Author: Kuri, Deep Tide TechFlow

Recently, you may have noticed one thing: Everyone has started discussing what Tokens should be called.

Professor Yang Bin from Tsinghua University published an article titled "Determining the Chinese translation for Token is already urgent"; related translation questions on Zhihu have garnered 250,000 views, with comments providing ideas.

For the past two to three years, the AI circle in China has directly used the term Token, with no one feeling that there was an issue. Why is there suddenly a need for a Chinese name?

The direct reason may be that, after this year's Spring Festival, the general public learned for the first time that Tokens cost money.

OpenClaw allows AI to transition from chatting to working; a single task can burn tens of thousands of Tokens, and the bills are skyrocketing; cloud vendors are all announcing price hikes, with pricing units also being Tokens.

At the same time, Tokens have begun to appear in places where they previously should not.

At the GTC conference, NVIDIA President Jensen Huang mentioned that there are already people in Silicon Valley asking in interviews, "How many Tokens does this job pay?", and he suggested incorporating Tokens into engineers' salaries;

OpenAI founder Sam Altman went even further, believing that Tokens will replace universal basic income, with everyone receiving not money, but computing power.

Data from the National Bureau of Statistics indicates that China's daily Token consumption has risen from 100 billion at the beginning of 2024 to over 40 trillion by September 2025, reaching 180 trillion this February. The People's Daily published an article earlier this year titled "A Chat about Tokens", explaining what the term means to readers.

Once a technical term enters cloud service bills, recruitment salary packages, and official statistical scopes, it can no longer continue to be called by its English name.

The question is, what should it be called?

If this were merely a translation issue, there would have been answers long ago. In 2021, the academic community in China assigned a name to Token: Token.

But no one paid attention because, at that time, Token was still just an internal term within the tech circle.

Now it's different.

The term Token itself is a universal container; previously, people in the cryptocurrency space referred to it as a digital currency, those in security called it a token, and those in AI called it a linguistic unit. Depending on which direction the Chinese translation leans, it belongs to that domain.

Thus, a battle for naming Tokens has begun.

Business Needs Discourse Power

How a word is translated is usually the concern of linguists. But among those involved in the naming this time, there are almost no linguists.

The most prominent name currently is "Smart Unit".

The one promoting it the hardest is an AI media company called "New Smart Unit". If the Chinese name for Token is set as "Smart Unit", this company's brand name would overlap with the industry's foundational term, essentially meaning every article discussing Tokens is helping advertise for them for free.

They candidly wrote at the end of their promotional article: "We suggest translating Token into the new industry consensus: Smart Unit, leaving the word 'new' for us."

According to the same article, the founder of Baichuan Intelligence, Wang Xiaochuan, commented: "Calling it Smart Unit is quite good."

For those who work on large models, it's certainly good if Tokens are called Smart Units. Each computation output is no longer a billing unit but rather an "intelligent basic unit".

Selling Tokens is selling traffic, while selling Smart Units means selling intelligence; the valuation story is entirely different.

Professor Yang Bin from Tsinghua University proposed "Model Unit", with "Model" corresponding to models; whoever owns the large models controls the production rights of "Model Units". The name leaning towards models means that the pricing power shifts to the model companies.

There are also proponents of calling it "Symbol Unit", returning to the most fundamental definition in computer science, stating that a Token is simply a unit of symbol processing, unrelated to intelligence and models.

This is technically the cleanest, but the proposer is an independent technical author without a company backing or capital support, resulting in minimal volume in this discussion.

Whichever direction the name leans, the industry's narrative will follow that direction, and money will flow in that direction.

A distant example is when Facebook rebranded as Meta, the term "metaverse" went from a sci-fi concept to a company's valuation story; a closer example is that China consumes 180 trillion Tokens daily, leading the world, yet what this term is called, how it is defined, and who defines it has yet to have a conclusion...

The world's largest consumer of Tokens has yet to figure out what to call its own consumed items.

However, this term actually already has a Chinese name.

In 2021, Professor Qiu Xipeng from Fudan University's School of Computer Science translated Token as "Word Unit", and the academic community accepted it, including it in textbooks. At that time, no one discussed this matter because, at that time, Tokens were not valuable.

Now Tokens have become valuable.

They are the billing unit for cloud services, the revenue source for large model companies, and the core indicator for the state to measure the scale of the AI industry. Hence, the media has come, big names have appeared, and professors have joined, each bringing their preferred names and rationales.

Translation has never been the issue. The question is when did this term start being valuable.

Jensen Huang did not participate in the Chinese naming discussion at GTC. He did a simpler thing: raised a championship belt marked with Token King and declared that data centers are Token factories.

Whoever produces Tokens defines Tokens. What the name is, he does not care.

Token, Enclosing Territories, and Minting

Thus, the aspect of this matter that is truly worth considering seriously is not which translation is better.

Once the term "Calorie" was established, the entire pricing, labeling, and regulatory system of the food industry was built around it. After the definition of "Traffic" was established in the Chinese telecom industry, operators began billing by traffic, competing by traffic, and designing packages by traffic; the entire business model has revolved around these two words for over a decade.

Token is now walking the same path.

It has already become the billing unit for cloud services, a revenue metric for large model companies, and a core indicator at the national level to measure the scale of the AI industry. The VC circle has even started discussing whether investments can be given directly in Tokens.

Once a word becomes a measure of money, naming it is no longer translation, but minting.

If it is called "Smart Unit", the minting power belongs to AI narratives; whoever tells the intelligent story benefits. If it is called "Model Unit", the minting power belongs to model companies; whoever has the large models prints the money. If it is called "Symbol Unit", the minting power returns to the technology itself, but the technology itself will not speak for itself.

The "Word Unit" defined by academia in 2021 did not attract attention, not because the translation was incorrect, but because at that time this "coin" was still not valuable.

Now it is valuable, and everyone wants to carve their name on it.