OpenAI Is Closing the Gap on Anthropic in Business AI, Ramp Data Suggests

OpenAI is narrowing the business-user gap with Anthropic in the United States, according to fresh data from spend-management company Ramp. The figures matter because they suggest enterprise AI buying remains highly fluid, with customers switching models as new releases, pricing changes and compliance concerns reshape the market.

Ramp’s latest analysis of more than 70,000 U.S. companies shows Anthropic still leading among its paying business customers, but OpenAI has been growing faster in recent weeks and is regaining ground after losing the lead earlier this year.

The trend offers one of the clearest outside signals yet that the corporate AI market is still expanding rather than settling into a fixed winner-take-all pattern. It also underscores how difficult it remains to measure the real commercial strength of OpenAI and Anthropic ahead of any potential public listings, since both companies keep their financial details private.

What Ramp’s data says about the race

Ramp’s dataset tracks spending across its bill pay and corporate card products, giving analysts a window into how businesses are using AI tools. Because Ramp serves a broad customer base but is especially popular in the tech sector, the numbers are not a complete picture of the market. Still, they provide a useful read on purchasing trends among active corporate buyers.

According to the company’s latest figures, OpenAI once held a clear lead among Ramp’s business users. That changed in May, when Anthropic overtook it. Since then, Anthropic has held the top position, while OpenAI has continued to close the distance.

By July, Anthropic had just under 44% market share among Ramp’s paying business users, while OpenAI was at nearly 40%. In May, the gap was narrower, with Anthropic at 41% and OpenAI at 39%. Ramp says OpenAI has been growing faster in the current quarter so far, though there is still time left for the pattern to shift again.

Why business buyers keep switching

Business AI purchases are not behaving like a stable utility contract. Instead, they appear to move with product launches, model quality, pricing changes, retention policies and compliance concerns.

That volatility helps explain why one company can pull ahead one month and fall behind the next. For many teams, the decision is less about loyalty to a brand than about which model best fits a specific workflow at that moment.

How product releases influence buying

New model launches can quickly reshape purchasing behavior, especially when developers and technical teams are the ones making or influencing the buying decision. A better coding assistant, a more capable reasoning model or a more attractive enterprise package can prompt rapid switching.

Ramp economist Ara Kharazian suggested on X that OpenAI’s newest model is increasingly attractive to developers, while Anthropic’s latest higher-tier offering has faced a softer reception in some use cases. That view is only one interpretation of the market, but it reflects a broader reality: feature changes can move spending quickly.

Ramp economist Ara Kharazian said OpenAI’s newest model appears to be gaining favor among developers, while Anthropic’s premium tier has seen weaker adoption in some scenarios, especially when pricing and data-retention requirements become part of the conversation.

That said, product competition is not the only factor. Enterprise buyers also weigh data handling rules, internal security policies and whether a model is suited for general use or for a narrower set of tasks.

Who is actually represented in Ramp’s numbers?

Ramp’s data reflects a substantial slice of the U.S. business market, but not the whole thing. The company says its analysis includes more than 70,000 American businesses that spend billions of dollars through its card and payment products.

That makes the sample meaningful, but it also introduces limits. Large enterprises that rely on expense systems from providers such as American Express are not included, so the data may skew toward smaller and mid-sized firms as well as technology-forward customers.

In other words, the numbers are best understood as a strong market indicator, not a definitive national census of corporate AI use.

Why the sample matters

The composition of the sample may help explain why OpenAI and Anthropic appear to rise and fall so quickly. Ramp’s customer base includes many firms that are comfortable adopting new software early and experimenting with multiple tools. Those users tend to move faster than traditional procurement-heavy enterprises.

That means the data can capture the earliest shifts in sentiment among technical teams, even if it does not fully reflect the slower-moving purchasing cycles of Fortune 500 companies.

How big is the enterprise AI market becoming?

