This marks the first time Anthropic has surpassed OpenAI in terms of enterprise subscriptions. Ramp's chief economist noted that historical data shows the months when Anthropic faced government crackdowns coincided with its highest rates of enterprise adoption. "When your model is specifically labeled as 'too dangerous to use,' that itself creates a significant halo effect."
Ongoing friction with the Trump administration has not only failed to hinder Anthropic’s business performance but appears to have added a distinctive luster to its brand. Recent data shows that the AI company is rapidly eroding its rivals’ enterprise market share at an astonishing pace.
On June 16, tech media outlet TechCrunch reported that, according to corporate spending data platform Ramp, Anthropic’s share of the enterprise AI subscription market rose to 41% in May—a 2.5 percentage point increase from the prior month—surpassing OpenAI’s 39.5% for the first time. Meanwhile, Anthropic completed a funding round valuing the company at $965 billion and confidentially filed its IPO prospectus, reportedly backed by its first-ever profitable quarter.
Yet amid these developments, as previously reported by Wall Street Journal, the Trump administration sent a letter to Anthropic last Friday demanding that it prohibit non-U.S. persons—including its own employees—from accessing its latest flagship models, forcing Anthropic to fully delist both its newly launched Mythos 5 and Fable 5, which had been publicly available for just three days.
This move marks the latest escalation in tensions between the two sides. In March, the Trump administration had already designated Anthropic as a “supply chain risk” after the company refused to cooperate with government efforts to deploy its models for mass surveillance and fully autonomous weapons.
According to Ara Kharazian, Chief Economist at Ramp, this controversy is unlikely to harm Anthropic; on the contrary, it could further enhance its commercial appeal. Historical data already supports this logic: the month when Anthropic saw its highest enterprise adoption rate coincided precisely with the month the Department of Defense labeled it a supply chain risk.
Subscription Share Overtakes OpenAI for First Time, Reshaping Enterprise Market Landscape
Ramp’s data is drawn from over 70,000 enterprise clients on its platform, lending significant representativeness to these figures.
In May, Anthropic captured 41% of the enterprise AI subscription market, up 2.5 percentage points from April, while OpenAI’s share remained largely flat at 39.5%. This marks the first time Anthropic has surpassed OpenAI in enterprise subscriptions.
Notably, subscription data represents only a portion of enterprise AI spending. Ramp notes that the vast majority of enterprise AI expenditure comes from API calls—usage billed per token—for core business applications such as code generation. Anthropic’s Claude Code has earned strong reputation in enterprise programming tools, which is considered one of the key drivers behind its growth in the enterprise segment.
However, OpenAI still maintains a substantial lead in overall consumer usage, according to the latest data from Sensor Tower, a gap that remains unchallenged for now.
Delisted models versus persistently growing real-world demand
The immediate trigger for this government pressure was two of Anthropic’s latest models: Mythos 5, which had been made available to a limited user base, and Fable 5, which was publicly launched just three days before the incident.
The Trump administration invoked a little-known export control directive, requiring Anthropic to prohibit non-U.S. persons from accessing these models—effectively forcing the company to fully withdraw both products from the market.
Reports indicate that it was widely rumored hackers could easily bypass Fable 5’s safety safeguards to access the core capabilities of Mythos—a model so powerful in identifying software code security vulnerabilities that Anthropic itself labeled it as "dangerous" in its marketing materials and restricted its public release.
However, according to Ramp’s data, enterprise customers are predominantly using Anthropic’s Opus series models. In roughly one-third of transactions where model information could be identified, corporate spending has been concentrated on various versions of Claude Opus, particularly newer iterations.
Mythos had only recently launched—opening to a limited user group in April—and Fable 5 lasted merely a few days. In late May, Anthropic further solidified this product line’s market position by releasing Opus 4.8.
This suggests that even if Mythos and Fable 5 were forced off the market, the actual impact on Anthropic’s current enterprise revenue may be quite limited. Ramp noted that its data granularity is insufficient to precisely quantify this financial effect.
Ramp’s chief economist, Ara Kharazian, told TechCrunch:
"If anything, this incident is likely to boost Anthropic. The best month ever for Anthropic’s enterprise adoption was precisely the month the Department of Defense labeled them as a supply chain risk. When your model is singled out as 'too dangerous to use,' that itself creates a significant halo effect."
This logic is not without precedent. In March of this year, Anthropic was designated a 'supply chain risk' by the Trump administration after refusing to allow the government to use its models for mass surveillance of U.S. citizens or fully autonomous weapons systems. Yet subsequent enterprise sales data showed that this crackdown failed to produce the intended deterrent effect.
This latest friction has, to some extent, once again validated the technical strength of the Mythos series models—the government's firm stance has objectively served as an endorsement.
Editor/Lambor