How OpenAI makes money: the 2026 revenue breakdown
OpenAI is on track to clear $20 billion in revenue this year, up from $11 billion in 2025. Almost all of it comes from four products. Here's the breakdown, including the segment that's quietly growing 4x faster than ChatGPT.
The short version. OpenAI is on track to do roughly $20 billion in revenue in 2026, nearly doubling from $11.2 billion in 2025. The mix is changing fast: ChatGPT consumer subscriptions still dominate but their share is shrinking as enterprise and API revenue scale. The story under the surface is that enterprise has gone from a sideline to OpenAI's fastest-growing business in 18 months. Here's the four-product breakdown.
The four revenue lines
OpenAI's reported revenue mix for full-year 2025, with our projections for 2026 based on disclosed run-rates and reporting in The Information and Bloomberg:
| Segment | 2025 revenue | 2026 projected | YoY growth |
|---|---|---|---|
| ChatGPT Plus / Pro / Team | $7.4B | $11.2B | +51% |
| API (developer + business) | $1.6B | $3.5B | +119% |
| Enterprise (custom contracts) | $1.4B | $3.8B | +171% |
| Microsoft licensing | $0.8B | $1.5B | +88% |
| Total | $11.2B | $20.0B | +79% |
The headline: ChatGPT consumer is still the biggest line, but it's now growing slower than the total. The growth story is enterprise.
ChatGPT consumer: the slowing flywheel
ChatGPT Plus ($20/month) and ChatGPT Pro ($200/month) together represent OpenAI's biggest revenue line, projected at $11.2 billion for 2026. The growth rate matters more than the absolute number: Plus subscriptions grew 73% in 2025 and we project 51% growth in 2026.
The slowdown is real and predictable. The total addressable market of people who will pay $20/month for an AI chatbot is finite, and OpenAI is approaching saturation in major English-speaking markets. The continued growth comes from emerging markets (where Plus is now accepted with local cards), Pro tier upgrades (the $200/month plan grew 4x in 2025), and incremental conversion from free.
The interesting question is what happens when consumer ChatGPT plateaus. The $200/month Pro tier is OpenAI's answer: if you can't add more subscribers, increase ARPU. The Pro tier's deep-research and operator features are explicitly positioned to justify the 10x price difference. Our reporting suggests Pro now accounts for ~23% of consumer ChatGPT revenue despite being <5% of subscribers.
API: the developer business that grew up
The API business — pure consumption-based billing for developers building applications on top of OpenAI's models — is OpenAI's second-largest line. Importantly, this is the line most exposed to competitive pricing pressure. DeepSeek, Gemini Flash, and Claude Haiku have all been undercutting OpenAI's API for over a year now.
OpenAI's response has been to compete on capability, not price. GPT-5 is priced 30-40% above competitors but offers measurably better reasoning and multimodal performance. The bet is that developers will pay a premium for the best-in-class model rather than route to the cheapest. So far, that bet is working: API revenue is on track to more than double in 2026.
The under-discussed part of the API business: batch and committed-use discounts drive 35% of API volume. Customers who commit to spending a minimum monthly amount get 25-50% off list pricing. This is how OpenAI prevents large customers from defecting to DeepSeek on price; it's also why the headline API price is misleading as a competitive comparison.
For builders, our AI Pricing Tracker shows live published rates across all major providers.
Enterprise: the growth story nobody is writing about
OpenAI's enterprise business — custom contracts with Fortune 1000 companies for ChatGPT Enterprise, dedicated model capacity, and on-premise deployments — was a sideline in 2024. It is now growing 171% year-on-year and on track to overtake the API business in late 2026.
The unit economics are different from consumer. A Fortune 500 customer might pay OpenAI $5-25 million per year for ChatGPT Enterprise across 30,000 seats plus reserved API capacity. Gross margins are higher (consumer subscription gross margin is roughly 75%; enterprise is 80%+). Customer acquisition cost is enormous but the contracts last years.
What's driving the growth: every Fortune 500 board has "AI strategy" on the agenda for 2026, and the safest answer to the board question "how are we using AI" is "we have ChatGPT Enterprise deployed." OpenAI is the default; competitors (Anthropic Enterprise, Google's Gemini Enterprise) are catching up but still trail in deal count.
The risk: this revenue is concentrated. We estimate that 60% of OpenAI's enterprise revenue comes from the top 50 customers. Losing five of them would meaningfully dent the line.
Microsoft licensing: the quiet line
OpenAI's commercial relationship with Microsoft is the most underdiscussed line because the terms aren't publicly disclosed. What we know from reporting: Microsoft pays OpenAI a per-query licensing fee for serving OpenAI models through Azure, plus a separate licensing fee for using OpenAI technology in Copilot and other Microsoft products.
The estimate of $0.8B for 2025 (growing to $1.5B for 2026) is consistent with reported deal terms and Microsoft's disclosed Copilot revenue figures, but the exact number is opaque. The relationship is complicated by the recent renegotiation following the original 49% profit-share arrangement's expiration in late 2025.
What this tells us about the AI business broadly
Three lessons from OpenAI's revenue mix that apply to the AI industry as a whole:
-
Consumer AI hit its first plateau in 2026. ChatGPT growth is slowing. Claude consumer growth is slowing. The B2C AI market is approaching saturation in developed markets. Future growth comes from emerging markets and from selling higher-priced tiers to existing users.
-
Enterprise is where the next 10x is. Every frontier lab is shifting investment from consumer to enterprise. The dollar size of the enterprise opportunity dwarfs consumer; the deal cycles are longer but the LTV is enormous.
-
API pricing race-to-the-bottom is real but bounded. Cheap models (DeepSeek, Gemini Flash) are taking volume share, but premium models keep winning on revenue share because the highest-value use cases pay for quality. OpenAI hasn't lost the dollar war yet, though it's clearly lost the volume war.
What we're watching next
- OpenAI's rumoured IPO has been pushed to late 2027 according to reporting in The Information. The 2026 revenue trajectory will determine pricing if it happens.
- The Microsoft renegotiation terms, when fully disclosed in OpenAI's next investor update, will tell us the real shape of the Microsoft relationship.
- The enterprise customer concentration risk: OpenAI hasn't disclosed top-10 customer concentration. When (if) they do, that number will tell us a lot about the durability of the enterprise line.
Sources
- OpenAI investor disclosures (May 2026)
- The Information, multiple reports on OpenAI revenue and customer concentration
- Bloomberg, ChatGPT Enterprise deal reporting
- Public statements from Sam Altman at AI Summit (March 2026)
Related on Meridian48
One email. The week in AI, Pakistan tech, and global business.
Curated by Faizan Khan. No filler. Unsubscribe in one click.
Faizan Ali Khan is the Founder and Editor of Meridian48 and the Founder of Cubitrek, a technology consulting practice. He writes about AI, Pakistan's technology economy, and the business of innovation.
More from this author →