The AI industry faces an uncomfortable reality: even as models become more capable and efficient, usage costs are climbing sharply for businesses deploying advanced agents. The culprit is a phenomenon called token amplification, where complex multi-step tasks consume millions of processing units behind a single user request.

What You Need to Know

AI token costs grow exponentially with conversation length and task complexity due to mandatory reprocessing of all previous context. A single agentic report generation that uses Excel and CRM data can burn through millions of tokens, translating to $1 to $10 per task. When scheduled repeatedly, such tasks can cost thousands per month per user. Companies like Anthropic and Microsoft have moved to usage-based pricing after many clients experienced sticker shock.

The Mechanics of Token Amplification

Token amplification arises from a fundamental limitation of current AI models: they lack persistent memory. Each interaction in a conversation requires reprocessing the entire exchange, including all prior messages, uploaded files and model responses. This means that a simple follow-up query that costs 500 tokens may trigger an additional 600 tokens because it must reprocess the previous 500. Over a long conversation, costs accumulate dramatically. For agentic tasks that involve accessing Excel spreadsheets and a CRM system to generate customer reports, the token count can explode into the millions for a single workflow.

  • Conversation length: Each new question reprocesses the entire history, so the 10th question costs far more than the first.
  • Task stages: Agentic workflows break a report into multiple stages, each requiring separate token expenditures for data retrieval, analysis and web searches.
  • Retries and corrections: When a model produces a flawed result, repeating the process multiplies token usage further.

Why This Matters

The financial implications are severe for enterprises adopting AI agents. A single report generation task might use 280,000 tokens at a cost of $1. That seems trivial until it runs every 15 minutes for a dashboard: $96 per day or nearly $2,880 monthly. For more complex reports requiring millions of tokens, monthly costs can exceed $28,800 for one department. This has forced companies like Uber and Amazon to rein in AI spending, as agentic workloads can become costlier than human employees.

For AI providers, the challenge is twofold. They must reduce per-token costs through techniques like prompt caching, model routing and batch processing while also helping clients manage token consumption. The move to usage-based pricing by Anthropic, Microsoft and others in April reflected this reality, as fixed-rate plans proved unsustainable when power users burned through vast token allocations.

A New Cost Calculus

Businesses now treat token budgets as a critical operational metric, akin to cloud compute costs. A prompt redesign can become a margin event, and a poorly designed agent loop risks running up an exorbitant bill with no output. The era of treating AI as a cheap utility is over for enterprise deployments.