(TNND) — An artificial intelligence price war may be brewing between OpenAI and Anthropic, with the maker of ChatGPT considering “significant” price cuts for enterprise customers, the Wall Street Journal reported.
OpenAI also expects Anthropic to reduce prices, the report said, citing unnamed people familiar with the matter.
The Journal recently reported that companies using AI are feeling the strain on their budgets and are looking for ways to retire their use of AI or use powerful new technologies more purposefully and efficiently.
OpenAI CEO Sam Altman acknowledged that he has heard cost concerns from customers, with some enterprise users burning through their annual AI budgets in a fraction of the fiscal year.
“At the beginning of this year, people were completely satisfied with how much they were spending, and all of a sudden, this issue that wasn’t even talked about became a big issue,” Altman said at the Intelligence at Work event earlier this month.
Altman said OpenAI is working on ways to deliver “more value for less” to enterprise customers.
File – Open AI CEO Sam Altman speaks at Snowflake Summit 2025 at Moscone Center in San Francisco on June 2, 2025. (Photo by Justin Sullivan/Getty Images)
Potential price reductions relate to OpenAI’s enterprise products. These are AI tools that customers use to improve or streamline their products, services, and workflows.
The potential price cuts, which the Journal describes as “significant” in scope, seem unlikely to affect the flat monthly fee that individual users pay for upgraded versions of ChatGPT.
This reduction applies to the cost of AI tokens used by OpenAI Business customers.
Many of OpenAI’s enterprise customers pay based on token usage. That is, the cost increases as more employees enter prompts and get responses. This is similar to how a utility company’s customers pay for the amount of electricity or water they use.
“The best way to describe it is just a measure of the computational resources required to perform operations on a large language model,” said Anton Dervla, an AI expert and co-director of the Johns Hopkins Institute for Autonomy Assurance.
OpenAI said at a recent event that the company serves 2 million enterprise customers.
And Altman said the company’s top token consumers spend about 100 billion a month, a rapid increase from about 100,000 a month for token leaders just about six years ago.
“Although it’s being deployed and adopted at scale, it’s important to remember that this is still a technology in its infancy,” Dabula said.
Derbra said companies are experimenting with different applications of AI, evaluating costs and determining where the technology will provide the most value.
He described this as an “establishment” stage approaching a “disruption period” where early adopters react to what works and what doesn’t.
“So there are massive adjustments being made, especially as businesses look at their bills for using this technology,” Dabula said.
The Journal reported that some companies expect their AI bills to double or triple as they learn more about their use of AI.
Mark Muro, a senior fellow at Brookings Metro and an expert on the digital economy, questioned what the price cuts would mean for OpenAI and Anthropic as both technology companies seek to go public.
“As we spend heavily on development, training and data centers, soft pricing and the potential for price competition raise new questions about our ability to monetize,” Muro said in an email. “Still, the potential for price competition points to a deeper problem. Do these companies have a moat? Are they essentially selling the same product? I think this is the fundamental question for companies to step up. Because products are interchangeable and easily switched, companies become more like social media apps, and more susceptible to intense competition when they need to keep spending.”
Dabula said he does not expect price competition to slow the development or deployment of AI.
He said business demand remains strong and the main driver of growth for AI developers remains investor capital rather than customer revenue.
“If there is a legitimate business case, I think such a development will happen anyway,” Dabula said.
