This article is compiled from the WeChat mini program: Encrypted Cloud Travel
In the summer of 1858, a copper-core cable crossed the Atlantic Ocean floor, connecting London and New York.
The significance of this matter has never been about transmission speed, but about power structures. Whoever lays the undersea cables can siphon off the flow of information. The British Empire relied on this global telegraph network to hold onto intelligence from the colonies, the prices of cotton, and news of wars.
The strength of the empire lies not only in its fleet but also in that cable.
More than one hundred and sixty years later, this logic is being replayed in an unexpected way.
In 2026, China's large models are quietly consuming the global developer market. The latest data from OpenRouter shows that Chinese models account for 61% of the Token consumption among the top ten models on the platform, with the top three all coming from China. API requests from developers in San Francisco, Berlin, and Singapore are crossing the Pacific undersea cable to reach data centers in China, where computing power is consumed, electricity flows, and results are sent back.
Electricity has never left the Chinese grid, but its value has been delivered across borders through Tokens.
The Great Migration of AI Models
On February 24, 2026, OpenRouter released a weekly report: the total Token consumption of the top ten models on the platform was about 8.7 trillion, with Chinese models accounting for 5.3 trillion, or 61%. MiniMax M2.5 topped the list with 2.45 trillion Tokens, followed closely by Kimi K2.5 and Zhishu GLM-5, with all top three coming from China.
Latest data from February 26
This is not a coincidence; a fuse ignited everything.
Earlier this year, OpenClaw emerged, an open-source tool that truly allows AI to 'work,' capable of directly controlling computers, executing commands, and completing complex workflows in parallel, with GitHub stars surpassing 210,000 within weeks.
Financial professional John installed OpenClaw at the first opportunity and connected to the Anthropic API, starting to automatically monitor stock market information and promptly report trading signals. A few hours later, he stared in shock at his account balance: dozens of dollars, gone.
This is the new reality brought by OpenClaw. In the past, chatting with AI consumed a few thousand Tokens, and the cost was negligible. After integrating OpenClaw, AI runs dozens of subtasks simultaneously in the background, repeatedly calling context and iterating in cycles; Token consumption is not linear but exponential. Bills accelerate like a car with its hood open, with the fuel gauge dropping and unable to stop.
A 'brilliant idea' quickly spread in the developer community: use OAuth tokens to connect Anthropic or Google's subscription accounts directly to OpenClaw, turning the monthly fee's 'unlimited' quota into free fuel for AI Agents, which is also a method many developers use.
Official countermeasures soon followed.
On February 19, Anthropic updated its protocol to explicitly prohibit the use of Claude subscription credentials for third-party tools like OpenClaw. To access Claude's features, one must go through the API billing channel. Google further banned accounts that accessed Antigravity and Gemini AI Ultra via OpenClaw.
'The world has long suffered under Qin,' John quickly embraced domestic large models.
On OpenRouter, the domestic large model MiniMax M2.5 scored 80.2% on software engineering tasks, while Claude Opus scored 80.8%, with the difference almost negligible. But the price difference is astronomical; the former charges $0.3 per million Tokens at the input end, while the latter charges $5, about 17 times more.
John switched over, and the workflow continued to operate, but the bills shrank by an order of magnitude. This migration is synchronously occurring globally.
Chris Clark, COO of OpenRouter, stated directly that the reason Chinese open-source models have captured a large market share is that they occupy an unusually high proportion in the agent workflows operated by American developers.
Electricity Export
To understand the essence of Token export, one must first clarify the cost structure of a Token.
It seems light; one Token is approximately equal to 0.75 English words. In a typical conversation with AI, only a few thousand Tokens are consumed. But when these Tokens stack up in trillions, the physical reality behind them becomes heavy.
Breaking down the cost of Tokens, there are only two core components: computing power and electricity.
Computing power is the depreciation of GPUs. If you buy an NVIDIA H100 for about $30,000, its lifespan translates to a depreciation cost for each inference. Electricity is the fuel for the continuous operation of data centers. Each GPU consumes about 700 watts at full load, and with the cooling system overhead, a large AI data center's electricity bill can easily exceed hundreds of millions of dollars annually.
Now, let's map this physical process.
An American developer sends an API request from San Francisco. Data departs from California, travels via the Pacific undersea cable to a data center somewhere in China, where the GPU cluster begins to work, electricity flows from China's grid to those chips, inference is completed, and the results are sent back. The entire process may take only a second or two.
Electricity has never left China's grid, but the value of electricity has been delivered across borders through Tokens.
Here lies a magical aspect that ordinary trade cannot reach: Tokens have no physical form, do not need to go through customs, will not be hit by tariffs, and are not even included in any current trade statistics. China has exported a large amount of computing power and electricity services, but in official commodity trade data, it is almost invisible.
Tokens have become a derivative of electricity; the export of Tokens is essentially the export of electricity.
This is also due to China's relatively low electricity prices, with comprehensive electricity prices about 40% lower than those in the United States, which is a physical cost difference that competitors can easily replicate.
