OpenAI is now growing its spending from enterprise customers at a pace that matches, and in some cases exceeds, Anthropic’s, according to data from Ramp, a corporate payments and expense platform that tracks business spending on technology. TechCrunch first reported the data on Wednesday, and the shift is being read across the industry as evidence that the competitive balance in corporate AI has moved.
The Ramp data shows corporate spending on OpenAI’s products accelerating through recent months, narrowing a gap that had favored Anthropic since last year. Anthropic had built a reputation among enterprise buyers for data security and conservative deployment practices, and it used that reputation to win contracts with banks, insurers, and government-adjacent businesses that were wary of sending their data to a less trusted provider. OpenAI’s gains suggest that reputation advantage is eroding.
The change has several causes. OpenAI has expanded the security certifications and compliance features it offers to business customers, and it has introduced products aimed at regulated industries, where data handling requirements are strict. The company has also invested heavily in its sales organization, building a team that can compete with Anthropic’s on the ground. And OpenAI’s brand, still the best-known in AI, appears to be pulling enterprise buyers who were waiting for the company to address their concerns.
The competition between the two companies has become the defining rivalry in commercial AI, and the enterprise market is its most important battlefield. Consumer subscriptions and developer APIs generate headlines, but corporate contracts are where the real revenue is, and both companies have built their growth plans around winning business from large organizations. For years, Anthropic had the clear edge in that market; the Ramp data suggests the edge has narrowed, if not disappeared.
The same week brought a move from Anthropic aimed at defending its position. Reuters reported that Anthropic is planning to change its enterprise data retention policy, allowing customers to store the data processed by its AI systems on their own infrastructure. The change is a direct response to OpenAI’s pledge to offer zero data retention for its business customers, a commitment that has been one of OpenAI’s most effective tools in winning over cautious buyers.
Data retention has become the flashpoint of enterprise AI. Companies that process sensitive information, from legal documents to patient records, have demanded assurances about what happens to the data they feed into AI models: where it is stored, who can see it, and whether it is used to train future models. Anthropic built its enterprise franchise on promising strong protections; OpenAI’s zero-retention pledge went further, and the new Anthropic policy appears designed to match it.
The shift in the competitive balance has consequences beyond the two companies. Enterprise AI buyers are now negotiating from a stronger position, with two large suppliers offering comparable data protections and competing on price. The spending data suggests that price competition is real: as OpenAI has closed the security gap, the decision for corporate buyers has increasingly come down to cost and performance, the categories where OpenAI’s models have traditionally been competitive.
The rivalry is also shaping product development. Both companies have introduced enterprise-specific features at a rapid pace, from fine-tuning options to dedicated deployment environments, and the data retention fight is part of a broader contest over who can offer the most control to corporate customers. Analysts who follow the market say the competition is good for buyers, who are getting better products at lower prices, but it is squeezing margins for the suppliers, both of which spend heavily on computing capacity and research.
The Ramp data has limitations. It measures spending that flows through its platform, which skews toward smaller companies and does not capture the largest enterprise deals, most of which are negotiated directly. But the direction of the trend matches what industry executives describe anecdotally: OpenAI has become a more credible enterprise supplier, and its sales pipeline has grown accordingly. The data is a useful proxy, if not a complete picture.
The broader question is whether the enterprise market will support two companies of this scale. Both OpenAI and Anthropic are spending billions on the computing power needed to serve corporate customers, and both are pricing their products below what the underlying costs suggest. The market is growing fast enough that both can win for now, but the data retention arms race is a sign of how intense the competition has become. Each new feature, each new policy, is aimed at the other company as much as at the customer.
For corporate technology buyers, the effect is welcome. Two well-funded suppliers competing on security and price means better terms for the companies that sign contracts, and the pressure on both providers to prove their data handling is safe has raised standards across the industry. The question of who wins the enterprise AI market will be settled over years, by the same forces visible this week: spending data, retention policies, and the slow accumulation of trust.
