Tokens Become "New Money" in the AI Era, Companies Forced to Change How They Calculate Costs

JAKARTA - In the era of artificial intelligence, tokens are starting to rise in class. No longer just a technical unit behind the AI model, but increasingly treated as a core resource that determines costs, efficiency, and business strategies. Citing a report by China Daily, Thursday, April 2, industry players are now encouraging companies to include token consumption in budgets and operational decisions.

The spotlight on the token intensified after Nvidia CEO Jensen Huang called it a "new commodity" in his keynote speech at the recent GTC conference. According to the Nvidia boss, tokens are starting to function like currencies that support recruitment, budgeting, and productivity.

In the world of AI, tokens can be understood as the basic unit that the machine uses to read and process information. One Chinese character requires about 0.7 tokens. A high-resolution image can cost thousands of tokens. While a 15-second video can take about 300,000 tokens.

The number continues to soar. The China National Data Administration recorded that, as of March, the average daily consumption of tokens in China had reached 140 trillion. This number has increased more than 1,000 times compared to the beginning of 2024.

OpenRouter data, as reported by China Daily, also shows the world's top four models based on token usage in the week of March 16-22 all come from Chinese companies. The models are Xiaomi's MiMo-V2 Pro, StepFun's Step 3.5 Flash, MiniMax-M2.5, and DeepSeek-V3.2. Total weekly token usage of China's large models also rose 56.91 percent compared to the previous week, and for the third consecutive week surpassed models from the United States.

International Data Corporation (IDC) assessed that this surge was supported by cost advantages. Thanks to lower green energy prices, the cost of China's main model token is only one-sixth to one-tenth compared to foreign competitors. IDC China Research Director Lu Yanxia said that the price advantage gives greater purchasing power to each Chinese company's token budget.

"Just as companies in the industrial era have to budget for electricity, companies in the AI era must learn to budget for tokens," Lu said, quoted by China Daily.

According to Lu, the use of tokens in the future can be included in core financial indicators. This means that companies may begin to know terms such as token budgets, token quotas per department, to returns on tokens.

A similar push came from Zhang Ran, an analyst at China CITIC Bank's information technology management department. He assessed that it was time for companies to build an AI evaluation framework centered on tokens, covering costs, efficiency, quality, returns, and strategies.

A number of major players are starting to move. Alibaba Group this month formed the Alibaba Token Hub (ATH) business unit, which is directly led by CEO Eddie Wu. The token is a strategic core indicator, and all operations are directed towards its creation, distribution, and implementation.

Chinese technology company Inspur, referring to the same source, also sees the same direction. The company's AI strategy head, Liu Jun, said that the system architecture needs to be rethought with the main goal of lowering the cost of tokens. According to him, the AI system that has been built large and comprehensive must now be disassembled again to be more lean and efficient.

Liu emphasized that in order for AI to truly expand and be more accessible, the cost of the token still needs to fall even further from its current position. Because, the cost of the token also determines the competitiveness and profitability of AI agents.