Anthropic Passed OpenAI in Revenue — What the Shift Means

The AI revenue throne just changed hands

On April 7, 2026, Anthropic announced its annualized revenue run rate had crossed $30 billion, surpassing OpenAI for the first time. OpenAI sits at roughly $25 billion. Eighteen months ago, the comparison would have been laughable — Anthropic was at $1 billion in January 2025, OpenAI was already past $5 billion and pulling away.

That milestone is more than a leaderboard entry. The Anthropic-vs-OpenAI numbers shape what the AI tools small businesses depend on actually cost, how fast they improve, and which vendors absorb the next wave of infrastructure pressure. If you run a five-person service business in the Appalachian region, this is the kind of “industry news” that quietly shows up on your bill three months from now.

What the numbers actually say

The headline figure: Anthropic crossed a $30 billion annualized run rate in early April 2026, up from $9 billion at the end of 2025. The number of customers spending more than $1 million per year doubled from 500 to 1,000 in roughly two months, tracking from Anthropic’s February Series G announcement.

A few details matter:

Coding drove the swing. Claude Code crossed $2.5 billion in annualized revenue on its own, and Anthropic now holds roughly 54% of enterprise coding AI usage versus OpenAI’s 21%, per the Menlo Ventures 2025 enterprise AI report and the April updates that followed.
Eight of the Fortune 10 are paying Claude customers.
OpenAI disputes the comparison. OpenAI argues Anthropic’s gross revenue accounting overstates the figure by roughly $8 billion. Under OpenAI’s preferred net calculation, Anthropic’s comparable number would land closer to $22 billion — still extraordinary, but behind OpenAI. Multiple outlets, including Trending Topics, have flagged this caveat.
An IPO is reportedly on the table as early as October 2026, with Reuters and other outlets covering the planning timeline.

Whether or not you accept the gross-vs-net argument, the trajectory is real. Anthropic went from $1 billion to $30 billion run rate in roughly fifteen months. That kind of growth doesn’t happen because of marketing — it happens because real businesses are signing real contracts.

Why this matters for small business AI tools

Three labs, balanced power

A year ago, a few people argued OpenAI would win AI the way Google won search. The market just answered that question. Anthropic, OpenAI, and Google now sit at roughly comparable scale — three labs with frontier capabilities, each with enough revenue and capital to keep shipping. The previous post on Anthropic capturing 40% of enterprise LLM spend traced how that share shifted; this revenue milestone is the dollars catching up to the share.

For a small business, three balanced labs is the best possible outcome. A monoculture would mean one company setting prices and shipping on its own timeline. Three roughly equal competitors creates pricing discipline and shipping urgency. Every quarter, one of them undercuts the others on the mid-tier models that small business tools actually run on. That’s why the workload that cost $500 a month two years ago now runs $20 to $50.

Coding leadership is the quiet beneficiary

Anthropic’s dominance in coding is not just a developer-tool story. The small business SaaS you depend on — your booking platform, your accounting software, your inventory app, your Shopify plugins — gets shipped by engineering teams that increasingly use Claude or GPT to write the code. When the model leading those workloads gets meaningfully better, the tools you actually buy ship faster and break less. We’ve written before about how AI coding tools affect software you use; this revenue milestone means that effect compounds.

Vendor stability matters more than vendor branding

Seventy-six percent of enterprise AI use cases are now purchased rather than built, up from a near-even split in 2024. Big enterprises trust outside vendors enough to hand over real workloads. The same logic applies down-market. The AI tool you sign up for in 2026 is far less likely to disappear in a funding cliff than its 2024 equivalent — its upstream provider is one of three companies with durable revenue and a path to public markets.

That stability is invisible until it matters. It matters the day a tool you depend on for after-hours customer intake renegotiates its model contract or has its primary vendor cut off API access. The strongest signal that won’t happen to your stack is that the underlying model providers are now too big and too profitable to vanish.

Our take

The growth rate is the real story, not the leaderboard.

Going from $1 billion to $30 billion in fifteen months means Anthropic is signing contracts faster than its operations can keep up. That’s a different kind of risk profile than pure scale. Capacity gets tight. Pricing changes. Service terms tighten on the lowest-cost tiers as the company prioritizes the customers paying $1 million a year. Small businesses on the cheapest plans are the first to feel that.

The bottom line: Don’t pick your AI vendor based on which lab leads this quarter. Pick based on whether the tool solves your problem and whether your provider can route between models if one gets expensive or capacity-constrained.

What’s missing from most coverage of this milestone is the practical question: what should a small business actually do differently? The headlines are about valuation and IPO timing. The useful question is whether the tool sitting between you and the model — the AI Employee, the chatbot, the scheduling assistant — gives you any choice in which model it uses. If yes, you’re insulated from any one lab’s pricing decisions. If no, you’re exposed to whichever provider that vendor locked in.

It’s also worth flagging the accounting dispute. Anthropic’s $30 billion is gross revenue, before the cut paid to AWS and Google Cloud (where most of Anthropic’s compute runs). OpenAI uses net accounting. A clean apples-to-apples comparison probably puts the two roughly even, with Anthropic ahead on growth velocity. Either way, the era of one dominant AI lab is over.

What you should do

You don’t need to make a portfolio decision this week. But three things are worth doing while the dust settles:

Check whether your AI tools are model-portable. Ask your vendor — explicitly — which model their tool runs on, and whether it can switch providers. Vendors that built on a single API are more fragile than vendors that abstract the model layer. Our AI Employees are built to route between providers because we’ve watched too many small business tools get caught flat-footed by an upstream pricing change.
Stop overpaying for capabilities you’re not using. The mid-tier and budget-tier models from all three labs are now genuinely competent at the tasks small businesses need: customer intake, review responses, scheduling, basic content generation. If you’re paying for premium model access for tasks a budget model handles, you’re subsidizing someone else’s experimentation.
Watch for capacity tightening on free and cheap tiers. As Anthropic prioritizes million-dollar enterprise customers, expect rate limits and quality changes on the cheapest API tiers. The tool you use today should publish a status page and have a fallback story if its primary model API goes degraded.

Watch for

The IPO timeline. If Anthropic files in October as reported, expect a quiet period that affects pricing announcements and partnership news through Q4.
OpenAI’s response. A revenue swap usually triggers a price war or a major product launch. Expect one or both before fall.
Google’s Gemini share. The third member of the top three has been gaining quietly. If Google takes share at the bottom tier, Claude and GPT prices have to follow.

The shift you’ll actually feel

Most small business owners will never see Anthropic’s revenue numbers. But if your AI scheduling tool gets more reliable in Q3, or your AI receptionist drops its monthly bill by a third, or a competitor in your market suddenly starts answering after-hours calls when you don’t — that’s the downstream effect of three competitive labs racing for share. The leaderboard changed. The work hasn’t.

If you’re trying to figure out which AI tools belong in your stack right now, explore our AI Employees or get in touch — we’ll help you pick what fits, model-portable by default.