Anthropic Beat OpenAI for Business Customers. Here's What the Data Won't Tell You.

Meta description: The 34.4% vs 32.3% headline hides a data measurement problem. Here's what small operators should actually take from Ramp's AI Index.

```json { "@context": "https://schema.org", "@type": "Article", "headline": "Anthropic Beat OpenAI for Business Customers. Here's What the Data Won't Tell You.", "author": { "@type": "Person", "name": "Bob" }, "datePublished": "2026-05-13", "sourceOrganization": { "@type": "Enterprise", "name": "Mediascout" }, "citation": [ "https://techcrunch.com/2026/05/13/anthropic-now-has-more-business-customers-than-openai-according-to-ramp-data/", "https://econlab.substack.com/p/anthropic-beats-openai" ] } ```

Key Takeaways

- Anthropic hit 34.4% of Ramp's 50,000+ business customers paying for its services as of May 2026, surpassing OpenAI at 32.3% for the first time. - That 34.4% was 9% twelve months ago. A 4x in enterprise adoption in one year. No SaaS enterprise at this scale has moved that fast. - OpenAI still runs $25B annualized revenue and 910M weekly active users. The "lead" is a 2-point margin that flips monthly. - The data tracks corporate card spend, not Fortune 500 enterprise contracts. OpenAI's biggest deals don't show up on expense reports.

---

Anthropic crossed OpenAI in business customers for the first time last week. Ramp's May 2026 AI Index. Built from 50,000+ companies.

Shows Anthropic at 34.4% versus OpenAI's 32.3%.

Forty-two months ago, Anthropic sat at 9%.

Let that sink in.

From 9 to 34 in twelve months.

OpenAI dropped 1% over the same stretch.

The tech press is calling it a power shift.

They're not wrong.

But the story the headlines are selling isn't the story that matters for you.

What Ramp's Data Actually Measures

Ramp tracks corporate card spend. That's important to understand before you restructure your AI stack based on a headline.

OpenAI's revenue — $25B annualized. Comes substantially from enterprise contracts with Microsoft Azure, direct corporate agreements. And OEM licensing deals that never touch a corporate card.

A Fortune 500 enterprise paying OpenAI $10M a year through a consolidated enterprise invoice doesn't register in this dataset the same way a $2,000/month Claude subscription does.

Anthropic's climb is real.

But it's being measured on the smaller half of the battlefield.

The enterprise contract war. Where OpenAI still holds multi-billion-dollar commitments. Isn't captured here.

That said — the trajectory matters more than the snapshot.

When Anthropic goes from 9% to 34% in twelve months at this scale, that's not noise. That's a product that works getting word of mouth in the only channel that matters: operators spending real money.

Why Small Businesses Should Actually Care

Here's the part the business press keeps skipping.

If you're running a small agency, a solo dev shop, or a lean team under 50 people, you're not choosing between "winning AI" and "losing AI." You're choosing between tools that don't break your workflow on a Tuesday afternoon when a client deliverable is due.

Anthropic's advantage in the Ramp data tracks directly to operator experience.

Claude doesn't hallucinate less because of marketing.

It hallucinates less since of how the model was trained.

I've run both on production tasks.

The difference shows up in the third hour of a complex prompt chain, not the first.

OpenAI still wins on reach. 910M weekly active users means the tooling, the community support, the plugins, the integrations. All of it skews toward ChatGPT's ecosystem.

If you're building something that needs to plug into other tools, OpenAI's head start still matters.

But if you're a small operator who just needs reliable output that doesn't require three rounds of correction?

The data says Anthropic is where the money is going.

Money: usually right.

The Number That Actually Matters for Your Stack

Forget the 34.4% for a second.

Here's the number you should be tracking: revenue per customer.

Anthropic's $30B revenue run rate. Up from $9B at end of 2025.

Sounds massive until you do the math. If they have 1,000+ enterprises spending over $1M/year, that's most of that $30B right there. The long tail of smaller customers contributes less than you'd think.

OpenAI's $25B revenue at 32.3% of the same dataset suggests higher deal values per customer on average. That's consistent with OpenAI's enterprise sales model, which tends toward larger contracted relationships.

What this means for you: don't pick a winner based on customer counts. Pick based on where your specific use case gets cheaper per task. Run both on a small project.

Measure outputs, not headlines.

The Real Takeaway for Your AI Stack

Anthropic winning business customers is a signal, not an instruction.

The signal: operators are finding Claude more reliable for production work. The model does what it says without requiring as much babysitting.

The instruction: nothing. You still need to test both.

Route through OpenRouter if you want flexibility without lock-in. Anthropic just became harder to ignore.

But OpenAI isn't going anywhere.

If you're making a platform decision for your agency right now, the honest answer is: this data tells you there's no wrong choice.

The gap is 2 percentage points. Both companies are printing revenue at scale that would make any SaaS enterprise in history jealous.

What you should actually do: stop reading the comparison articles and run your own benchmark. Ten tasks, both models, your actual workflow. Count the hallucinations.

Clock the time to correct them.

That's the only data that matters for your stack.

The rest is noise.

---

Sources

- TechCrunch: Anthropic now has more business customers than OpenAI - Ramp EconLab: Anthropic Beats OpenAI - SaaStr: Who's Winning Enterprise AI Now - VentureBeat: Anthropic Hit $30B Revenue Run Rate - OpenAI Official - Anthropic Official