The Enduring Paradox of the AI Economy: Models Improve but Costs Can Spiral Out of Control

This article from Tom's Hardware analyzes a paradox in the AI economy: while AI models become more capable and efficient, usage costs are soaring due to 'token amplification.' As models handle complex agentic tasks—like generating reports by querying multiple systems—they consume millions of tokens per request, far more than simple Q&A. For example, a single task might cost $1, but if run every 15 minutes, it can reach $2,880 per month. More complex tasks can cost $28,800. This has led companies like Anthropic, Microsoft, and OpenAI to shift to usage-based billing, causing sticker shock for developers. Major firms like Uber, Microsoft, Amazon, and Walmart are curbing AI spend. The article warns that for agent-heavy companies, prompt redesign is a margin event, and poorly bound agent loops are outages with credit cards attached.