In addition, China's AI large models also have algorithmic and 'involution' advantages.
The MoE architecture of DeepSeek V3 activates only part of the parameters during inference. Independent tests show that its inference cost is about 36 times lower than GPT-4o, and MiniMax M2.5 also has 229B total parameters but only activates 10B.
At the top level is involution, with Alibaba, ByteDance, Baidu, Tencent, Dark Side of the Moon, Zhishu, MiniMax... dozens of companies stepping on each other in the same track. Prices have long fallen below the reasonable profit range, and losing money for publicity has become the norm in the industry.
Looking closely, this is similar to the export of Chinese manufacturing, utilizing supply chain advantages and industry involution to sharply drive down Token prices.
From Bitcoin to Tokens
Before Tokens, there was another electricity export.
Around 2015, power station managers in Sichuan, Yunnan, and Xinjiang began to receive a peculiar group of guests.
These people rent abandoned factories, fill them with dense machines, and run them 24/7. The machines produce nothing; they just keep solving a math problem, and occasionally, they might calculate a Bitcoin from this endless math problem.
This is the first generation of electricity export: converting cheap hydroelectric and wind power through the hashing operations of mining machines into globally circulating digital assets, then cashing them out as dollars on exchanges.
Electricity has not crossed any borders, but its value has flowed into the global market through Bitcoin.
In those years, China's computing power accounted for more than 70% of the global Bitcoin mining computing power. China's hydropower and coal power participated in a global capital redistribution in such an indirect way.
In 2021, all of this came to a halt. Regulatory hammers fell, miners scattered, and computing power migrated to Kazakhstan, Texas in the U.S., and Canada.
But this logic has never disappeared; it was just waiting for a new shell until ChatGPT emerged, and the large model competition began. Former Bitcoin mining sites transformed into AI data centers, mining machines became computing power GPUs, and the Bitcoins once produced became Tokens, with only electricity remaining unchanged.
The export of Bitcoin and Tokens is structurally identical at the foundational logic level, but Tokens have more commercial value at present.
Mining with mining machines is a purely mathematical calculation, and the produced Bitcoin is a financial asset whose value comes from scarcity and market consensus, unrelated to 'what it calculated.' Computing power itself has no productivity; it is more like a byproduct of a trust mechanism.
Large model inference is different. GPUs consume electricity, and the output is real cognitive services—code, analysis, translation, creativity. The value of Tokens comes directly from their utility to the user. This represents a deeper embedding; once a developer's workflow relies on a certain model, the cost of switching will increase over time.
Of course, there is one key difference: Bitcoin mining was expelled by China, while the export of Tokens is actively chosen by global developers.
Token War
The undersea cable laid in 1858 represents the British Empire's sovereignty over the information superhighway; whoever owns the infrastructure can define the rules of the game.
The export of Tokens is also a war without a declaration, full of obstacles.
Data sovereignty is the first wall. An API request from an American developer is processed through a Chinese data center, with data physically flowing through China. For individual developers and small applications, this is not a problem, but in scenarios involving corporate sensitive data, financial information, and government compliance, this is a hard injury. This is also why the penetration rate of Chinese models is highest in developer tools and personal applications, but they have almost no presence in core corporate systems.
Chip bans are the second wall; China's AI development faces export controls on high-end GPUs from NVIDIA, and the MoE architecture and algorithm optimizations can only partially offset this disadvantage. The ceiling still exists.
But the resistance before us is just the prologue; a larger battlefield is taking shape.
Tokens and AI models have become a new dimension of strategic competition between China and the U.S., comparable to semiconductors and the internet of the 20th century, and even closer to an older metaphor: the space race.
In 1957, the Soviet Union launched Sputnik 1, shocking the United States, which quickly initiated the Apollo program, investing the equivalent of hundreds of billions of dollars today to ensure victory in the space race.
The logic of the AI competition is astonishingly similar, but the intensity will far exceed that of the space race. Space is, after all, a physical space that ordinary people cannot feel; AI penetrates the capillaries of the economy. Behind every line of code, every contract, and every government decision system, there may be a large model from a certain country running. Whoever's model becomes the default infrastructure for global developers will gain structural influence over the global digital economy.
This is exactly what makes the export of Chinese Tokens truly unsettling for Washington.
When a developer's code repository, Agent workflow, and product logic are built around the API of a certain Chinese model, the cost of migration will increase exponentially over time. By then, even if U.S. legislation restricts it, developers will resist by walking away, just as no programmer can abandon GitHub today.
Today's Token export may just be the prologue to this long game. China's large models have not claimed to overturn anything; they simply provide services to every developer with an API Key at a lower price.
This time, the teams laying the cables are those engineering teams writing code in Hangzhou, Beijing, and Shanghai, and the GPU clusters running day and night in a certain province in the south.
This competition has no countdown; it is ongoing 24 hours a day, measured in Tokens, and the battlefield is every developer's terminal.#加密市场反弹 $BTC