The Ramp AI Index for June 2026 shows Anthropic's Claude at 41% of paying U.S. businesses vs. OpenAI's 39.5%. This is real corporate spend data, and it changes how you should think about your AI stack.
The June 2026 Ramp AI Index landed this week with a clear finding: Anthropic has pulled ahead of OpenAI among paying U.S. businesses. This is not a consumer poll or a benchmark leaderboard. It is actual corporate card and billing data from tens of thousands of companies, and it deserves a straight reading.
What the Numbers Show
Anthropic rose 2.5 percentage points to reach 41% of businesses tracked by Ramp. OpenAI declined 0.1 percentage points to 39.5%. Anthropic leads by 1.5 percentage points.
The trajectory behind that gap is worth understanding. Anthropic was at 0.03% of businesses in June 2023, climbed to 7.94% by April 2025, then jumped to 34.44% by April 2026. OpenAI peaked near 36.5% in mid-2025 and has been slowly declining since. The June 2026 data confirms the crossover was not a one-month blip.
The Product Doing the Heavy Lifting: Claude Code
The numbers alone do not explain the shift. The engine behind much of Anthropic's growth is Claude Code, its agentic AI coding tool and the fastest-growing product in Anthropic's history. A recent analysis estimated that 4% of all public GitHub commits worldwide were being authored by Claude Code, double the share from just one month prior.
Claude Code is not a chat interface. It sits inside engineering pipelines, writes tests, reviews code, and submits pull requests autonomously. Once a development team adopts it, the spend shows up every month on the corporate card and compounds. Anthropic has also used it as a wedge into finance, legal, and research workflows, converting businesses that had never paid for AI before directly onto Claude.
What This Means for OpenAI
OpenAI is not collapsing. The company owns the consumer interface with roughly 900 million users. But the enterprise picture is a genuine concern for its leadership.
OpenAI launched Codex specifically to compete with Claude Code's momentum. The June Ramp data suggests Codex has not yet converted into meaningful business spend. OpenAI held essentially flat despite the launch.
Claude's enterprise advantages include a longer context window, strong coding performance, reliable instruction-following, and a safety-focused positioning that resonates with risk-conscious procurement teams. OpenAI still leads in multimodal capabilities, ecosystem depth, and established relationships at the Fortune 500 level.
One important caveat: Ramp captures corporate card and direct subscription spend. Large enterprises running OpenAI through Microsoft Azure, or Google AI through existing Cloud Platform and Workspace contracts, often never touch a corporate credit card. Those deployments may simply not appear in Ramp's figures at all.
The Real Risks Facing Anthropic's Lead
Ramp's own economist, Ara Kharazian, published the adoption data alongside a candid list of reasons why Anthropic's position is fragile. Business leaders should take these seriously.
Cost pressure is real. Anthropic earns more when businesses consume more tokens, which creates an incentive to push users toward expensive models even when cheaper, faster ones would do. Uber's CTO announced the company had already exceeded its full 2026 AI budget. When finance teams start auditing AI spend, token-heavy agentic workflows will be the first line item they scrutinize.
Reliability has been uneven. Users have reported outages, rate limits, and inconsistent output quality in recent periods. Anthropic has responded quickly, but the pattern is worth watching at scale.
The market is still moving fast. DeepSeek and other low-cost open-weight models are already pulling cost-sensitive workloads away from both Anthropic and OpenAI. In this industry, vendor stickiness is weak, and a new entrant can shift adoption figures in a matter of months.
What Business Leaders Should Do Right Now
The Ramp data confirms that the AI vendor decision is no longer a safe default. Here is how to approach it practically.
Audit your current AI spend by use case. Break down what your teams are actually using: chat, code generation, document review, API automation. Different tasks have genuinely different optimal models, and paying premium rates on tasks where a cheaper model performs equally well is straightforward budget waste.
Run a structured pilot on coding and agentic workflows. If your engineering team is not evaluating Claude Code, this data is a strong signal to start. A 4% share of global GitHub commits points to productivity gains that are measurable, not theoretical.
Avoid hard lock-in on either side. Architecture choices made today, such as whether your infrastructure calls a specific model directly or routes through an abstraction layer, will determine how flexible you are in 12 months.
Set cost controls before deploying agentic tools. Every autonomous agent spawning sub-tasks multiplies your token bill. Configure spend alerts before rolling out Claude Code or any agentic framework at scale, not after.
The Bigger Picture
Anthropic's annualized revenue climbed from $1 billion in early 2025 to $4.7 billion by May 2026, a 16-month run that is hard to overstate. The market is voting with real money, not benchmark scores.
The headline that Anthropic beat OpenAI in business adoption matters. But the real takeaway for decision-makers is simpler: the AI vendor market is now genuinely competitive, the best choice changes by use case and workflow, and the businesses treating this as a strategic procurement decision rather than a default will come out ahead on both cost and capability.
How 247techify can help
At 247techify, we help businesses cut through the noise on exactly these decisions, from evaluating Claude versus OpenAI for your specific workflows to deploying agentic tools with proper cost controls and governance in place. If your team is ready to make AI vendor decisions based on data rather than marketing, get in touch at https://www.247techify.com/ and we will map out the right approach for your business.
