The numbers arrived quietly. No press release, no executive tweet. Ramp’s April AI Index dropped into the market with a data point that contradicts the prevailing assumption of the past three years: more U.S. businesses are now paying for Claude than for ChatGPT.
34.4% to 32.3%. A 2.1-point gap. Anthropic up 3.8 points month-over-month; OpenAI down 2.9.
Before analyzing what this means, it’s worth being precise about what it is. Ramp runs a corporate spend management platform, it tracks what companies pay, not what they use. Its AI Index is a proprietary dataset derived from business transactions processed through Ramp’s system. The index measures spend share, not seat counts, session volumes, or task completion rates. A mid-market company upgrading from a ChatGPT Team plan to Claude Enterprise moves the needle the same direction as a thousand individual users switching defaults. These are different signals with identical index impact. That distinction is load-bearing for anyone making procurement decisions based on the data.
It’s also a single-primary-source data point. VentureBeat and Business Insider both covered the index release, but their coverage is derivative, they’re reporting on Ramp’s publication, not running independent verification against separate methodologies. This is the best available spend-side signal for enterprise AI adoption. It’s not the only signal, and it shouldn’t be the only input.
That said: the direction is meaningful. And direction requires explanation.
What Anthropic Did Differently in the Last 90 Days
Three things happened in Anthropic’s enterprise stack between February and April that don’t show up in a single data point but appear to be moving in the same direction as the spend share shift.
First, Claude Code’s adoption among development teams accelerated. Prior pipeline coverage documented the SpaceX Colossus infrastructure deal and institutional deployments that concentrate Claude in high-compute, high-frequency usage environments. Development teams that adopted Claude Code for agentic coding tasks are generating significant recurring spend at contract tier rates that index disproportionately in Ramp’s data. A single mid-size engineering organization on an enterprise Claude Code contract outweighs dozens of individual ChatGPT subscribers in spend share terms.
Second, Anthropic’s pricing architecture for high-volume API contracts tightened in ways that made enterprise budgeting more predictable. Procurement teams at regulated companies, finance, legal, healthcare, respond to pricing predictability more than they respond to benchmark performance. Claude’s positioning as a compliance-oriented model (constitutional AI framing, documented safety evaluations) has given it credibility in RFP processes where OpenAI’s consumer-first brand history has created friction. Enterprise AI is outperforming consumer AI on revenue metrics for reasons that map directly to this dynamic.
Third, Anthropic closed what has been reported as a $30 billion financing round in the cycle preceding this data. Large financing rounds affect enterprise procurement in ways that aren’t intuitive: they reduce perceived vendor risk. Procurement teams at large organizations run vendor stability assessments before signing multi-year AI contracts. A company that can point to $30 billion in institutional backing passes that assessment more easily than one that can’t, regardless of model performance. The financing close almost certainly accelerated deals that were in late evaluation stages.
Who This Affects
What OpenAI’s Position Actually Is
OpenAI isn’t losing. It’s navigating a more complex enterprise motion than it built for.
The company designed its go-to-market around individual and small-team adoption: ChatGPT consumer, ChatGPT Team, API access with minimal friction. That motion captured enormous market share fast. It also created a pricing architecture where enterprise customers, the ones who show up disproportionately in Ramp’s index, aren’t always on the most favorable contract structures.
The Deployment Company JV, confirmed in prior registry coverage as backed by a consortium of private equity investors, is OpenAI’s structural answer to this problem. The thesis is that deploying OpenAI models through a purpose-built enterprise delivery vehicle, with implementation support, compliance documentation, and long-term contract structures, closes the gap that Anthropic has been exploiting. The Deployment Company isn’t built to compete on the Ramp index today. It’s built to shift enterprise contract flow over the next 12-18 months.
That timeline matters. Enterprise procurement cycles run 6-18 months for meaningful contract sizes. The deals that show up in the April Ramp data were likely initiated in Q4 2025 or Q1 2026. The Deployment Company’s pipeline won’t appear in spend data until late 2026 at the earliest. Investors reading the April index as a permanent market share shift are reading 90 days of data as a multi-year verdict.
The Methodology Question Enterprise Buyers Should Actually Be Asking
Ramp’s index is useful precisely because it’s transaction-based, not survey-based. Companies lie on surveys about what tools they use; they don’t lie on invoices. But transaction-based data has its own blind spots.
It doesn’t capture internal deployment of open-source or self-hosted models, a growing share of enterprise AI infrastructure that doesn’t generate SaaS spend at all. It doesn’t capture API calls that are embedded in third-party products and don’t appear as direct AI tool purchases. And it doesn’t capture the difference between an organization that uses Claude as its primary intelligence layer versus one that has a handful of teams piloting it alongside their primary OpenAI deployment.
The 34.4% figure means Anthropic leads on Ramp’s specific measure. It doesn’t mean 34.4% of enterprise AI decision-makers have chosen Anthropic as their primary vendor. Those are different facts.
What to Watch
Analysis
Enterprise procurement cycles run 6-18 months for meaningful contract sizes. The deals appearing in April's Ramp index were initiated in Q4 2025 or Q1 2026. Anyone reading April spend data as a permanent market share verdict is collapsing a 90-day signal into a multi-year conclusion.
For procurement teams using this data to justify a vendor evaluation, the right question isn’t “is Claude winning?” The right question is: what does your organization’s actual usage profile, API calls, seat count, task type, compliance requirement, look like, and which model’s pricing and capability architecture fits that profile? The Ramp data tells you what other companies are spending. It doesn’t tell you what you should spend, or on what.
The Trend Signal That Actually Matters for the Next Quarter
One month of data showing Claude ahead isn’t a trend. Two months is a pattern. Three months is a strategic signal.
The May Ramp AI Index, due approximately four weeks from now, is the single most important follow-on data point for investors and procurement teams tracking this question. If Claude’s lead widens past 3 percentage points in May, the shift is structural and Anthropic’s enterprise moves are compounding. If the gap narrows back below 1 point, April was noise, contract timing, a large renewal cohort, or seasonal procurement patterns.
Watch also for Q2 earnings commentary from Salesforce, ServiceNow, and enterprise software vendors who sit adjacent to AI deployment. These companies have direct visibility into which AI models their enterprise customers are integrating. Their public earnings comments on AI attach rates will provide independent triangulation on whether Ramp’s April signal has legs. The first of those calls arrives in July. Mark it.
Anthropic’s lead in the April Ramp index is real, narrow, and methodologically bounded. It’s also the first spend-side confirmation of what prior qualitative signals suggested: Anthropic is executing a serious enterprise motion, and it’s beginning to show in purchasing behavior. Whether it’s durable depends on product velocity, pricing discipline, and whether OpenAI’s Deployment Company structure can close enterprise contracts faster than Anthropic can compound its current lead. Watch Q2 Ramp data. Watch Salesforce July earnings. That’s where the verdict lands.