Global Daily Average Token Consumption Increases Nearly 300-Fold in Two Years, China's Low-Cost Token Drives Mining Power Expansion
BlockBeats News, March 3rd, Huawei's ICT BG CEO Yang Chaobin delivered a keynote speech at the World Mobile Communications Conference today, stating that the global daily Token consumption has increased nearly 300 times in the past two years, triggering AI applications such as lifelike videos and smart shopping, with over 30 million AI Agents worldwide collaborating.
Against the backdrop of Yang Chaobin's remarks today, the current surge in global calls to China's domestically produced large-scale model APIs and Token outflow driving the expansion of computing power have led to a sharp increase in electricity demand. Yesterday, Musk retweeted a tweet stating, "Electricity production is the best single indicator of industrial capacity." China's electricity generation in 2024 is about 100 trillion kilowatt-hours, 40% more than the sum of the United States and the European Union (27 countries). China's electricity generation surpassed that of the European Union around 2007 and the United States around 2010. Electricity generation not only represents industrial strength but has also become a hardcore indicator of computing power, directly determining the ceiling of AI development. Among the three major bottlenecks to AI scaling—chips, electricity, transformers—the global data center's electricity shortage will be the first to arrive.
Today, as the United States continues to impose tariff blocks, the effectiveness of a single physical blockade is gradually diminishing. The U.S. can block logistics but is finding it difficult to restrict the uninterrupted global supply of low-cost electricity-produced Tokens. Recently, the AI programming subscription package GLM Coding Plan under the Zhipei umbrella went online and sold out immediately, while the domestically produced Chinese AI programming model's paid package, rarely seen, was quickly grabbed. Within just one week of its release, Kimi K2.5 climbed to the top of the global large-model API aggregation platform OpenClaw call list, surpassing giants like Gemini and Claude. China's domestically produced large-scale models, leveraging low prices as a starting point, combined with high performance for user retention, are bringing a price-performance double kill to the AI market.
Outside of the increasingly fierce China-U.S. AI competition, Token outflow has also brought mutual benefit and common development, as China's low-cost Token has, for the second time, indirectly eased U.S. inflation. In the 2000s, China's product outflow relied on cheap labor and textiles; today, it relies on low-cost Tokens. With a stable supply cost of only $0.1 per million Tokens, it serves as an invisible subsidy supporting the Silicon Valley application layer.
