This piece gives you a quick introduction to OpenAI. You’ve almost certainly used, or at least heard of, ChatGPT, but most people can’t really explain who the company behind it is or how it grew to its current size.
OpenAI was founded in 2015, starting out as a non-profit AI research lab, with founders including Sam Altman and Elon Musk. A decade later, it has become one of the world’s most valuable AI companies. In the single year from 2025 to 2026, its post-money valuation jumped from about US$300 billion to over US$800 billion; its ChatGPT is used by nearly 900 million people each week, and its product line stretches from chat and image generation all the way to coding (Codex).
Its biggest difference from rivals is being “consumer-facing.” Compared with the enterprise-focused Anthropic, OpenAI takes the mass-market consumer route: first get hundreds of millions of people using ChatGPT every day, then layer enterprise subscriptions, the API, and ecosystem monetization on top.
Remember it in one line: the company that wants to turn AI into an everyday consumer staple and wants to do everything.
It also carries an ever-debated issue: the structural shift from non-profit to for-profit, and the governance turmoil that came with it. This shift drew a lawsuit from Elon Musk, and to this day the case isn’t fully settled; in late 2023 CEO Sam Altman was even briefly ousted by his own board. All of which makes the question “can a lab with a mission to benefit humanity also be one of the world’s most valuable AI companies?” one it can’t shake off.
Key Data Snapshot
Let’s put the key numbers in one place first. OpenAI isn’t public yet and hasn’t released complete financials, so quite a few figures below are the company’s own public framing or third-party estimates. We try to make clear which are officially confirmed and which are outside estimates.
| Item | Data |
|---|---|
| Year founded | 2015 |
| Company type | Two-tier structure: non-profit foundation (OpenAI Foundation) + public benefit corporation (OpenAI Group PBC), not public |
| Co-founders | Sam Altman, Greg Brockman, Ilya Sutskever, Elon Musk, and others |
| CEO / Headquarters | CEO Sam Altman; headquartered in San Francisco, California, USA |
| Latest valuation | Post-money about US$852 billion as of March 2026, among the highest in the world; that round in March 2025 was post-money about US$300 billion, nearly tripling in a year |
| User scale | ChatGPT around 900 million weekly active users, about 50 million paying subscribers, the vast majority free users; figures are the company’s public framing from early 2026 |
| Annual recurring revenue (ARR) | About US$20 billion for 2025, the company’s annualized estimate framing, not the actual full-year intake |
| Actual annual revenue | About US$13.1 billion for 2025 per third-party estimates, unaudited, clearly lower than the annualized ARR |
| Flagship products | ChatGPT, Codex, OpenAI API; the flagship GPT-series models |
Two reminders for reading the numbers, useful for any AI startup. ① ARR (annual recurring revenue) is recent revenue multiplied by 12 to annualize, not the same as that much actually coming in over a full year; ② private companies have no audited financials, so revenue and user counts are mostly estimates, getting the “order of magnitude and trend” right is more solid than chasing exact values.
A Quick Tour of Seven Dimensions
You can get to know an AI company through seven dimensions. We’ll have more detailed standalone pieces on each later.
① Technology and product direction: The flagship models are the GPT series. Its product line is the broadest among peers, covering chat, coding (Codex), and image generation (GPT Image); in 2025 it also released an open-weight model for the first time, breaking from its previously purely closed-source approach. It is also working with Broadcom on in-house inference chips to reduce its dependence on Nvidia; outsiders sometimes refer to this chip by code names like “Titan,” but the company hasn’t officially named it. It has also extended its reach from software into hardware, acquiring Jony Ive’s device startup io, and according to analyst reports there is also talk of it negotiating partnerships with Qualcomm and MediaTek on AI phone chips.
② Customer structure and market positioning: It started in the mass consumer market, with ChatGPT’s hundreds of millions of users as its biggest asset, then extended to enterprise subscriptions and API developers. Together with the enterprise-focused Anthropic and Google, which binds AI into search and the cloud, they are effectively pushing toward the middle from different ends of the market. It’s worth noting that while ChatGPT’s user numbers are staggering, the vast majority don’t pay, and the real revenue mainstay comes from enterprises and the API.
③ Ecosystem and partnership strategy: Behind it stands a lineup of deep-pocketed backers who are also compute partners, including Microsoft, Amazon, Nvidia, and SoftBank. It is also the core of Stargate, a massive data center alliance. Its relationship with Microsoft has also loosened from “exclusive” to “multi-cloud,” so it can offer services on more cloud platforms, but Microsoft remains its primary cloud partner to this day.
