Just two years ago, the market for generative artificial intelligence resembled an arms race. Technology companies were outdoing each other in presenting ever more advanced models, with the number of parameters, the quality of the response or the ability to perform complex tasks determining the advantage. Today, however, there are increasing indications that the AI industry is entering a new phase of development. This phase may be less spectacular technologically, but much more important from a business perspective.
Reports that OpenAI is considering lowering the price of tokens used to bill AI services may be the first clear sign of the start of a price war in the artificial intelligence market. This is a phenomenon we have seen many times before in the technology sector – from cloud services to streaming platforms. In each case, it has led to a change in the rules of the game.
Competition is becoming more demanding
Until recently, OpenAI was in an extremely comfortable position. ChatGPT had built global brand recognition and GPT models had become the natural choice for many companies when implementing generative AI-based solutions.
Today, the situation is different. Anthropic, the developer of Claude models, is no longer seen as a niche competitor. The company is gaining a stronger foothold in the enterprise segment, where it is not only the capabilities of the model that count, but also the predictability of performance, security and compliance with organisational requirements.
At the same time, companies are becoming less and less dependent on a single supplier. A multi-model strategy is gaining popularity, where companies use OpenAI, Anthropic, Google or open source models in parallel. This means that customer loyalty is declining and price is becoming one of the key elements of competitive advantage.
This is a fundamental change. In the first phase of the AI market, the most important question was: ‘Which model is best?’ Today, companies are increasingly asking: “Which model will provide the best value for money?”.
AI is starting to resemble the cloud market
The history of the technology sector shows that as the market matures, technological advantage gradually gives way to price competition. We have seen a similar process with cloud computing services.
Initially, market leaders were able to maintain high margins due to their technological advantage and limited number of competitors. However, as the infrastructure developed and the number of providers increased, prices started to fall steadily.
In the case of artificial intelligence, a similar mechanism is taking place. The cost of inference, i.e. the generation of responses by models, is decreasing as hardware evolves, architectures are optimised and suppliers become increasingly large. At the same time, the quality differences between the leading models are no longer as pronounced as they were just a year or two ago.
This is causing the market to start treating AI models more and more like an infrastructure service. What matters to many customers is no longer the model itself, but the ability to deploy quickly, integrate with existing systems and the predictable cost of ownership.
The pressure is not just coming from Anthropic
While the media focus is on the rivalry between OpenAI and Anthropic, the biggest challenge for market leaders may come from a completely different direction.
The increasing quality of open source models is causing more and more organisations to consider building their own AI environments. Projects such as Llama, Mistral and DeepSeek have proven that high-quality models do not have to be available exclusively through the commercial APIs of the major vendors.
For enterprises, this means more freedom of choice, and for companies like OpenAI, additional pressure to reduce the cost of services. Even if the majority of organisations continue to use commercial solutions, the mere possibility of an alternative affects the negotiating position of customers.
In practice, this means that competition in the AI market is becoming multidimensional. It is no longer just about the quality of the models, but also about cost, availability, integration and data control.
IPOs and the battle for scale
News of OpenAI’s potential IPO adds an extra dimension to the whole situation. According to reports, the company could be aiming for a valuation of up to a trillion dollars. Achieving such a value, however, requires more than just impressive technological demonstrations.
Above all, investors expect the ability to build a dominant market position. From this perspective, lowering prices may be a rational part of the strategy. Smaller margins in the short term can be compensated by faster customer growth and greater market share.
This approach is reminiscent of the strategy previously used by major technology companies. Amazon has for years focused on expansion and market capture at the expense of profitability, with investors rewarding it for its scale of operations and growth potential.
A similar mechanism may be at work today for OpenAI.
The AI market is entering a new era
Possible token price reductions are more than a decision about one company’s sales policy. It signals that the market for generative artificial intelligence is beginning to mature.
Over the past years, advantage has been built primarily through technological innovation. Today, deployment economics, cost of ownership and the ability to scale the business are increasingly important. Success will not only be determined by who can create the most advanced model, but also by who can deliver the technology most efficiently and cheaply.
If OpenAI does indeed decide to cut prices, this could mark the beginning of a new phase in the development of the entire sector. A stage where artificial intelligence will cease to be solely a technological revolution and become a fully-fledged, competitive market fighting for customers, margins and participation in the global digital economy.
