OpenAI Just Hit $25B in Annualized Revenue

OpenAI crossed $25 billion in annualized revenue and is taking early steps toward a late-2026 IPO. Anthropic is at $30 billion. The AI business is now materially bigger than most headline writers have caught up to.

$25B
OpenAI ARR (Apr 2026)
$30B
Anthropic ARR (Apr 2026)
$500B
OpenAI last private valuation
Q4 '26
Possible IPO window

In the same week Anthropic disclosed a $30 billion annualized revenue run rate, new reports confirmed that OpenAI has surpassed $25 billion in annualized revenue and is taking early steps toward a public listing as soon as late 2026. The two headline numbers from these disclosures — $25 billion and $30 billion — deserve more attention than they are getting, because they quietly rewrite the story everyone tells about the AI business.

For most of 2024 and 2025, the dominant narrative was "AI companies are burning cash on compute and nobody knows if the revenue model works." That narrative is now outdated. OpenAI is running at roughly the revenue scale of Salesforce in 2015, and Anthropic is running at the revenue scale of Salesforce in 2018. Both companies got there in less than five years from founding. That is one of the fastest revenue ramps in software history.

The 5-Second Version

01

The Revenue Numbers, In Context

The easiest way to feel how big $25 billion actually is: compare OpenAI's current revenue to major public software companies at various points in their history.

$25B
OpenAI 2026
$25B
Salesforce 2018 (19 years old)
$25B
Oracle 2005 (28 years old)

OpenAI is 10 years old (founded December 2015). It took Salesforce nearly two decades and Oracle nearly three decades to reach the same revenue. That is the compression of this cycle in concrete numbers.

The revenue mix is roughly: ChatGPT subscriptions (Plus at $20/mo, Team at $30/user/mo, Enterprise custom) represent the majority — somewhere north of $16-$18 billion of the $25B. The API business contributes another $5-$7 billion. Enterprise deployment deals with large customers like Morgan Stanley, Klarna, and Moderna contribute the rest. The subscription mix is the story: OpenAI is closer to a consumer SaaS company with an enterprise arm than a pure infrastructure play.

02

Anthropic Is Now Bigger Than OpenAI on Revenue

This is the part that is going to keep showing up in business coverage over the next several months. For most of the past three years, OpenAI has been ahead of Anthropic on basically every metric — users, revenue, mindshare, valuation. As of April 2026, Anthropic's disclosed run rate is $30 billion versus OpenAI's $25 billion. That is the first time Anthropic has led on revenue.

Two important caveats. First: the comparison is noisy. Anthropic and OpenAI recognize revenue differently, and Anthropic's $30B figure may weight recently-signed enterprise contracts more heavily than OpenAI's reporting. Second: the order is likely to flip back and forth as both companies grow and disclose new numbers throughout 2026.

But the story is that Anthropic has caught up, and the gap that existed in 2023 (when Anthropic was 1/10th of OpenAI's size) is now gone. That reflects three things: Claude has eaten the developer-and-engineer market for AI coding; Anthropic's enterprise deals with Amazon, Google, and others have ramped aggressively; and OpenAI has been less focused on enterprise relative to consumer growth.

03

What an OpenAI IPO Actually Looks Like

The reports about OpenAI "taking early steps" toward an IPO are worth reading carefully. Early steps in banker language specifically means things like hiring underwriters, cleaning up governance and cap table, doing dry-run audit work, and starting the long process of turning a private company into a public-company-ready entity. None of that locks OpenAI into a 2026 IPO. But it does strongly suggest 2026 or 2027 is on the table.

An OpenAI IPO is complicated by a few factors that don't affect most public listings:

01

The Microsoft Partnership Structure

OpenAI's relationship with Microsoft is not a normal investment — Microsoft gets a share of OpenAI's profits up to a cap, plus preferred access to model releases, plus revenue share from Azure. Unwinding or disclosing this structure cleanly for public markets is legally intricate. Expect this to be the slowest-moving piece of IPO prep.

Watch for a simpler MS-OpenAI deal
02

The Nonprofit-Holds-Company Structure

OpenAI Inc (the nonprofit) still technically controls OpenAI LP (the capped-profit subsidiary). Public markets require clean corporate governance and clear voting rights. OpenAI has been quietly restructuring toward a simpler public-company structure since late 2024; this is likely part of the "early steps" being reported.

Corporate structure simplification is coming
03

Profitability Question

OpenAI is growing fast and burning fast. Public markets in 2026 are more skeptical of "growth at any cost" than they were in 2021. OpenAI likely needs to show a credible path to operating profitability, or at least unit economics that work, before they can list. That may push the actual IPO into 2027 even if prep starts now.

Profit narrative matters, not just revenue
04

Valuation Expectations

OpenAI's last private round was at ~$500B. Public markets typically apply a discount to private valuations at IPO. A $300-400B listing is more plausible than $500B+, unless AI premium multiples stay extreme through 2026. Expect a range of $300B to $600B depending on market sentiment the week of listing.

Realistic range: $300B-$600B
04

What This Means for Everyone Who Is Not OpenAI

Three takeaways if you are a builder, an operator, or a working professional trying to figure out how to position yourself.

The AI business is real. For the past two years, the most common pushback on AI has been "nobody is actually making money, this is all VC subsidy." That argument is now over. Two companies are doing $25B and $30B in revenue. This is a real industry, not a hype cycle. Anyone who was waiting for AI to "prove itself" before investing learning time should stop waiting.

The best AI skills are sticky. When an industry is growing at this speed, people who have hands-on experience with the tools get paid significant premiums. The gap between "I use ChatGPT sometimes" and "I can design, ship, and debug an AI-integrated workflow" is now worth $20K-$60K per year in salary across most professional roles. That gap is going to persist for several more years.

The infrastructure war is over. OpenAI and Anthropic have won. Google is a close third. Everyone else — Meta, Mistral, xAI, various Chinese labs — is now in a second tier competing for capabilities, specific niches, or cost advantages. If you are building AI products, you are building on top of four or five foundation models for the rest of 2026, and the names don't change.

05

The Bottom Line

The Verdict
The AI business is now bigger than most people think, growing faster than almost any software business in history, and concentrating into fewer hands than anyone predicted. If you are not yet building your skill set around it, the window to catch up is still open — but it is closing faster than it was a year ago.

OpenAI at $25B and Anthropic at $30B is the headline. The subtext is that working on top of these models is now the most lucrative professional skill in software, and the demand for people who can build AI-integrated products isn't slowing down. Every AI news story this week, whether it was about compute, cheap APIs, or video generation, is another data point that the industry is concentrating, not collapsing. Build accordingly.

Stop Reading AI News. Start Building With It.

The 2-day in-person Precision AI Academy bootcamp. 5 cities. $1,490. 40 seats max. Thursday-Friday cohorts, June-October 2026.

Reserve Your Seat
PA

Published By

Precision AI Academy

Practitioner-focused AI education · 2-day in-person bootcamp in 5 U.S. cities

Precision AI Academy publishes deep-dives on applied AI engineering for working professionals. Founded by Bo Peng (Kaggle Top 200) who leads the in-person bootcamp in Denver, NYC, Dallas, LA, and Chicago.

Kaggle Top 200 Federal AI Practitioner 5 U.S. Cities Thu-Fri Cohorts