Ramp Data Shows Anthropic Topping OpenAI in Paid Business AI

Anthropic now leads OpenAI in paid US business AI adoption — 41% to 39.5% — according to Ramp's June AI Index, which tracks real corporate billing data from more than 70,000 companies. This is not a survey. It is receipts. And it arrives weeks before both labs file for IPO while OpenAI's leadership is treating Anthropic's rise as a code red.

The numbers:

• Ramp lead economist Ara Kharazian reported on June 9 that Anthropic rose 2.5 percentage points to 41% of US businesses with paid AI subscriptions.
• OpenAI was essentially flat, slipping 0.1 points to 39.5%.
• The gap widened to 1.5 percentage points after Ramp updated its methodology to better capture enterprise spend.
• In June 2023, Anthropic held 0.03% of business subscriptions. Three years later, it leads.

Why this isn't a blip:

Ramp's index measures which AI products companies actually pay for — corporate card and bill-pay transactions, not free-tier signups or analyst estimates.

The May reading already showed Anthropic crossing OpenAI at 34.4% versus 32.3%. The June update confirmed the lead held and widened.

IDC's separate enterprise survey still shows OpenAI ahead on a different metric — organizational penetration. Both can be true. Ramp measures who is writing checks. IDC measures who has deployed. The procurement vote has flipped.

Claude Code broke the lock:

SiliconANGLE, citing Ramp data, identifies Claude Code as the primary driver — the agentic coding tool that lets developers delegate multi-step software tasks with minimal hand-holding.

Anthropic never chased Sora, browsers, or hardware. It chased workflows. OpenAI spread across a dozen "side quests," as applications CEO Fidji Simo called them in a Wall Street Journal report.

Simo told staff Anthropic's gains should serve as a "wake-up call" and that OpenAI needed to reclaim its position with developers and enterprise customers.

Her message was blunt: "We are very much acting as if it's a code red."

The IPO stakes:

Anthropic raised $65 billion at a $965 billion valuation in late May and filed confidential IPO paperwork in June.

OpenAI is racing the same window with its own listing — but the enterprise revenue story investors will underwrite is no longer uncontested.

On June 23, Anthropic doubled down with Claude Tag — an AI teammate that lives inside Slack, shares context across teams, and operates like a virtual employee. The product push is designed to convert developer enthusiasm into organization-wide lock-in before Codex catches up.

The cost ceiling:

Ramp's same dataset reveals a more than 650-fold spending gap between AI's heaviest adopters and everyone else.

The top 1% of firms spend $7,449 per employee per month on AI tools. The median firm spends $11.38 — roughly one enterprise seat.

That top tier grew spend 14.1% last month while also routing routine work to cheaper models from DeepSeek and open-source platforms.

Anthropic's per-token pricing drives revenue growth — and creates the first churn risk when CIOs audit the bill.

What to watch:

• July's Ramp AI Index — whether Anthropic's lead widens or OpenAI's Codex push narrows it
• OpenAI's reported enterprise PE joint venture talks valued near $10 billion pre-money
• Whether Claude Tag accelerates adoption beyond developer teams into legal, finance, and operations
• IPO pricing as both labs convert enterprise share into public-market narratives

Convina's view: The model war was always the wrong frame. Businesses do not buy benchmarks — they buy workflows that ship code, close tickets, and draft documents. Anthropic bet on agents. OpenAI bet on everything. Ramp's billing data just rendered the verdict in dollars. The lead is real, but it is also thin — 1.5 points, one product cycle, one pricing rethink away from reversal. What Anthropic has won is not permanence. It is proof that enterprise AI is contestable — and that is the most dangerous number in the entire IPO season.