Tokenomics: Why making AI pay is tricky
Buyers of AI services are struggling to control costs and sellers are not sure how much to charge.
ChatGPT, Claude and Gemini are free AI services, but firms behind them want to charge for enhanced features. Agentic AI, tailored for specific tasks, poses pricing challenges. Breaking prompts into tokens, the cost is unpredictable as the same prompt may yield different outputs from different models. Goldman Sachs forecasts a surge in token usage, but companies struggle to gauge their consumption, often facing surprise bills.
Microsoft has limited engineers' access to certain tools, while Uber spent through its AI coding budget quickly. Companies are exploring workarounds, like flat fee personal accounts, but this may not be sustainable. Larger providers may eventually clamp down on these practices. Precision in prompts and careful task selection are recommended.
When AI is integrated into products for many users, costs can escalate rapidly. While token costs are uncertain, AI may deliver better results for the investment. However, these costs will likely be passed onto customers. Companies are debating various pricing strategies, but changes in LLM providers' pricing could further complicate matters.
Written by urgent.news from BBC World's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
- Tokenomics: Why making AI pay is tricky bbc.co.uk
