AI Companies Are Losing Billions: Their Simple Plan to Recover? Pricier Subscriptions

Using ChatGPT, Claude, Copilot, or Gemini to generate AI-powered answers to our questions feels like second nature now. For many users, these tools seem free. But behind the scenes, those answers — let alone AI-generated images or videos — come at a staggering cost to the companies operating these systems. Just how much? That’s a question these companies are reluctant to answer.

Burning Through Cash

AI companies like OpenAI are burning money at an unprecedented rate. OpenAI’s CEO, Sam Altman, recently revealed that the company isn’t expecting profitability until 2029. Until then, they will continue operating at a massive loss. For instance, OpenAI’s projected losses for 2024 alone are estimated at $5 billion. This isn’t unique to OpenAI — the operational costs for maintaining AI systems are astronomical across the board.

The True Cost of AI Remains Elusive

While companies remain tight-lipped about their internal costs, a report by The Washington Post in mid-2023 highlighted a critical truth: AI chatbots lose money with every single use. Independent studies in 2023 suggested that each query on ChatGPT cost around $0.36. OpenAI reportedly spent $700,000 daily just to keep its language models running.

More Complex Models, Higher Costs

The release of OpenAI’s o1 model, designed to “reason,” underscores the escalating costs. Unlike earlier versions, this model generates multiple responses, evaluates them, and selects the best one to display. While this results in more accurate answers, it significantly raises operating costs. For users, the introduction of features like o1 has already led to more expensive subscription tiers. OpenAI’s $200-a-month subscription for unrestricted access to its reasoning model may just be the beginning.

Images and Videos: Even Costlier

Text generation might be the least resource-intensive task for AI systems, but it’s still expensive. Generating images is significantly costlier, and creating videos increases the cost exponentially. The complexity of the request, server efficiency, and the resolution of…