From a nonprofit research lab to the most valuable private company in history. Scroll through every round, every dilution event, every structural twist that created the most chaotic ownership structure in tech.
Sam Altman, Elon Musk, Reid Hoffman, Peter Thiel, Greg Brockman, and others announce OpenAI as a nonprofit research lab. They pledge $1 billion, though only ~$130M is actually collected by 2019.
The mission: "ensure that artificial general intelligence benefits all of humanity." The Foundation owns everything. There are no investors, no equity, no cap table. Just a research lab and a dream.
One month after releasing GPT-2 — which they initially called "too dangerous to release" — OpenAI creates "OpenAI LP," a capped-profit subsidiary. Microsoft invests $1B for a minority stake and becomes the exclusive cloud provider.
The profit cap: first-round investors can earn at most 100x their investment. Everything above that reverts to the nonprofit. This sounds reasonable in 2019. It will not age well.
This era covers three major releases: GPT-3 (Jun 2020), DALL-E (Jan 2021), and ChatGPT (Nov 2022). Each one raised the stakes. Microsoft adds $2B in 2021, then another $10B in January 2023 — weeks after ChatGPT hit 100M users. Total investment: ~$13B. Microsoft now holds ~32.5%.
GPT-4 dropped in March — the first model to pass the bar exam. One month later, Thrive Capital leads ($1.3B eventual total), joined by a16z, Sequoia, Khosla Ventures, and others. Every fund in Silicon Valley wants allocation.
Early angels — Hoffman, Thiel, Kutcher's Sound Ventures — are sitting on 30-140x paper returns from their founding-era checks. Khosla's early bet is already worth 30x.
The nonprofit board fires Sam Altman on November 17, citing he "was not consistently candid." Within 48 hours, 95%+ of employees threaten to leave. Microsoft offers to hire the entire team.
Altman is reinstated 5 days later. The board is reconstituted with Bret Taylor as Chair and Larry Summers joining. This episode proves the nonprofit board has real power — and that power is destabilizing.
GPT-4o (May) made the flagship model free. o1 (September) introduced "reasoning" and outperformed PhDs on science benchmarks. One month later, Thrive Capital leads again. SoftBank enters with $500M, Nvidia with $100M, plus a16z, Fidelity, MGX, and Tiger Global. Valuation 5x's in 18 months.
In the background, SoftBank has been buying employee shares on secondary markets at an ~$80B valuation. The cap table is getting crowded.
GPT-4.5 (February) and o3-mini dropped weeks before the round closed. The pattern is now unmistakable: ship a model, then raise. SoftBank leads a $40B round at $300B. Blackstone, TPG, and T. Rowe Price join. OpenAI is now worth more than Goldman Sachs, Morgan Stanley, and Citigroup combined.
OpenAI ARR has tripled from ~$6B to ~$20B. Monthly revenue hits $1.5B. The AI revenue machine is real, but it's still burning cash on compute.
The defining structural event. OpenAI Inc becomes the "OpenAI Foundation" (nonprofit). The for-profit arm reforms as "OpenAI Group PBC". The 100x profit cap is removed entirely.
The Foundation retains ~26% equity and controls 100% of board appointments. California AG Bonta and Delaware AG Jennings both approve.
o4-mini and GPT-4.1 ship in quick succession. Then: the largest private funding round in history. Amazon ($50B), SoftBank ($30B), and Nvidia ($30B) lead. OpenAI also raises $3B from retail investors via bank channels and ARK Invest.
Microsoft — diluted from 32.5% to 26.79% — sits on a $228B position from $13B invested. The Foundation's 25.8% is worth $220B, the most valuable nonprofit asset in human history.
Paper returns at the $852B valuation, sorted by multiple. The range from 140x to ~1x tells the story of when you got in.
The CEO of the most valuable private company in history owns exactly zero percent of it. In September 2024, Bloomberg reported a 7% stake was being discussed — worth ~$10.5B at the time. The board said "no specific figures have been discussed." At the PBC conversion, he received nothing.
