More U.S. businesses paid for Anthropic's Claude than for OpenAI's ChatGPT in April 2026 — the first time in the AI industry's short history that the San Francisco-based safety lab has held the top position, according to the May 2026 Ramp AI Index, published this week. The crossover matters directly to the tens of thousands of companies deciding right now which AI vendor to put on contract: it signals that the market has moved, that Anthropic's pricing model is changing how CFOs budget for software, and that OpenAI is responding with a structural counter-offensive that could re-shape the race within months.
Ramp Index Shows Anthropic at 34.4%, OpenAI at 32.3%
The Ramp AI Index draws on corporate card and invoice-based payments across more than 50,000 companies on Ramp's platform. In April, Anthropic's adoption climbed 3.8 percentage points to 34.4% of participating businesses, while OpenAI's fell 2.9 points to 32.3%. Overall AI adoption across the sample rose a further 0.2 points to 50.6% — meaning nearly half of U.S. businesses on the platform still run no AI in production.
"Today, for the first time, Anthropic passed OpenAI in business adoption," wrote Ara Kharazian, Ramp's lead economist, calling the shift "a stunning reversal in the competitive market dynamics for AI model providers." The reversal caps a twelve-month run that Kharazian described as exceptional even by the standards of a fast-moving industry: in May 2025, only 9% of businesses on Ramp were paying for Anthropic products; that figure rose 26 percentage points in the following year, while OpenAI's share declined by 1%.
The index has real limits. Ramp's sample skews toward tech-forward, venture-backed companies already predisposed to Anthropic's developer-first products. The index measures paid subscriptions, not usage intensity, and likely undercounts free-tier adoption of both vendors. On OpenRouter's leaderboard, which samples a different population of users, OpenAI last ranked above Anthropic in December 2025 — a corroborating signal, but still a partial one.
Claude Code Drove the Reversal — and Blew Up at Least One Enterprise Budget
The engine behind Anthropic's climb is a single product. Claude Code, the company's terminal-native agentic coding tool, has become the fastest-growing product in Anthropic's history. A recent analysis estimated that 4% of all GitHub public commits worldwide were authored by Claude Code — double the share from one month prior. By February 2026, the product was generating more than $2.5 billion in annualised revenue, according to Anthropic's Series G announcement, and business subscriptions had quadrupled since January 1.
The scale of real-world consumption is not abstract. Uber CTO Praveen Neppalli Naga told The Information that the company burned through its entire 2026 AI budget in four months, driven by Claude Code spreading across its 5,000-engineer organisation faster than any budget model could absorb. Adoption jumped from 32% to 84% of Uber engineers between December 2025 and March 2026. Individual monthly costs ranged from $500 to $2,000 per engineer; Naga himself spent $1,200 in a two-hour personal demonstration session. "I'm back to the drawing board," he said, "because the budget I thought I would need is blown away already." About 70% of committed code at Uber now comes from AI, and roughly 11% of live backend updates are written by AI agents without human review.
That cost dynamic is not unique to Uber. NPI Financial, an IT procurement advisory firm, warned enterprise buyers in April that Anthropic's shift from bundled seat-based pricing to per-token billing would increase total cost of ownership for most organisations. Older enterprise contracts included a generous flat-rate token allowance; new contracts charge a $20-per-employee monthly access fee for platform access only, with all usage billed separately at standard API rates on top. "The seat fee no longer bundles any usage allowance at all," said Adrien Laurent, CEO of AI consultancy IntuitionLabs, in comments to The Register. "Every token gets billed at standard API rates on top of the base seat."
Anthropic's Own Economist Flags Three Risks to the Lead
The same report that put Anthropic in first place offered a detailed case for why the lead may not hold. Kharazian identified three headwinds specific to Anthropic, in writing published alongside the index figures.
First, a structural pricing conflict: Anthropic earns more revenue when businesses consume more tokens, which gives the company a commercial incentive to push customers toward more expensive flagship models even when cheaper alternatives would serve the task. "Anthropic's incentives are misaligned with those of business customers," Kharazian wrote. "Anthropic is incentivized to drive users to more expensive models, even when cheaper models are sufficient and faster for many tasks." Ramp economist Rafael Hajjar found that Anthropic's latest model update tripled token costs for any prompt containing an image — a change that Kharazian described as difficult to reconcile with the company's compute constraints.
Second, reliability. In the weeks before publication, Claude users experienced frequent outages, rate limits, and rising complaints about degraded output quality. Anthropic responded by resetting usage limits for all paid users in April and striking a compute agreement with SpaceXAI — the entity formed from SpaceX's merger with Elon Musk's xAI — to access the entire Colossus 1 data centre in Memphis, Tennessee: more than 220,000 Nvidia GPUs across 300 megawatts of capacity. Claude Code's five-hour rate limits were doubled immediately after the deal closed on May 6.
