The Vendor Crossover Doctrine: Anthropic Just Passed OpenAI In Business Spend, And Your Default AI Vendor May Be Wrong For The Next 12 Months

For the first time since ChatGPT launched in November 2022, more American businesses are paying Anthropic than OpenAI.

Read that sentence again.

Two years ago that sentence would have sounded like a fan fiction headline. Today it is the documented finding of the Ramp AI Index for May 2026, published June 9 and reconfirmed this week, drawn from corporate card and bill-pay transactions across more than 70,000 US companies (Ramp via Nerd Level Tech).

Anthropic now holds 41% of paid US business AI subscription spend.

OpenAI holds 39.5% (AI Weekly).

A 1.5-point Anthropic lead.

The number is small. The shape of the move is not.

Why is this Ramp number such a big deal for business owners?

Because the metric is not benchmarks, not vibes, not analyst slide decks.

It is actual spend from corporate cards and bill-pay invoices at over 70,000 US companies (TechCrunch via FuTu News).

When real budgets flip, your competitors are quietly changing their tooling.

If you do not run the same audit they did, you ship slower on a vendor losing share while they ship faster on the one gaining it.

Here are the numbers that matter for your next quarter.

Anthropic was at 0.03% business adoption in June 2023, a rounding error (ChatForest).

By April 2025 it had reached 7.94%.

By April 2026 it hit 34.44%.

By May 2026, in the revised methodology, it sat at 41% (247Techify).

That is a roughly 1,100x increase in paid business customers over three years (ChatForest).

Over the same year that Anthropic roughly quadrupled, OpenAI grew its business adoption by 0.3 percentage points (ChatForest).

Among businesses purchasing AI services for the first time, Anthropic was winning approximately 70% of head-to-head comparisons against OpenAI by early 2026 (ChatForest).

The revenue picture confirms what the adoption data is signaling.

Anthropic's annualized revenue went from approximately $1 billion in January 2025 to $47 billion in May 2026, a roughly 30x climb in fifteen months (eCorp IT).

OpenAI moved from $13 billion to $24 to $25 billion in the same window (eCorp IT).

Claude Code by itself crossed $2.5 billion in annualized revenue as a standalone product, larger than most public SaaS companies (Motley Fool).

That single product holds 54% of the enterprise AI coding market against OpenAI's 21%, and coding accounts for 51% of all enterprise AI spending (Motley Fool).

That is not a close race. That is a market leader and a challenger.

What changed in the buyer's brain to drive this flip?

Buyers stopped purchasing "the most powerful AI" and started purchasing "the AI they can trust for their specific use cases" (Note IT News).

That is the line that explains everything.

When a buyer thinks in capability terms, they pick whoever leads the latest benchmark.

When a buyer thinks in reliability terms, they pick whoever ships the cleanest enterprise contract, the most explainable behavior, and the lowest blast radius when something goes sideways.

That shift turned Anthropic from a runner-up into a leader and turned Claude Code into the spending crown jewel of enterprise AI.

The same dynamic is now playing out in Asia.

Anthropic opened its Seoul office on June 17 and 18, its third Asia-Pacific office after Tokyo and Bengaluru, with named deployments at NAVER, Samsung SDS, LG CNS, Nexon, Hanwha Solutions, and Channel Corp (Anthropic Newsroom).

Claude Code weekly active users in Korea grew 6x in four months, and Asia-Pacific large-business accounts above $100,000 in annualized revenue grew 8x (Let's Data Science).

Channel Corp, a Korean customer AI platform, alone serves over 230,000 businesses on Claude infrastructure (Anthropic Newsroom).

That is the same pattern Ramp captured in US data, repeating in Korea, three quarters later.

But isn't there a survey saying OpenAI is still ahead?

Yes, and you should know about it before you make any decisions.

The IDC March 2026 enterprise survey puts OpenAI at approximately 42% of organizations and Anthropic far lower (Nerd Level Tech).

The Menlo Ventures cohort puts enterprise LLM spend share at OpenAI 25% and Anthropic 32% (eCorp IT).

An alternative survey puts OpenAI at 27% and Anthropic at 40% (eCorp IT).

Three datasets, three different leaders, depending on whether you measure spend, organizations, or seats.

The honest read is this.

In bottoms-up SMB and mid-market spend, where new dollars are flowing right now, Anthropic is ahead by a small margin and growing fast (247Techify).

In top-down enterprise seat count, where multi-year contracts already exist, OpenAI is ahead by a wider margin but barely growing (Nerd Level Tech).

Your business is almost certainly in the first cohort.

Which means the Ramp data is the more honest read of your competitive landscape than any single enterprise survey.

So what do you actually do as a business owner this week?

You run The Vendor Crossover Doctrine.

I built this five-question audit because most operators are running one vendor on lock-in habit, paying a premium for capability they no longer use, and missing the actual product that is driving the market shift.

Use this audit on every AI vendor charging you more than $100 a month.

One. Which workloads in your business are coding, code review, or agentic engineering tasks?

If the answer is "any," check whether you are on Claude Code or its competitor.

The market data is clear. Claude Code holds 54% of enterprise AI coding spend, and that share is still expanding (Motley Fool).

If your engineers and ops people are still using a tool that has been losing share in the category that represents 51% of all enterprise AI spending, you are paying for the runner-up.

Two. When was the last head-to-head test of your two top model vendors on your real production prompts?

If your answer is more than 90 days ago, that test is stale.

