OpenClaw Frenzy Triggers 34% Alibaba Cloud Price Increase
On March 18, Alibaba Cloud announced significant price increases for its AI-related services, citing a surge in global demand. The price for its Pingtouge Zhenwu 810E computing cards will rise by 5% to 34%, while its CPFS file storage product will increase by 30%. While the official reason points to supply chain pressures, insiders report the primary driver is a massive spike in AI 'token' consumption. This is a direct result of the viral adoption of OpenClaw, an open-source AI agent framework colloquially known as "lobster" in China, which has led to record growth for Alibaba's Model-as-a-Service (MaaS) business between January and March 2026.
Unlike traditional chatbots that answer queries within a single interface, OpenClaw operates as an "AI executor," capable of controlling a computer to perform complex, multi-step tasks across different applications. This functionality causes a dramatic increase in resource usage, with a single agent generating hundreds or even thousands of model API calls daily—an order of magnitude greater than a standard AI chat application. By March 15, the framework had already accumulated over 315,000 stars on GitHub, signaling its rapid integration into the developer ecosystem.
Weekly AI Token Use Triples to 16 Trillion as Chinese Models Dominate
The popularization of OpenClaw has ignited a token consumption revolution. According to data from API aggregator OpenRouter, global weekly AI token usage has soared to approximately 16 trillion, nearly tripling from its pre-OpenClaw levels in January 2026. Chinese AI models are the largest beneficiaries of this boom. For the month ending March 9, models from Chinese labs accounted for the highest usage globally, with MiniMax's M2.5 at 7.5 trillion tokens, followed by Moonshot's Kimi K2.5 at 4.2 trillion.
This market capture is fueled by an aggressive pricing strategy that Barclays analysts describe as a "cost nuclear weapon." Chinese AI models are priced dramatically lower than their U.S. competitors, with most costing less than 10% of OpenAI's GPT-5.4. For example, running MiniMax M2.5 for an hour costs just $1, approximately 1/10th to 1/20th the price of using a comparable model like Claude Sonnet. Despite these low prices, Chinese AI firms are maintaining high profitability, with UBS reporting that inference services at companies like Zhipu and MiniMax carry gross margins of 60-70%.
'China's SaaS Moment' Unlocks New Revenue for Tencent and Baidu
The OpenClaw phenomenon is doing more than just driving cloud and model revenue; it is fundamentally reshaping the software market in China by creating a viable path to Software-as-a-Service (SaaS) monetization. Chinese tech giants have quickly productized OpenClaw into "ChinaClaw" offerings, moving away from the historically free model for consumer AI. These new agent-based services are being introduced with subscription fees, typically ranging from 40 to 500 yuan per month. Barclays has labeled this shift "China's SaaS moment," suggesting the country could leapfrog traditional SaaS and move directly into an era of AI-driven productivity subscriptions.
Investment banks like UBS and Barclays have identified a clear, multi-layered investment thesis. At the infrastructure level, cloud providers like Alibaba and Kingsoft Cloud are poised to benefit from the escalating compute shortage and pricing power. At the model layer, labs like MiniMax and Moonshot are directly monetizing the explosion in token demand. Finally, at the application layer, platform giants like Tencent and Baidu are positioned to capture new, high-margin SaaS revenue by integrating these AI agents into their vast ecosystems, from WeChat to Baidu Search.