He earns a $76,001 annual salary. His cap table entry reads "TBD/Pending." Prediction markets on Kalshi are actively trading on whether he'll receive equity in 2026.
He's not poor — he's worth ~$2B+ from Helion Energy, Worldcoin, Reddit, and other personal investments. But the optics are unprecedented.
Four structural eras in eleven years. Each one more complex than the last.
501(c)(3) research lab. No equity, no investors, no profit motive. Funded by donations.
Simple"OpenAI LP" — a for-profit arm under the nonprofit. Investors capped at 100x returns. Nonprofit retains control and gets all excess profits. No one fully understood this structure.
The cap was quietly modified in 2023 to increase by 20% annually starting 2025, effectively inflating the ceiling toward infinity.
Complex"OpenAI Group PBC" (Public Benefit Corporation) replaces the capped-profit. The 100x cap is removed entirely. The nonprofit becomes the "OpenAI Foundation" with 25.8% equity and 100% board control.
Required approval from California and Delaware AGs. Over 50 orgs protested, calling the deal "full of holes." The Foundation's $130B+ stake is paper wealth, not cash for grantmaking.
UnprecedentedS-1 expected Q3 2026. IPO Q4 2026 or Q1 2027 on NASDAQ. Potential valuation: $830B–$1T. The nonprofit Foundation will be the largest non-government shareholder of a public company.
Meanwhile, Elon Musk's fraud lawsuit goes to trial on April 27, 2026 — four weeks before the expected S-1 filing.
TBDEvery major tech IPO had a recognizable cap table. OpenAI's is in a category of its own.
| Company | Structure at IPO | Founder Equity | Top Investor | Time to IPO |
|---|---|---|---|---|
| C-corp, dual-class shares | ~30% (Page + Brin) | Kleiner & Sequoia ~15% | 6 years | |
| C-corp, dual-class shares | ~28% (Zuckerberg) | Accel ~10% | 8 years | |
| Uber | C-corp, standard | ~8.6% (Kalanick, post-ouster) | SoftBank ~16% | 10 years |
| Stripe | C-corp, dual-class | ~27% (Collisons) | Sequoia ~?% | 14+ years |
| SpaceX | C-corp, standard | ~42% (Musk) | Fidelity ~?% | Not yet public |
| OpenAI | Nonprofit → PBC. Foundation controls board. | 0% (Altman) | Foundation 25.8%, Microsoft 26.8% | 11 years |
Every company above had a founder with significant equity, a standard corporate structure, and investors who understood what they owned. OpenAI has none of these.
OpenAI's revenue is real and growing fast. But at 42x forward revenue, it's priced for perfection.
Buried in the Microsoft partnership is one of the strangest contractual provisions in corporate history.
Under the original 2019 deal, if an "expert panel" appointed by OpenAI's board certifies that the company has achieved Artificial General Intelligence, Microsoft loses access to AGI technology. The revenue-sharing agreement ends. The API access ends. The exclusive cloud partnership — functionally ends.
The definition of AGI in the contract has never been made public. No one outside OpenAI's board and Microsoft's legal team knows what threshold triggers the clause. And both sides have enormous financial incentives to interpret it in their favor.
In December 2025, Microsoft renegotiated its deal as part of the PBC conversion. The new terms reportedly soften the AGI cutoff — but the exact language remains under NDA.
The man who helped create OpenAI is now suing it for fraud. The trial starts April 27, 2026 — weeks before the expected S-1 filing.
A nonprofit → capped-profit → uncapped PBC, with a CEO who owns nothing, a foundation worth $220B that controls the board, retail investors buying private shares through ETFs, a co-founder suing for $134B weeks before the S-1, and 69 investors across 12 rounds totaling $168B. The lawyers are billing more per hour than most engineers earn per month.