Across China's technology sector, a quiet but consequential transformation is underway: companies like ByteDance and Alibaba have begun issuing AI tokens as a form of internal corporate currency, making artificial intelligence consumption measurable, budgeted, and tied to employee performance. Driven by a fierce price war — punctuated by DeepSeek's dramatic 75% price cut — AI has crossed the threshold from specialized tool to everyday infrastructure. In this transitional moment, the token economy serves less as a permanent system than as a bridge, training a generation of workers to think of A
China's tech giants embrace AI tokens as corporate currency in price-war economy
Using AI has become part of almost every task now
Mark
Why would a company bother tracking something as granular as token consumption? Isn't that just adding bureaucracy?
Mimi
It's not really bureaucracy—it's visibility. When AI is free or nearly free, people use it without thinking. But when you can see the cost, even an internal cost, behavior changes. It's like how companies track cloud computing usage. The token budget forces employees to think about efficiency, not just capability.
Mark
So it's a way of rationing something that's becoming abundant?
Mimi
Exactly. Right now, AI is cheap but not free. Companies are in this awkward middle ground where they want everyone using it, but they also want to avoid waste. The token system lets them have both—encourage adoption while maintaining some discipline.
Mark
What happens when prices drop further, like after DeepSeek's cut?
Mimi
The token budgets become less meaningful as a constraint. But by then, the habit is formed. Employees have internalized the idea that AI is a tool you use deliberately, not carelessly. That mindset persists even after the scarcity disappears.
Mark
Is this unique to China, or are other tech companies doing this?
Mimi
It's most visible in China right now because the price war is most intense there. But the underlying logic—treating AI as a measurable, budgetable resource—that's going to spread wherever AI becomes central to daily work. It's not about Chinese culture; it's about economics.
Mark
What does this say about how we'll measure productivity in the future?
Mimi
It suggests productivity metrics are going to become much more granular and AI-centric. Instead of measuring output alone, companies will measure how efficiently employees use AI to generate that output. It's a fundamental shift in what "being good at your job" means.
Der Puls
- ByteDance employees are consuming AI tokens by the billions each month, signaling that artificial intelligence has become as embedded in daily work as email or spreadsheets.
- DeepSeek's 75% price slash on its flagship model sent shockwaves through China's AI industry, accelerating a race to make AI so affordable it functions like a public utility.
- Major firms including Alibaba and Tencent are now issuing token allowances to employees, creating a new layer of corporate accountability where AI literacy is tracked, benchmarked, and expected.
- The token system is widely understood to be temporary — a rationing mechanism for a resource that is cheap but not yet free — yet the behavioral habits it instills may outlast the budgets themselves.
- What begins as an internal accounting tool is quietly reshaping how productivity, capability, and professional value are defined across the Chinese technology workforce.
Across China's technology sector, a quiet but consequential transformation is underway: companies like ByteDance and Alibaba have begun issuing AI tokens as a form of internal corporate currency, making artificial intelligence consumption measurable, budgeted, and tied to employee performance. Driven by a fierce price war — punctuated by DeepSeek's dramatic 75% price cut — AI has crossed the threshold from specialized tool to everyday infrastructure. In this transitional moment, the token economy serves less as a permanent system than as a bridge, training a generation of workers to think of AI fluency not as an advantage, but as a baseline expectation.
A ByteDance employee in Beijing tracks a metric that would have seemed fantastical just two years ago: the number of AI tokens he consumes each month. The figure approaches a billion — and still leaves him somewhere in the middle of his department's rankings. "Using AI has become part of almost every task now," he said, speaking anonymously because his employer had not authorized him to comment publicly.
These tokens — known as ciyuan in mainland China — are not spendable currency. They are the basic unit of AI service consumption, the way electricity is measured in kilowatt-hours. Every query, every generated response, every logged interaction draws from a pool. Across ByteDance, Alibaba, Tencent, and dozens of other major firms, those pools have become formal corporate allowances: budgets that employees must manage alongside their time, with usage tracked, benchmarked by department, and treated as a measure of AI literacy.
