OpenAI Eyes Steep Price Cuts as Anthropic Competition Intensifies

A Wall Street Journal story citing people familiar with the matter says that OpenAI is thinking about lowering the prices it charges for tokens, which are used to track and bill AI usage.

This is because it wants to compete more fiercely with Anthropic. The talks are still going on, but the time is perfect: OpenAI is said to be thinking about making the move because they think Anthropic will also cut prices, which makes the possible action seem proactive instead of reactive.

There is a chance of a price war between the two biggest AI companies just as they are getting ready to go public. This means that investors who are looking at pre-IPO valuations that have already hit hundreds of billions of dollars need to be aware of margin visibility.

Why This Is Happening Now

There is clear economic pressure. Anthropic's software creation tool, Claude Code, has made real progress in the enterprise market, which OpenAI has long seen as its main growth driver.

When enterprise users look at large-scale AI deployment, they decide what to buy based on both the model's capabilities and the total cost of ownership. As the use of AI grows from a pilot program to full production across big companies, small costs add up quickly. CEO Sam Altman has said in public that AI costs have become "a huge issue" for customers.

The story of the IPO is being told in the business world. Both OpenAI, which is said to be worth more than $850 billion, and Anthropic, which just raised about $965 billion, have said that their enterprise income is the best way to tell if they are commercialising at scale.

Anthropic's recent $65 billion fundraising round and the news of its Singapore Applied AI Lab show that the company is speeding up operations in businesses around the world. Cutting the price of the core product as OpenAI's response is the most direct way to protect and grow corporate market share before either company's shares can be bought and sold by anyone.

The Economics of an AI Price War

Both companies are having a hard time because cutting prices on small items cuts into margins in a business that is already losing money at the base level. It costs a lot for OpenAI and Anthropic to keep building the computers they need to train and run complex models.

As part of its Colossus deal, OpenAI pays xAI $1.25 billion a month for access to its computers. This agreement lasts until May 2029. Anthropic just recently said that it will be adding Colossus 2 to the same relationship with Colossus. Price cuts aren't possible for either company unless they agree to lower economics per unit or reach a large enough size that fixed infrastructure costs are spread across a large enough usage volume to keep the overall margin.

The way competition works has been seen before in cloud computing. In the middle of the 2010s, AWS, Google Cloud, and Azure started competing hard on price. This made unit economics tighten across the industry, but usage volumes grew a lot, eventually favouring the players with the lowest infrastructure costs and the most enterprise integrations.

Both OpenAI and Anthropic are hoping that the same thing will happen with AI tokens: lower prices will make people more likely to buy them, and that will lead to more usage. Eventually, more usage will lead to a margin recovery that makes the short-term compression worth it.

What's at Stake for the Enterprise Market

Enterprises don't just decide which AI to use based on coin price; they also look at how hard it is to integrate, how reliable the models are, what tools developers have access to, how well they can follow regulations, and how stable their relationships with vendors are.

Anthropic's success in a high-value enterprise use case where developers are making a lot of tokens every day and adding up real costs at scale is a good example of Claude Code's popularity in software development. If OpenAI cuts its prices in a way that makes similar code-generation models much cheaper, it will directly compete with Claude Code's lower prices in that market area.

ChatGPT Enterprise and OpenAI's enterprise API are both used by different types of customers. ChatGPT Enterprise is used by organisations to handle AI assistant deployments, while OpenAI's enterprise API is used by builders to access large language models. Price cuts could happen in different ways for each group: aggressive token cuts for API customers that are aimed at Claude Code's developer base, while ChatGPT business's business licensing structures stay mostly the same.

The IPO Dimension

The price talks are taking place against a background that makes the margin trend very clear. Both OpenAI and Anthropic are getting ready to go public. When they do, institutional investors will look closely at the unit economics of AI model adoption, just like they do with cloud infrastructure companies.

If there is a price war that cuts short-term profit margins, even if it is strategically necessary to gain market share in the long term, there needs to be a clear story about how to get back to positive margins as the business grows.

The fact that OpenAI is worth more than $850 billion suggests that the market sees it as having a dominant business position in enterprise AI. Similar hopes can be seen in Anthropic's $965 billion. If there is a long-lasting price war, access to cutting-edge AI models will become a low-margin commodity business instead of a high-margin platform business. Companies at both values will have to deal with the same structural problem that cloud infrastructure companies did ten years ago.

OpenAI's plan to lower the price of its tokens is a sign of competition, a move to protect its market share, and a move to place itself for an IPO. Anthropic is really getting a lot of success with businesses with Claude Code, and the best way to stop them is to match or beat their prices.

The risk for both companies is that the price war will make it harder to defend the valuations they are starting with for their IPOs as each cut in prices leads to a reaction. This will make margins tighten across the industry. As a strategic bet, lower prices will lead to enough adoption to finally create the margin structure that makes frontier AI development economics make sense.

In the cloud, that bet has paid off. At the rate that infrastructure costs are rising now, not even the company's IPO prospectus can fully answer the question of whether it works in AI model distribution.