④ Valuation and financial model: It has closed one of the largest private rounds in history, lifting its valuation among the highest in the world; in just one year, its post-money valuation rose from about US$300 billion to over US$800 billion, nearly tripling. But there’s a gap between ARR and revenue actually booked, and valuation multiples and scenario projections are all built on assumptions, more of an exercise in understanding finances than a market price.
⑤ Commercialization risk and regulation: Its unit economics are unusual: every time a user asks a question, compute gets burned behind the scenes, and the larger the scale, the heavier the compute bill. Variables to watch over the long term include the cash-burn rate and the timing of turning cash-flow positive, the pricing pressure from open-weight models, and regulatory currents like copyright lawsuits and AI rules.
⑥ Geopolitics and supply chain: OpenAI’s appetite for compute is among the largest in the world, and it can’t get around Nvidia’s GPUs, TSMC’s advanced processes, CoWoS (advanced packaging), and HBM (high-bandwidth memory), and it also drives enormous data-center power demand, the Stargate data center alliance exists precisely for this. Geopolitical variables like export controls also indirectly affect the pace of its expansion. For how the whole chain works, see The AI Hardware Supply Chain, End to End.
⑦ Leadership and governance: OpenAI’s governance history is dramatic in its own right. CEO Sam Altman was abruptly removed by the board in late 2023, only to return within five days under pressure from employees and investors, with the board reshuffled accordingly; co-founder Ilya Sutskever, former CTO Mira Murati, and other core figures later left one after another to start their own ventures or join rivals. In terms of governance structure, the operating public benefit corporation is about a quarter owned by the non-profit OpenAI Foundation, which holds the power to appoint and replace the entire board. A non-profit parent with a “mission to benefit humanity” controlling one of the world’s most valuable AI companies, this tension is a throughline you can’t get around when trying to understand OpenAI.
Key Milestones
Let’s pick out the pivotal moments that brought OpenAI to where it is today:
| Time | Milestone |
|---|---|
| 2015 | Founded as a non-profit AI research lab, with founders including Sam Altman, Elon Musk, and others |
| 2018 | Elon Musk leaves the board; there are two accounts of why, one citing a conflict of interest with Tesla, the other a failed bid for control |
| 2019 | Establishes a for-profit subsidiary; Microsoft invests and becomes the primary cloud partner |
| 2022 | Launches ChatGPT, igniting the global generative AI boom |
| 2023 | The board briefly fires CEO Sam Altman, who is reinstated five days later as the board is reorganized |
| 2024 | Co-founder Ilya Sutskever, CTO Mira Murati, and other core figures depart one after another |
| 2025 | Closes a large private round at a post-money valuation of about US$300 billion; releases an open-weight model for the first time; completes a year-end restructuring, establishing the two-tier “non-profit foundation + public benefit corporation” structure |
| Early 2026 | Closes one of the largest private rounds in history, at a post-money valuation of about US$852 billion, reaching among the highest in the world; ChatGPT around 900 million weekly active users, about 50 million paying |
| First half of 2026 | Shifts to a non-exclusive, multi-cloud arrangement with Microsoft, though Microsoft remains the primary cloud partner; a new generation of GPT models is unveiled; the video product Sora shuts down its web and app and keeps only the API; OpenAI wins the first instance of the Elon Musk lawsuit but it’s still under appeal |
Milestones will keep being added, with figures based on the latest announcements; this table was last compiled in May 2026.
Further Reading and Upcoming Standalone Pieces
If you want to read deeper next, Penchan will break each dimension into a standalone piece, rolling them out over time:
- Why Was Sam Altman Fired and Then Brought Back? OpenAI’s Five-Day Board Storm (leadership and governance)
- Who Runs OpenAI? The Founders, Current Executives, and the Geniuses Who Left (leadership team)
- What Products Does OpenAI Have? The Full Map from ChatGPT and Codex to In-House Chips (technology and products)
- How Much Is OpenAI Worth? The IPO Timeline and How to Read “$800 Billion” (valuation and finance)
- OpenAI vs Anthropic: The Two Playbooks Behind ChatGPT and Claude (customers and competition)
- What Is Stargate? OpenAI’s Compute Gamble and Multi-Cloud Map (ecosystem and partnerships)
- OpenAI’s Legal Battles: The Musk Lawsuit, Copyright, and Regulatory Pressure (commercialization risk and regulation)
- OpenAI’s Compute Achilles’ Heel: Supply-Chain Exposure from Nvidia and TSMC to Power (geopolitics and supply chain)
- The Complete Pre-IPO Deep-Dive Report (2026 Q1 Edition)