Alibaba’s $100 Billion Bet on Token Economics
Original article 阿里巴巴千亿美元token经济学 March 20
After Jensen Huang outlined a future of trillion-dollar AI revenues, Alibaba Group responded with a concrete target: $100 billion in annual revenue from its Cloud and AI businesses within five years. In practical terms, this means rebuilding more than half of Alibaba’s current business. In the last fiscal year, total group revenue was roughly RMB 1 trillion, making this one of the most ambitious transitions attempted by any global technology company.
What is the token economy?
Alibaba plans to build an entire AI ecosystem where everything revolves around tokens, as units of work, labor, everything. For this to work, Alibaba plans to control every layer needed to produce and sell those tokens. It designs its own chips through T-Head, runs the computing infrastructure through its cloud division, and builds AI models and applications through its newer AI business units. This approach closely mirrors the framework described by Jensen Huang, who breaks AI into multiple layers from hardware to applications.
What makes Alibaba stand out is that it is trying to control all of these layers at once, rather than focusing on just one part of the system. Compared with other Chinese tech companies, it has the broadest reach across the entire AI stack and operates at the largest scale. That puts Alibaba in the strongest position to make this model work, but it also means the company is under the most pressure to prove it can succeed.
That ambition is not optional. Recent financial results have increased the urgency. Revenue growth has slowed to low single digits, while operating profit has fallen sharply. The market reaction has been swift, with a significant correction in its Hong Kong-listed shares. As in the United States, where investors closely scrutinize the returns on AI spending at companies like Microsoft and Google, Alibaba now faces the same core question: how quickly can AI investment translate into durable revenue.
Where Alibaba differs from its US counterparts is in how it intends to monetize. American firms have largely pursued a layered model. NVIDIA captures value at the hardware layer, Microsoft and Google at cloud and platform layers, and companies like OpenAI focus on model access and applications. Revenue streams are more modular, and ecosystems are relatively open.
Tight integration
Alibaba, by contrast, is pursuing a tightly integrated system in which tokens become the central unit of value across all layers. Chips, cloud infrastructure, model services, and applications are designed to reinforce one another within a single ecosystem. This reflects a broader pattern in China’s technology sector, where platform companies tend to internalize more of the value chain.
The foundation of this strategy lies in computing supply. Alibaba is betting that global and domestic compute capacity will remain constrained over the next three to five years. If that assumption holds, it creates pricing power at the infrastructure layer. Recent increases in the pricing of certain compute products suggest the company is already testing this thesis. In the US, by contrast, hyperscalers are investing aggressively to expand supply, which may limit sustained pricing leverage over time.
The real burden of execution, however, sits at the application and model layer. Alibaba’s consolidated AI business units are responsible for driving token consumption at scale. This is not just a technical challenge but a structural one. The company is attempting to shift enterprise spending on AI from traditional IT budgets into core operational and production costs. If successful, this would significantly expand the total addressable market for AI services.
This is a critical distinction with the United States. In the US, AI adoption has been led by productivity tools, developer platforms, and premium enterprise services. Monetization is still evolving and often tied to subscriptions or usage-based APIs. In China, Alibaba is attempting to embed AI directly into transaction flows across e-commerce, enterprise collaboration, and industrial systems. The goal is not just to sell AI tools, but to make token consumption inseparable from everyday business activity.
Achieving this requires deep integration into workflows. AI agents must move beyond experimentation and become trusted operators within real business processes. This is less about model capability and more about engineering, reliability, and organizational change. It also requires companies to restructure internal systems to accommodate AI-driven execution.
There are precedents for this kind of transformation. Google DeepMind and Google Brain were merged to align research and product more closely, helping Google regain momentum in consumer AI. Microsoft has attempted similar integrations, though with mixed results. OpenAI is now moving toward a unified “super app” model, combining multiple products into a single interface.
Reorganizing around the token
Alibaba’s effort is comparable in scale but different in direction. Rather than separating layers and monetizing each independently, it is reorganizing the company around a single economic primitive: the token. If cloud and AI revenues come to dominate the business, this will not just change Alibaba’s revenue mix. It will reshape its internal structure, incentives, and balance of power.
From the perspective of global investors, the next phase is straightforward. The question is no longer whether AI will drive growth, but whether companies like Alibaba can convert technical capability into sustained, large-scale monetization.
In both China and the United States, the focus has shifted to the same three variables: investment, application, and returns.