Not long ago, corporations handed out AI access as corporate perks, encouraging employees to "burn" as many tokens as possible. Today, the situation has changed dramatically: executives are hastily revoking access, imposing strict limits, and revising budgets. Tokens, once considered "free fuel," have become the main line item for unforeseen expenses. Let's break down why this happened and how corporations missed this crisis.
What is a token and why do you need to pay for it
In the context of artificial intelligence, a token is the smallest unit of memory processed by a language model (LLM). One token can correspond to several characters or an entire word. Models use them to process user requests and generate responses. On average, 100 words are broken down into 130–140 tokens. Tokens are divided into two types: input (prompt) and output (model response).
For a long time, tokens were perceived as an abstract unit of information until their monetization began. In March 2023, OpenAI released GPT-4 and set public pricing: $30 per 1 million input tokens and $60 per 1 million output tokens. From that moment, tokens became the unit of measurement for the cost of AI tools.
From 2023 to 2025, the "era of subsidies" lasted: companies distributed subscription services to regular users, but corporate developers were already paying for tokens. Since spring 2026, all major providers have switched to monetizing requests. Anthropic did this in April, and GitHub Copilot in July. The current period is called the beginning of the "tokenapocalypse" — an era where, in addition to computing costs, companies urgently need to include additional expenses in their budgets.
Price drops, demand rises
The paradox is that over the past three years, the actual cost of tokens has decreased. The average price of an input token has fallen by about 98%. However, enterprise AI spending has surged by 320%. The average annual AI budget has increased from $1.2 million in 2024 to $7 million in 2026. The reason is the explosive growth in token usage volumes, especially due to agentic technologies. A simple linear workflow in 2023 cost about $0.04 per interaction, while a request to an agentic system already costs $1.2 (30 times more). For the most complex processes, costs increase by 50 to 500 times.
This is a classic example of the Jevons paradox: a decrease in the price of a resource leads to an increase in its consumption. In the case of AI, the effect has been hypertrophied.
Who miscalculated?
The token crisis is not just on paper. Major companies have already publicly acknowledged the problems.
Uber actively implemented Claude Code and Cursor. In May, the CEO stated that autonomous agents write 10% of the company's code. However, less than a month later, Uber set a limit of $1,500 per month per employee. The CTO admitted that he personally spent $1,200 on tokens in one two-hour demo, exhausting the annual programming budget.
Microsoft gave thousands of developers access to Anthropic's Claude Code in December 2025. By spring, the tool had spread far beyond engineering teams. Six months later, Microsoft ended the experiment, canceling most licenses. Engineers preferred Claude Code over their own Copilot, but its excessive use impacted costs.
Amazon restructured its partnership with Anthropic, switching to pay-per-token instead of compute hours. This could increase costs for Claude models used in Alexa, Kiro, and other services.
Cases of small companies are particularly telling. One firm forgot to set limits on Claude Opus and received a $500 million bill. Another developer spent over $40,000 on tokens in a month. According to analysts, token consumption per developer increased 18.6 times over nine months, and most companies still cannot measure the return on investment.
Not the foggiest future
The term "tokenapocalypse" clearly exaggerates the problem. Large corporations will pay the bills without difficulty, while small companies will adapt through reasonable consumption. Against this backdrop, the Linux Foundation introduced the Tokenomics Foundation project — a new standard for managing AI costs.
My expert opinion: Tokens have become a new line item in corporate budgets, and this is here to stay. Companies that want to keep pace with technology will have to learn to forecast these expenses as easily as they do cloud service costs. Ignoring this trend risks not only financial losses but also strategic uncompetitiveness.