The economic force behind this shift is a price war of remarkable intensity. In May, Hangzhou-based DeepSeek cut the price of its V4-Pro model by seventy-five percent — not as a promotion, but as a declaration about the industry's direction. In China, building AI capability and making it cheap have become the same project. The goal is a world where AI is as routine and unremarkable as running water.
When AI becomes that affordable, it stops being a specialist's tool and becomes infrastructure — and infrastructure must be measured and managed. The token economy is, in this sense, a creature of transition: AI is cheap enough to be everywhere, but not yet so cheap it is free. The budgets may eventually dissolve as prices fall further. But the expectations they are encoding — that workers will integrate AI into their thinking, that fluency in these tools is a professional baseline — are likely to endure long after the tokens themselves are forgotten.
A ByteDance employee in Beijing sits down each month to track a consumption metric that would have seemed absurd just two years ago: the number of artificial intelligence tokens he has burned through. The figure hovers near a billion. It is, by any measure, staggering. Yet when he compares his usage to colleagues in the same department, he finds himself nowhere near the top of the list.
These tokens—called ciyuan in mainland China—are not currency in any traditional sense. You cannot spend them at a café or use them to pay a mortgage. They are, instead, the basic unit of measurement for artificial intelligence services, the way electricity is measured in kilowatt-hours or data in gigabytes. Every question typed into an AI model, every response generated, every interaction logged consumes a certain number of tokens. The employee, speaking on condition of anonymity because his employer had not authorized him to discuss the matter publicly, described the integration as seamless and total: "Using AI has become part of almost every task now."
What is happening inside ByteDance is happening across China's technology sector. Alibaba Group Holding, Tencent Holdings, and dozens of other major firms have begun treating AI tokens as a form of corporate allowance—a new kind of budget that employees must manage alongside their time and attention. The shift reflects a deliberate corporate strategy: as these companies push AI deeper into daily workflows, they are also creating accountability for its use. Employees are no longer simply expected to be productive; they are expected to demonstrate AI literacy, and that literacy is now quantifiable, trackable, and subject to departmental benchmarking.
The driver behind this transformation is economic pressure so intense it has reshaped the entire Chinese AI industry. In May, DeepSeek, a Hangzhou-based artificial intelligence company that had emerged as a serious competitor to larger incumbents, made a move that sent shockwaves through the market: it cut the price of its flagship V4-Pro model by seventy-five percent. The reduction was not a promotional gesture or a temporary discount. It was a statement about the direction of the entire industry. In China, the race to build artificial intelligence capability has become inseparable from the race to make it cheap—so cheap that it becomes a utility, something as routine and unremarkable as electricity or water.
That price war has consequences that ripple far beyond quarterly earnings reports. When AI becomes affordable enough to integrate into nearly every corporate task, the economics of knowledge work begin to shift. Companies can no longer treat AI as a specialized tool deployed by specialists. Instead, it becomes infrastructure, embedded in the daily operations of ordinary employees. And when something becomes infrastructure, it must be managed, measured, and rationed—at least until supply catches up with demand and the scarcity that makes rationing necessary disappears entirely.
The token economy emerging inside Chinese tech firms is, in many ways, a temporary phenomenon. It reflects a moment of transition: AI is cheap enough to be ubiquitous but not yet so cheap that it is free. Employees are being trained to think of AI as a resource with a cost, even if that cost is internal and abstract. Over time, as prices continue to fall and AI becomes truly commodified, the token budgets may disappear. But the habits they are creating—the expectation that workers will integrate AI into their thinking, that productivity will be measured partly through AI consumption, that literacy in these tools is a job requirement—those habits are likely to persist. The token economy is not the future of corporate life. But it is a bridge to that future, and right now, in Beijing and Hangzhou and Shenzhen, millions of workers are learning to cross it.
Bemerkenswerte Zitate
Using AI has become part of almost every task now— Anonymous ByteDance employee
The defining feature of China's AI industry is an accelerating race to make intelligence cheap— Industry observation
Möchten Sie die ganze Geschichte?
Das Original lesen bei South China Morning Post ↗
Economy & Finance