Anthropic shipped Claude Fable 5 at $10 per million input tokens and $50 per million output tokens, half the price of Mythos Preview, with an 80.3% SWE-Bench Pro score against GPT-5.5's 58.6%, a 22-point lead (eCorp IT).

OpenAI is reportedly weighing steep price cuts to win business back (AI Weekly billing alert).

Either way, your old benchmark spreadsheet is wrong this month.

Three. What percent of your AI spend goes to a single vendor?

If the answer is above 70%, you have a concentration problem.

The market has now flipped once in 12 months, with Anthropic going from 7.94% adoption to 41% in the same year OpenAI's adoption was essentially flat (ChatForest).

If it flipped once, it can flip again.

Lock annual contracts at hard caps, do not promise a single vendor 100% of your workloads, and keep a parallel sandbox account on the second-place model open at all times.

Four. Are you measuring "AI cost" or "AI cost per shipped output"?

This is the question almost no operator runs.

Cost per output is the right metric. It accounts for tokens, model speed, retry rate, error rate, and engineer time.

A model that is 30% cheaper per token but 60% slower or 2x more error-prone is more expensive per shipped output.

Build a tiny spreadsheet this week. Three columns. Vendor. Tokens per output. Total cost per output, including engineer review time.

Five. What is your "vendor crossover" trigger for switching defaults next quarter?

Write down the trigger right now.

Something like: "If our second-place vendor wins more than 60% of head-to-head tests on our top three workflows for two consecutive months, we switch our default."

That single written rule is worth more than any benchmark spreadsheet.

It removes vendor switching from gut-feel debate and turns it into a calm operational decision.

How is this connected to the IPO timing that just got announced?

The vendor crossover is happening in the same quarter both companies filed to go public.

OpenAI confidentially submitted a draft registration statement to the SEC on June 8, with valuation around $852 billion based on its March 2026 raise (Jangwook Kim).

Anthropic confidentially filed its S-1 on June 1 following a $65 billion Series H at a $965 billion post-money valuation, briefly making it the most valuable startup in the world (Motley Fool).

Both companies need to defend or grow their enterprise share before they ring the bell.

That is why the Anthropic Agent SDK billing reset was reversed on June 15 the day it was to go live (AI Weekly billing alert).

That is why OpenAI is reportedly weighing steep API price cuts (AI Weekly billing alert).

You sit in the negotiating middle for the next 60 to 90 days.

If you have $50,000 or more in annual AI spend across vendors, this is the moment to negotiate.

Tell both vendors the Ramp data is on your desk and ask for your next year's pricing in writing.

TL;DR

• Anthropic reached 41% of US paid business AI subscriptions in May 2026 against OpenAI's 39.5%, the first time it has led the Ramp AI Index (AI Weekly)
• The shift is real spend data from 70,000+ companies, not survey opinion (TechCrunch via FuTu News)
• Anthropic grew adoption from 0.03% in June 2023 to 41% in May 2026, roughly 1,100x in three years, while OpenAI grew 0.3 percentage points in the last year (ChatForest)
• Anthropic annualized revenue: $1B Jan 2025 to $47B May 2026, a 30x climb in 15 months (eCorp IT)
• Claude Code alone is at $2.5B annualized revenue and holds 54% of enterprise AI coding spend against OpenAI's 21% (Motley Fool)
• Anthropic Seoul office opened June 17 and 18 with NAVER, Samsung SDS, LG CNS, Nexon, Hanwha, and Channel Corp deployments (Anthropic Newsroom)
• Anthropic filed its S-1 on June 1, OpenAI on June 8, both racing to lock enterprise share before IPO
• Run The Vendor Crossover Doctrine, a five-question audit on every AI vendor charging more than $100 a month, before the next quarter starts

FAQ

What does the Ramp AI Index actually measure?

Ramp tracks corporate card transactions and bill-pay invoices across more than 70,000 US businesses, then computes the share of those businesses paying for each AI vendor. It is bottoms-up spend data, not survey opinion (TechCrunch via FuTu News).

Is OpenAI still the leader in consumer AI?

Yes. Sensor Tower data confirms OpenAI maintains a substantial lead in overall consumer usage, separate from the business spend race (FuTu News). The crossover is specifically in business subscriptions.

Should I switch my company off OpenAI today?

Not on this data alone. Run the five-question Vendor Crossover Doctrine audit. If your workloads are coding-heavy, the case for testing Claude Code in parallel is strong. If your workloads are consumer-facing voice, multimodal, or video, OpenAI's product lead matters more than the spend data.

What happens if Anthropic's Fable 5 export-control issue is not resolved soon?

The June 22 free-trial window for paid subscribers closes that day. After June 23, subscription users will need paid usage credits (Build Fast with AI). Keep your fallback model live until the export-control matter is fully resolved.

How long do I have to renegotiate my AI contracts?

Roughly 60 to 90 days. Both vendors filed for IPO in the same month, both need to defend enterprise share before listing, and both have signaled willingness to move on price (AI Weekly billing alert). Your negotiating window is the next two months.

What to do this week

Open a blank document.

Write down every AI vendor charging you more than $100 a month.

Run all five questions of The Vendor Crossover Doctrine against each.

When you see the first vendor that fails three or more of the five, you have your switch candidate for the next quarter.

If you want a calm strategic partner to walk you through this audit on your real numbers and pricing data, book an AI Implementation Session at go.8fig.ai/1-on-1. We will sit with your vendor contracts, run the crossover audit live, and ship you a 90-day plan that protects your margins through the IPO window.

The data flipped. Your default may have followed.

Run the audit before the next quarter starts.