alibaba Cloud’s profit margin stands at 9%, compared to AWS’s over 30%, leaving more than 20 percentage points of potential margin expansion.

Both volume is exploding and prices are rising simultaneously, putting Alibaba Cloud’s financial model into a phase where operating leverage is rapidly amplifying.

At its foundation, the scaled mass production of Pingtouge's self-developed GPU underpins this narrative. In this regard, the wording in the earnings report was quite restrained, stating that it has “already made substantial contributions to cloud infrastructure supply.” However, this statement carries significant implications: global shortages in computing power are expected to persist for another two to three years, and nearly all cloud providers are scrambling for NVIDIA’s production capacity.The self-developed chips allow Alibaba to internalize this cost.。

Others raising prices are passing upstream pressure downstream, but Alibaba’s price hike capitalizes on the cost differential brought by its self-developed chips. Morgan Stanley has separately valued Pingtouge between $28 billion and $86 billion, a value that is currently not reflected in Alibaba's stock price. This earnings report also marks the first systematic disclosure of the full-stack layout of large models, cloud, and chips, logically对标ing Google's TPU-supported cloud business path.

With the underlying support in place, the question for Alibaba's business model becomes:Who will consume these tokens?？

The Alibaba Token Hub business group, established on March 16, provides Alibaba’s answer—not by separately seeking consumption scenarios, but by bringing chips, model platforms, Qwen, and Wukong into the same business unit, directly reporting to the CEO. Tokens are produced on Pingtouge's chips, priced and distributed through Alibaba Cloud’s MaaS platform, and ultimately consumed at both the C-end and B-end.

On the C-end, there's Qwen, whose total MAU has exceeded 300 million. About 140 million users have completed AI-driven shopping through Qwen's agent function, covering dining, daily necessities, ticketing, and travel booking. Qwen has been integrated with Taobao, Tmall, Taobao Flash Purchase, AutoNavi, Fliggy, and Alipay, making it the first AI assistant in China capable of executing complex real-world tasks at scale. Every transaction simultaneously consumes tokens and generates GMV, with the consumption scenario itself acting as a token burner.

On the B-end, there's Wukong, embedded in DingTalk, which is used by over 20 million enterprise organizations. It launched the 'Openclaw' solutions covering ten industries including e-commerce, cross-border trade, legal, finance, taxation, and headhunting. Instead of delivering a general framework, it offers scenario-based skill kits with pre-orchestrated workflows. Alibaba's B-end capabilities will be gradually integrated in the form of skills, aiming to create the largest enterprise-level AI skill marketplace.

Speaking of Wukong, what left the deepest impression at this week's product launch wasn't the product itself but a statement made by Wu Zhao, CEO of DingTalk: In the past, people used DingTalk for work; in the future, AI will use DingTalk for work. DingTalk's infrastructure underwent CLI command-line transformation, enabling agents to directly invoke thousands of capabilities without simulating human actions like RPA.This decision speaks louder than any strategic PowerPoint presentation about their genuine belief that the main actors on the internet are shifting from humans to AI.。

03Conclusion

Over the past two decades, the fundamental unit of measurement in China’s internet economy has been clicks. One search, one impression, one redirection—advertisers pay for every click, while platforms take a cut. The entire business model rests on the assumption that user attention is a scarce resource, and whoever controls attention can levy taxes.

This assumption is being rewritten. As users begin to complete shopping, ticket booking, and inquiries through AI assistants, and as companies start using agents to replace human labor in handling processes, the basic unit of economic activity is no longer a click but a token. A token represents the smallest unit of computational power consumption and serves as the new pricing unit in the AI era. Whoever possesses the ability to produce tokens, the platforms to distribute tokens, and the scenarios to consume tokens,will become the tax collector of the next infrastructure cycle.。

Understanding the significance of Alibaba's earnings report lies precisely here: as a giant that has fully participated in China’s Internet and mobile Internet eras, the re-acceleration of their business is not merely about catching up with a new trend, but about competing for the right to define the business model for the next twenty years. $Alibaba (BABA.US)$$BABA-W (09988.HK)$

Disclaimer: This article is intended for learning and communication purposes only and does not constitute investment advice.

173K Views