Third, the rise of cheaper alternatives. Some of the fastest-growing vendors on Ramp's platform in April were AI inference services giving companies access to open-source models at a fraction of frontier pricing. "OpenAI's Codex is pretty good, does the same tasks more cheaply, and the cost to switch between the models is minimal," Kharazian wrote.
OpenAI Calls the Ramp Data Incomplete and Launches a $4 Billion Counter
OpenAI did not accept the Ramp figures at face value. An OpenAI spokesperson told Axios: "We are driving enterprise transformation at scale. These are not engagements where customers pay with a credit card" — a pointed reference to the large, invoiced, six- and seven-figure contracts that Ramp's transaction-based methodology may underweight. OpenAI has previously said it is on track to generate more revenue than Anthropic in 2026.
The company followed its rebuttal with action. On May 11, OpenAI launched the OpenAI Deployment Company, a standalone unit backed by more than $4 billion from 19 private equity firms, consultancies and systems integrators including TPG, Bain Capital, Goldman Sachs, Capgemini, and McKinsey. The unit will place Forward Deployed Engineers directly inside client organisations to redesign workflows around AI — a model borrowed from Palantir, applied to frontier AI for the first time at this scale. OpenAI simultaneously acquired Tomoro, a London-based AI consultancy with roughly 150 engineers and clients including Tesco, Virgin Atlantic, and Mattel, to staff the new unit on day one.
At a company-wide meeting, OpenAI's CEO of applications, Fidji Simo, told staff that Anthropic's enterprise gains should serve as a "wake-up call" and that the company needed to "nail productivity" for business customers. "We cannot miss this moment because we are distracted by side quests," she said.
Revenue Has Crossed Too — But Anthropic's Pentagon Lawsuit Introduces Enterprise Uncertainty
The adoption crossover mirrors a financial one. Anthropic's annualised revenue run rate crossed $30 billion in early April 2026, up from roughly $9 billion at the end of 2025, placing it above the approximately $24 to $25 billion annualised figure OpenAI reported at the same time. More than 1,000 enterprise customers now spend over $1 million annually on Anthropic products — a number that doubled in under two months after the company's $30 billion Series G raise in February 2026. Eight of the Fortune 10 are now Claude customers, according to Anthropic.
The growth is shadowed by a serious legal dispute. In early March, the Department of Defense declared Anthropic a supply chain risk to national security, barring defense contractors from using Claude in military work after negotiations broke down over the Pentagon's demand for "all lawful uses" of the model — language Anthropic refused because it would have removed safeguards against fully autonomous weapons and mass surveillance. Anthropic sued the Trump administration in two federal courts. A DC Circuit appeals court denied the company's request for a stay in April, ruling that its interests "seem primarily financial in nature." A separate federal court in San Francisco reached the opposite conclusion, granting a preliminary injunction. The two rulings leave defense contractors with conflicting legal guidance. More than 100 enterprise customers reportedly raised concerns about their relationships with Anthropic following the blacklisting, according to Bloomberg.
Matt Schruers, CEO of the Computer & Communications Industry Association, which filed an amicus brief in Anthropic's favour, warned that "the Pentagon's actions and the DC Circuit's ruling create substantial business uncertainty at a time when US companies are competing with global counterparts to lead in AI."
What This Means for Companies Choosing an AI Vendor Today
For finance, legal, operations, and engineering teams currently evaluating AI contracts, the Ramp data and the OpenAI counter-move together suggest a market in transition. Anthropic's adoption lead is real but narrow — 2.1 percentage points separates the two vendors in Ramp's sample — and that gap could close or reverse if OpenAI's Deployment Company converts large invoiced contracts faster than Claude Code acquires new paying businesses.
Companies that have already adopted Claude Code face a pricing model that has materially changed since they signed. The shift to per-token billing without bundled allowances means that token consumption — not seats — drives the invoice. At Uber's reported per-engineer cost of $500 to $2,000 per month, a 1,000-engineer organisation could face between $6 million and $24 million in annual AI tooling costs from a single vendor. Finance teams that have not yet built consumption-based forecasting models for AI spend should treat the Uber disclosure as a warning.
Kharazian's closing note captures the condition of the market: "In a fast-moving AI industry, a one-month lead is not yet a moat." Whether Anthropic holds first place in June will depend on whether OpenAI's deployment push, cheaper open-source alternatives, or Anthropic's own pricing decisions move the 50% of businesses that have not yet paid for any AI product — and which vendor captures them when they do.
ⓒ 2026 TECHTIMES.com All rights reserved. Do not reproduce without permission.
