Artificial intelligence (AI) is becoming more expensive, and companies are rethinking their use of this technology. Many AI firms charged low prices at first to attract customers after the success of ChatGPT. Kevin Simback from Delphi Labs calls this time “subsidized intelligence,” meaning investors covered costs, allowing companies to sell AI cheaply.
Now, things are changing. Companies like OpenAI and Anthropic need to make money and aim to attract investors soon. Prices are rising, especially for AI agents. Unlike chatbots that only answer questions, AI agents can perform tasks like booking appointments and writing code. This makes them costly to operate, as one task can run many agents, leading to high charges.
For example, charges are based on tokens, the units used to bill for AI work. A single task can use many more tokens than a simple chat. Furthermore, the demand for computer chips and data centres supporting AI is causing shortages, adding to rising costs.
Some firms are now switching to free AI models or smaller, specific versions to save money. Meanwhile, companies are finding ways to do larger tasks in smaller, cheaper parts. This shows AI is becoming more like a regular product, where cost is very important.
