Anthropic Passes OpenAI in Business Adoption for First Time

Anthropic just crossed a line that few in the AI industry thought possible this fast. According to Ramp's May 2026 AI Index, 34.4% of participating businesses are now paying for Anthropic products, surpassing OpenAI's 32.3% share for the first time in the index's history. Twelve months ago, Anthropic's share sat at 9%. That is a 25-percentage-point gain in one year, while OpenAI's share fell by 1%.

What Actually Happened

Ramp, the corporate spend management platform used by tens of thousands of businesses, publishes a monthly AI Index tracking which AI labs are receiving corporate card and invoice payments from its customer base. The May 2026 data marks the first time Anthropic has topped the ranking. Claude Code, Anthropic's agentic coding tool, is the single biggest driver: it's the fastest-growing product in Anthropic's history, and the three sectors that most favor Anthropic, software and information services, finance, and professional services, are precisely the industries that code, model, and automate at scale.

The shift has been relentless. In May 2025, Anthropic was a distant third in enterprise adoption, with 9% versus OpenAI's 33.3%. By November 2025, Anthropic had reached 21%. By February 2026, it hit 28%. The May 2026 figure of 34.4% is not a blip; it's the conclusion of a 12-month trend driven by Claude Code's adoption among engineering teams, Anthropic's partnership with Goldman Sachs and Blackstone on a $1.5 billion enterprise services firm, and a sustained reputation advantage in safety-sensitive industries like finance and healthcare where Claude's more predictable behavior commands premium contracts.

Why This Matters More Than People Think

Enterprise AI adoption data is a leading indicator of where revenue concentration is heading. Ramp's index counts actual payments, not survey responses or stated preferences. When 34.4% of tech-forward businesses are routing dollars to Anthropic and only 32.3% to OpenAI, that signals a structural shift in enterprise purchasing that will show up in annual contract data, security audits, and vendor consolidation decisions over the next 12 to 18 months. CFOs don't swap primary AI vendors easily; they build compliance frameworks, data governance policies, and engineering integrations around them. Anthropic's lead, once embedded, is sticky.

The revenue implications are stark. Anthropic reported $1 billion in annualized revenue from Claude Code alone as of early 2026, and the company's total ARR hit approximately $3.5 billion following its Amazon partnership expansion that committed $100 billion in AWS compute. OpenAI's revenue remains larger in absolute terms, approximately $25 billion annualized, but it is disproportionately concentrated in consumer ChatGPT subscriptions and the SMB tier. The enterprise battle, which is where 80% of AI-related software margins will be captured over the next decade, is the one Anthropic is now winning.

The Competitive Landscape

Google trails both companies in Ramp's enterprise adoption data, with roughly 20% of businesses making payments to Google AI services, despite having the largest distribution network in the world via Workspace and Cloud. That gap reflects the fundamental tension in Google's AI strategy: it is simultaneously trying to defend its search advertising business, grow Gemini as a standalone product, and sell AI infrastructure to enterprises who may be competitors. Neither OpenAI nor Anthropic faces that structural conflict, which is one reason enterprises are increasingly choosing one of them as a primary AI vendor rather than defaulting to Google.

The comparison with historical technology transitions is instructive. When Slack passed Microsoft Teams in SMB adoption in 2019, Microsoft responded by bundling Teams into Office 365 at no extra cost, ultimately reversing the trend within three years. OpenAI's equivalent defensive move would be to bundle enterprise API access into Microsoft's Copilot contracts at favorable pricing, leveraging the $13 billion Microsoft has committed to OpenAI infrastructure. That playbook is available, and Anthropic's lead should be read as a signal of where the battleground is, not a settled verdict on who wins.

Hidden Insight: The Claude Code Effect Is Bigger Than It Looks

The conventional explanation for Anthropic's rise, better safety, more reliable outputs, fewer hallucinations, is true but incomplete. The deeper driver is Claude Code's ability to turn developers into advocates inside enterprises. When a software engineering team adopts Claude Code, it creates an internal constituency for Anthropic that bypasses the typical enterprise IT procurement process. Developers choose their tools, then IT ratifies what developers already use. This bottom-up adoption pattern mirrors how Slack displaced email, how GitHub displaced centralized version control, and how AWS displaced on-premise servers: through developer preference, not executive mandates.

The strategic implication is that Anthropic's enterprise advantage is self-reinforcing in ways that pure product quality cannot replicate. Every engineering team running Claude Code generates fine-tuning feedback, usage patterns, and integration demand that shapes Anthropic's roadmap. Claude Code users are functionally co-developing the product, and the switching cost grows with every workflow they build around it. OpenAI has Codex, its competing agentic coding product, and roughly 3 million developers on its platform. But Codex launched significantly later, and the developer loyalty it needs to build will take years, not months.

The second-order effect worth watching is Anthropic's valuation trajectory. The company is reportedly in talks for a funding round that would value it at $950 billion, up from $61 billion just 18 months ago. That 15x jump in implied value tracks almost exactly with the Ramp adoption curve: from 9% to 34.4% in enterprise share. Investors are pricing Anthropic as the enterprise AI standard-setter, not merely as a research lab that ships models. If that framing holds, the $950 billion figure may understate the eventual market outcome.

The bear case, however, is straightforward: Ramp's client base skews toward tech-forward, venture-backed companies that are structurally predisposed to Anthropic's developer-first positioning. The broader enterprise market, Fortune 500 procurement teams, heavily regulated financial services firms, healthcare conglomerates, still skews toward OpenAI's brand recognition and Microsoft's bundling power. Anthropic's Ramp-measured lead could overstate its position in the full enterprise market by as much as 10 to 15 percentage points. Winning Sequoia-backed startups is not the same as winning JPMorgan's AI contract.

What to Watch Next

The next 90 days will be decisive. Google I/O on May 19 will determine whether Gemini Ultra 2 can close the capability gap that drove engineers to Claude Code. Microsoft Build in late May will reveal how aggressively Microsoft plans to bundle Copilot with OpenAI's Codex to defend enterprise share. If either announcement produces a materially better coding product at a lower price point, Anthropic's Ramp numbers could flatten quickly. The adoption curve is steep in both directions when the primary driver is developer preference rather than contractual lock-in.

Track Anthropic's quarterly ARR disclosures and any updates to its Goldman/Blackstone enterprise services venture. If that $1.5 billion vehicle begins winning mandates from financial services firms, it would confirm that Anthropic's enterprise lead is extending beyond the tech sector into the regulated industries where contract sizes are largest. Also watch OpenAI's Deployment Company, launched in Q1 2026 to accelerate enterprise onboarding: if it gains traction among Fortune 500 companies, it signals OpenAI has identified the same structural problem and is fixing it faster than Anthropic's Ramp numbers suggest.

Anthropic didn't beat OpenAI at the frontier. It beat OpenAI at the place that determines the frontier: the engineering teams building everything on top of it.

---