The broader market is still growing, and that growth may be the most important takeaway from Ramp’s report. Even as OpenAI and Anthropic compete for share, the number of businesses paying for AI tools within Ramp’s customer base has continued to rise.

Ramp says more than half of its customers were paying for AI by March. By July, that figure had climbed to nearly 56%. That suggests the market is expanding fast enough for more than one company to grow at the same time.

For investors, that matters. A market that is both growing and unstable is very different from one where adoption has flattened and only one vendor is taking share. In the first case, both companies may keep adding revenue even if their relative positions fluctuate. In the second, gains for one would more clearly come at the other’s expense.

What the growth means for OpenAI and Anthropic

Both companies have reason to see the data positively. Anthropic still leads among Ramp’s business users, which points to strong commercial traction. OpenAI, meanwhile, is showing signs of renewed momentum after losing the lead in May.

That combination suggests the corporate AI market is far from settled. It is possible for one company to dominate consumer awareness while another performs better in certain enterprise workflows. It is also possible for a temporary advantage to fade as model quality changes or as buyers try alternatives.

Why enterprise AI may never look “sticky” in the usual way

Enterprise software often becomes sticky when companies invest heavily in workflows, integrations and training. AI tools, however, may be less predictable because the underlying technology improves so quickly and the best choice can change from month to month.

That creates a different pattern of loyalty. Instead of locking into one provider for years, businesses may test several vendors, move workloads around or keep multiple subscriptions active at once.

This dynamic appears to be visible in Ramp’s data. The fact that business users can swing between OpenAI and Anthropic suggests that enterprise AI spending is still highly responsive to product launches and pricing decisions.

How regulation can influence adoption

Compliance and data-retention policies can matter as much as raw model capability. Some businesses may avoid products that require broader data retention or more complicated legal review, even if the underlying model is powerful.

Anthropic drew attention when it told users of one of its premium offerings that data would need to be retained for 30 days. For regulated industries or privacy-conscious customers, that kind of policy can affect procurement decisions just as much as performance benchmarks.

Timeline of the business-user race

The competition between OpenAI and Anthropic among Ramp customers has moved quickly this year. The following timeline shows the major shifts based on Ramp’s reported figures.

What investors should watch next

For investors and analysts, the key question is not only which company has the largest share today, but whether either one can convert temporary user enthusiasm into durable business revenue.

A few indicators will be especially important over the coming months:

• Whether OpenAI’s faster recent growth continues through the end of the quarter
• Whether Anthropic can defend its lead among business users
• How many companies keep paying for AI overall as adoption expands
• Whether compliance or retention policies alter buying behavior
• Whether new model releases trigger another round of switching

Because Ramp declined to share actual dollar volumes, the report does not reveal how much each company is earning from this customer segment. Percentage share can show direction, but not absolute revenue. Still, the changes are significant because they imply that enterprise AI demand is both real and contested.

What this means for the broader AI market

The latest Ramp data points to a sector that is still in motion. OpenAI may be regaining momentum with business users, but Anthropic has not only held its lead in the most recent reporting period; it has also demonstrated that it can win over corporate buyers who were previously spending elsewhere.

That is important for a market that is still trying to determine whether a few dominant labs will control enterprise adoption or whether buyers will continue splitting their spend across multiple providers. At the moment, the evidence supports the latter view.

The most striking signal from Ramp is not just that one lab is ahead of the other. It is that both are benefiting from a bigger tide of AI adoption inside businesses. Even if market share changes again, the overall pie appears to be getting larger.

That leaves both companies with an opportunity and a risk. The opportunity is obvious: more businesses are paying for AI every month. The risk is that customers remain willing to move quickly when another model looks better, cheaper or easier to use.

Until OpenAI and Anthropic approach public market disclosure, outside observers will continue relying on partial indicators like Ramp’s. For now, those indicators suggest a competitive market, not a settled one — and one where OpenAI is once again pressing Anthropic from behind.