{
  "id": 109138,
  "title": "Tokenomics: Why making AI pay is tricky",
  "url": "https://urgent.news/2026/08/03/tokenomics-why-making-ai-pay-is-tricky-109138",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-03T23:21:10.000Z",
  "source": {
    "name": "BBC News",
    "slug": "bbc-news",
    "url": "https://www.bbc.co.uk/news/articles/c872r52x7jgo?at_medium=RSS&at_campaign=rss"
  },
  "original_language": "en",
  "account": "The economics of AI tokens and agentic AI pose significant challenges in determining appropriate pricing models. While firms like Microsoft, Google, and Anthropic have invested heavily in developing Large Language Models (LLMs), they seek to recoup their investments through paid versions of their AI services. These paid versions often include additional features for tasks such as coding or billing. Third-party firms are also building and selling services based on AI agents trained to perform specific tasks, but setting a price for these services proves difficult.\n\nThe cost of individual tokens, or credits used to pay for them, has decreased in recent years, but the number of tokens consumed by businesses and consumers has surged. Goldman Sachs predicts that external token consumption will increase 24 times between 2026 and 2030, reaching 120 quadrillion tokens per month as companies increasingly utilize AI agents. However, businesses often struggle to grasp the extent of token usage, leading to unexpected costs when they receive their monthly bills.\n\nMicrosoft, for instance, has limited the use of third-party coding tools by its engineers, while Uber reportedly exhausted its AI coding token budget within months earlier this year. Will Venters, an associate professor at the London School of Economics, highlights the non-deterministic nature of AI token costs, making it challenging for companies to manage expenses. Companies are exploring various solutions, such as using flat fee personal accounts, but the reliance on these accounts may not be sustainable as the big AI platforms face pressure to demonstrate profitability.\n\nCompanies should also carefully consider the choice of AI models and the precision of their prompts to optimize costs. However, when AI is embedded into products for broader use, the costs can quickly escalate. Managers may need to account for tokens not only for core software development but also for tasks like testing, security, and implementing guardrails. The unpredictability of token costs becomes even more pronounced when employing multiple AI agents, as expanding the workforce would involve complex discussions about headcount and hiring.\n\nWhile the cost of AI tokens may seem unpredictable, it could ultimately lead to better results. However, companies still need to pass these costs onto their customers, and finding the right pricing structure remains a challenge. Organizations are debating options such as raising prices, charging based on results, or bundling incidents, but these solutions may quickly become obsolete if the pricing strategies of LLM providers change. Until a more effective pricing model is established, companies will continue grappling with the costs of AI token consumption.",
  "summary": "Buyers of AI services are struggling to control costs and sellers are not sure how much to charge.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 2,
    "also_reported_by": [
      {
        "outlet": "BBC Technology",
        "title": "Tokenomics: Why making AI pay is tricky",
        "url": "https://urgent.news/2026/08/03/tokenomics-why-making-ai-pay-is-tricky",
        "published": "2026-08-03T23:21:10.000Z"
      }
    ]
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}