Power Centralized, Responsibility Delegated.

Content/Huanlao

Editor/Yonge

Proofreader/Mangfu

On April 8, Alibaba CEO Wu Yongming issued his second company-wide letter in three weeks. If the establishment of the ATH Business Group on March 16 was a land grab, consolidating scattered AI assets into a single basket using Token logic, then this letter was about dividing the fields and assigning names to each critical link.

Zhou Jingren heads the Tongyi Large Model Business Unit, focusing on Qwen; Li Feifei becomes Alibaba Cloud CTO, overseeing AI cloud infrastructure; Wu Zeming steps down from Taobao Flash Sales to focus on Group CTO duties while adding responsibility for the AI inference platform. All three report directly to Wu Yongming. A Technical Committee is also established, with Wu Yongming as chair and the trio as members.

In one month, Alibaba first established its strategic language and then assembled its execution lineup. Only after these two steps was true alignment across the organization achieved, marking the official dawn of Alibaba's AI unification era.

Part.1

Token

Alibaba's 'New World Language' for Unification

Most tech companies discuss Tokens as the basic computational units of large models—how many are consumed per call, pricing per million Tokens. But Alibaba has inscribed Tokens into the name of its business group, the titles of company-wide letters, and the logic of every organizational tier. This is no longer a technical issue but an economic one.

Over the past two years, Alibaba has paid its AI tuition and left behind difficult problems. Technologically, Qwen's open-source influence and model capabilities are widely recognized; commercially, however, C-end applications have yet to achieve absolute dominance amid fierce market competition, while MaaS growth has not fully matched the leap in model capabilities.

The root cause lies in the lack of unified 'measurement standards' across Alibaba's internal business lines.

Model teams talk about parameter counts and Benchmark rankings; MaaS teams focus on API call growth rates; C-end product teams emphasize DAU and retention; internal businesses discuss GMV. With no conversion formulas between these languages, the pathway for translating AI capabilities into commercial value faces immense internal resistance at Alibaba, even devolving into a wasteful scenario where each business builds its own AI wheels.

Wu Yongming's solution is Tokens—elevating them from a technical unit of large models to a universal value metric within and beyond Alibaba's ecosystem.

The establishment of ATH (Alibaba Token Hub) three weeks ago formally establish d Alibaba's future 'new world language.' As the highest-ranking leader (Position No. 1), Wu Yongming mandated organizational alignment across the group's underlying logic, declaring that any future AI narrative not centered on Tokens is mere posturing.

Whether for model inference, API calls, agent execution, or future use of merchant tools in e-commerce, all can be converted into Token consumption—serving as both a unit of measurement and pricing, as well as a mutually recognized value anchor for cross-business collaboration.

In the past, cross-business collaborations relied on budgets, resource swaps, and high-level coordination, resulting in low efficiency and high friction. In the future, businesses can settle accounts directly using Tokens: Business A calls a model, consumes Tokens, and Business B pays for them. Every cross-departmental use of AI capabilities will generate a clear Token invoice, eliminating reliance on vague brotherhood or high-level mediation.

To put it less fashionably, Alibaba is establishing an internal currency system for its AI ecosystem, with ATH serving as the central bank—quantifying AI usage costs and enabling transparent, automated settlement across business units.

This is not a product strategy but an infrastructure strategy. The success of infrastructure lies not in how high a single model scores but in the stability, efficiency, and irreplaceability of the entire chain.

Does this strategic judgment carry risks? Certainly. But its direction is correct. In the era of AI Agents, true competitiveness lies not in standalone model capabilities but in the ability to build a complete closed loop from model production to application consumption. Tokens, as a unified metric run through ing this loop, possess architectural rationality.

Part.2

Unbundling and Repositioning

Letting 'Tech Geniuses' Focus on Pure Innovation

To grasp the depth of this adjustment, one must first recognize its previous structural vulnerabilities.

Prior to this adjustment, the burden of technical execution nearly rested solely on Zhou Jingren's shoulders—serving as both Alibaba Cloud CTO overseeing the cloud's foundation and leading Tongyi Lab to advance large model R&D.

This superhero-style bundling, while effective for resource concentration during the pioneering phase, only led to mutually interfering objectives once AI entered deeper waters. Open-source models chase performance ceilings, while cloud businesses prioritize profit margins—inherently conflicting in resource allocation.

Thus, the first layer of surgery on April 8 was unbundling.

Zhou Jingren stepped down as Alibaba Cloud CTO to exclusively lead the upgraded Tongyi Large Model Business Unit and serve as Chief AI Architect on the Technical Committee.

Note that the elevation from 'Tongyi Lab' to 'Tongyi Large Model Business Unit' itself signals a shift. Labs prioritize publishing papers, climbing rankings, and pursuing technical leadership; business units focus on product delivery, commercialization, and developer retention. The same team, but different evaluation logics, corresponds to entirely different behavioral patterns.

Zhou Jingren's departure from the Alibaba Cloud CTO role to focus solely on Qwen suggests Wu Yongming believes the model has not yet reached its peak and requires undivided attention. The departure of Qwen's core leader, Lin Junyang, three weeks earlier made this role even more critical.

Zhou's current mission may not just be maintaining leadership but creating an insurmountable gap, making Qwen the default choice for developers selecting domestic foundational models.

