AI Agents' Token Consumption 5 Times Higher Than Human's - Research
AI agents, despite being an early-stage technology, are already significantly altering the consumption patterns of computational resources and tokens. According to data from OpenAI, OpenRouter, and Similarweb cited in a report by Andreessen Horowitz (a16z), agents consume nearly five times more tokens than humans, with their usage growing approximately 14-fold since February 2026.
The data reveals a widening gap between typical and leading-edge companies, especially in the tech sector. Companies in the top decile generate nearly 12 times more tokens than average users, with their volume surging 32.5 times compared to just over a year ago. These top performers are moving beyond basic chatbots to utilize more advanced tools like plugins, skills, and Codex at much higher rates.
A key finding is that over 85% of tokens consumed by AI agents are for cached prompts, reflecting their operational model of repeatedly reading, writing, and executing tasks within a retained context. While cached tokens are cheaper, they require substantial memory, potentially sustaining high demand for high-bandwidth memory (HBM).
The rise of agents is impacting the broader market. Traffic to traditional automation platforms like N8N, Zapier, and Make has seen double-digit declines in recent weeks, while AI-native platforms like Gumloop are growing. Analysts caution it's too early to declare the decline of traditional platforms, as they too are integrating AI, but the trend indicates autonomous agents are already reshaping the automation market.
This shift aligns with predictions of billions of personal AI agents emerging in the coming years, potentially forming a vast agent economy. However, researchers have also flagged risks, noting autonomous agents took undesirable or potentially harmful actions in 80% of tested scenarios.
cryptonews.ruHá 4m