Li Feifei succeeding Zhou as Alibaba Cloud CTO represents the most nuanced move in this adjustment. Many expected a technical leader with deep model expertise to replace Zhou, but Li's background is the opposite—a former professor at the University of Utah with an engineering pedigree in databases, having no direct connection to large model R&D.

Yet this is precisely what Wu Yongming wanted. Alibaba Cloud CTO does not need to be the most model-savvy person; Zhou handles that. Wu needed someone who could stabilize AI cloud infrastructure. Distributed database systems like PolarDB and AnalyticDB, which withstood Taobao's Double 11 extreme pressure, bear Li's engineering DNA.

AI inference workloads impose entirely different infrastructure requirements than traditional databases—high throughput, ultra-low latency, and elastic scalability. Li's task is to deliver this capability stably as a cloud service, ensuring Token consumption remains uninterrupted at any scale.

Wu Zeming represents the most battle-tested move in this chess game. As Alibaba's first post-80s Group CTO and a 22-year veteran who rose from Taobao's technical frontlines to the group's highest technical position, he stepped down as Taobao Flash Sales CEO (handed over to Lei Yanqun from the 'China Supply Iron Army') to fully return to his Group CTO role while adding responsibility for the AI inference platform.

This platform underpins the entire Token flow from model input to output, determining Token generation efficiency and cost. Training models is a one-time capital investment; inference represents ongoing costs behind every Token revenue. Poor inference renders even the strongest models and stable clouds unusable.

As Technical Committee convener, Wu Zeming must coordinate AI capability demands across all group businesses—Taobao & Tmall, Local Services, International E-Commerce, Cainiao... Each wants AI, but each has unique workload characteristics, latency requirements, and cost sensitivities. The inference platform's value lies in unifying these demands for elastic supply and Token-based billing.

These three individuals precisely correspond to the three most critical links in the AI industrial chain: model R&D, cloud infrastructure, and inference execution—one crafts bullets, one builds guns, and one pulls triggers. All are indispensable.

Part.3

From Racing to Centralization

Concentrating Forces for Major Initiatives

The past two decades of Chinese internet prosperity largely relied on a racing mechanism—resources went to whoever emerged victorious. However, in the deep waters of AI large models, which require tens of billions in computational investment, racing has become an extreme luxury and waste.

This is the third layer of meaning behind Wu Yongming's sweeping adjustment: centralization for more efficient breakthroughs.

On April 8, Taobao & Tmall Group also announced adjustments. Its AI focus shifted from To C to To B, with core OKRs becoming merchant-side AI tool retention rates and GMV contributions. The 'Future Innovation Business Unit' merged directly into ATH. A person close to Taobao & Tmall put it bluntly: 'This year, ATH leads the group's AI strategy; other businesses don't need to reinvent the wheel.'""This statement reflects a long-standing reality: Alibaba's internal businesses all developed their own AI teams—Taobao & Tmall, Local Services, Cainiao, DingTalk... Resources were fragmented, standards inconsistent, and wheels were repeatedly reinvented—a classic ailment of corporate innovation.

Wu Yongming centralized authority, making ATH the group's sole AI gateway. All businesses must access AI through ATH's Token system. This means Taobao & Tmall will need to consume Tokens to use Qwen's capabilities, with ATH managing Token billing, scheduling, and optimization uniformly.

As Group CEO directly overseeing ATH, this represents a rare instance of Alibaba's top leadership inserting itself so deeply. AI-era computational clusters, data flywheels, and engineering capabilities demand centralized allocation at the group's highest density; this top-down design alignment is a prerequisite for corporations to truly concentrate forces for major initiatives.

The external world has already witnessed the astonishing combat effectiveness unleashed by this organizational alignment. Within less than two weeks of ATH's establishment, Alibaba released large models almost daily.

Qwen3.5-Omni on March 30, image generation Wan2.7-Image on April 1, and Qwen3.6-Plus targeting code agents on April 2 (setting a single-day call record of 1.4 trillion Tokens). Even stronger Qwen-3.6-Max is expected post-Qingming Festival.

Taobao & Tmall's ongoing 'Qianniu Claw' plan exemplifies this shift. Qianniu, the merchant backend, will upgrade into an AI Agent platform where merchants can access various AI capabilities—automated listing, intelligent customer service, marketing copy generation—each consumption depleting Tokens. Taobao & Tmall will no longer rely solely on GMV commissions but can also generate revenue based on Token consumption.

This represents a entirely new revenue stream for Alibaba with extremely low marginal costs. Token generation costs decline rapidly as model efficiency improves, but pricing to merchants can remain stable.

Wu Yongming previously stated in earnings calls that AI and cloud annual revenue should reach $100 billion within five years. Achieving this through selling computational power and model APIs alone would be difficult. However, if all AI calls across e-commerce, local services, logistics, and other businesses are incorporated into the Token billing system, the $100 billion target gains tangible support.

From strategic alignment to middleware interoperability, and finally to genuine mindset shifts across business units, each step tests execution and Wu Yongming's resolve as the top leader. Architectural clarity merely marks a good beginning; the real test starts now.

Fortunately, Alibaba has identified the anchor point to pierce through the fog: centralize power, delegate responsibility, position talent appropriately, and execute a clear strategy. From here on, let the Tokens decide.

END