So there has never been a “start-up” like OpenAI. Already with 800,000,000+ users and a breathtaking $500 Billion (!) valuation.
But it’s also raised and spent epic amounts. So as the chart from FT below shows — all the historic investors together, as a group, will likely own 10% after the company converts to a for-profit. All of them. From Khosla to Thrive to Founders Fund and more.
OpenAI has now raised approximately $58 billion in primary funding across 11 rounds, with an additional $6.6 billion secondary sale in October 2025 at the $500 billion valuation.
But the cash burn? That’s where it gets interesting.
The company now projects it will burn $115 billion through 2029—that’s $80 billion higher than previously expected. Let that sink in. Not $115 billion raised. $115 billion burned.
Breaking it down:
- 2024: $8 billion burn (up $1.5B from earlier projections)
- 2025: $17 billion burn (more than doubled, $10B higher than projected)
- 2027: $35 billion burn
- 2028: $45 billion burn
This isn’t mismanagement. This is the cost of competing in frontier AI.
The Valuation Trajectory Looks Like a Rocket Ship
The numbers are staggering:
- 2023: $29 billion
- Early 2024: $80 billion
- October 2024: $157 billion
- March 2025: $300 billion (after $40B primary round)
- October 2025: $500 billion (secondary sale)
A 17x increase in less than three years. But when you’ve raised $58 billion and are burning over $100 billion to get there, the ownership math tells a different story.
Understanding the Ownership Reality
Looking at the projected ownership post-restructure (per FT Research and other sources):
- Microsoft: ~30% (after investing nearly $14 billion across multiple rounds, plus Azure cloud revenue)
- OpenAI employees: close to 30% (after multiple equity grants and secondary sales to retain talent)
- OpenAI nonprofit: 20-30% (the original controlling entity, which must maintain meaningful ownership)
- SoftBank: ~10% (after leading the $30 billion March 2025 round plus participating in the October secondary)
- Other investors: remaining ~10-15% (Thrive, Khosla, Tiger Global, Sequoia, a16z, Fidelity, Nvidia, T. Rowe Price, Dragoneer, and 50+ others splitting what’s left)
For founders and early investors who weren’t Microsoft, the dilution has been severe. Eight of the eleven original co-founders have left. Early VCs who got in at sub-$30 billion valuations have seen their ownership percentages compressed dramatically—even as the absolute dollar value has grown.
The cap table now includes 11 funding rounds.
The Investor Dynamics Are Unlike Most Anything We’ve Seen
What makes OpenAI’s story unique is who’s investing and why:
Microsoft isn’t just an investor—they’re existentially dependent on OpenAI. Their entire AI strategy—Azure growth, Office Copilots, enterprise AI positioning—runs through OpenAI infrastructure. They’ll keep funding to protect that relationship. But they also receive ~20% of OpenAI’s revenue through their partnership agreement, creating a circular funding dynamic.
Nvidia invested $100 million in October 2024 and gets paid back immediately through GPU purchases. They’ve invested in 50+ AI startups creating a circular economy where investment dollars flow back as compute revenue. Part of Microsoft’s and Nvidia’s investment effectively returns to their own coffers.
SoftBank went from a $500 million investor in October 2024 to leading a $30 billion round six months later. That’s not normal venture behavior—that’s Masayoshi Son making a macro bet on AGI. Their commitment is so aggressive that if OpenAI doesn’t complete its for-profit conversion by December 31, 2025, SoftBank’s investment could be slashed from $30 billion to as low as $20 billion.
Traditional VCs are playing a different game: raising larger funds on the AI narrative, marking up valuations in subsequent rounds, and hoping an exit materializes. But with 60+ investors on the cap table, exit dynamics will be complicated.
The Secondary Sale Tells Another Story
The October 2025 secondary sale is particularly revealing. OpenAI had authorized up to $10.3 billion in shares for sale but only about $6.6 billion ultimately changed hands.
Why did employees sell less than authorized? Two possibilities:
- They believe the upside justifies staying locked in
- The valuation is already so high that even smaller positions provide life-changing liquidity
Either way, this was the second major secondary in a year—a clear signal that liquidity events are necessary to retain talent competing with Meta’s nine-figure offers and public company stock compensation.
The For-Profit Conversion Remains Unresolved, For The Moment
Here’s the structural complexity that matters: current investors still don’t own actual equity. They hold convertible notes that only convert to shares after the nonprofit-to-for-profit restructuring completes.
This requires:
- Approval from Microsoft (who has special governance rights)
- California Attorney General approval (nonprofit oversight)
- Delaware Attorney General approval (incorporation jurisdiction)
- Surviving Elon Musk’s lawsuit challenging the conversion
The deadline? December 31, 2025.
If the conversion fails, SoftBank’s investment drops from $30 billion to $20 billion, and other investors can demand their money back with 9% interest. That’s a potential $40+ billion liability.
What This Means for AI Founders and Investors
The OpenAI model isn’t replicable—but the dynamics are instructive:
For AI infrastructure founders: If you’re building frontier models, you’re signing up for capital requirements that dwarf traditional SaaS. You’ll need to raise 4-5x your revenue run rate continuously. You might own 5-15% of something huge rather than 25% of something modest. And you’ll need to accept strategic investors who become customers, partners, and dependencies simultaneously.
For AI application founders: The good news is you probably don’t need the OpenAI model. If you’re building on top of foundation models, your capital requirements are dramatically lower. But you’re also betting that OpenAI, Anthropic, or another provider maintains reasonable API pricing as they burn billions.
For investors: Traditional venture math is broken in frontier AI. A 10x return on a $500 billion valuation requires a $5 trillion outcome. You’re not investing for returns—you’re investing for:
- Strategic positioning (Microsoft, Nvidia)
- Circular economics (vendors becoming investors)
- Market creation bets (SoftBank)
- Or you’re Thrive/Sequoia marking up your earlier positions
The Revenue Story Provides Some Justification
To be fair, the revenue growth is real:
- $4.3 billion in revenue during first half of 2025
- Projected $12.7 billion for full year 2025
- Targeting $100 billion by 2029
- 500 million weekly active users
- 20 million paid ChatGPT subscribers
At a $500 billion valuation on $12.7 billion revenue, that’s roughly a 39x revenue multiple. For context, Salesforce trades at 8-9x, ServiceNow at 20x. But both are profitable.
The bet is that by 2029, at $100 billion in revenue, a $500 billion valuation would be 5x revenue—reasonable for a high-growth software company. But that requires executing flawlessly for four years while burning $115 billion and fending off Google, Meta, Anthropic, and everyone else building AGI.
The Bottom Line
Yes, OpenAI scaled to a $500 billion valuation faster than any company in history.
Yes, they’re generating $12.7 billion in revenue with 500 million weekly users.
Yes, they’ve created a new market category and changed how the world thinks about AI.
But they’ve also:
- Raised $58 billion in primary capital
- Completed $6.6 billion in secondary sales
- Projected $115 billion in cash burn through 2029
- Diluted ownership across 60+ investors over 11 rounds
- Created a cap table so complex it requires nonprofit-to-for-profit conversion to unlock
- Built a dependency structure where major investors are also customers and vendors
The valuation is spectacular. The dilution is equally spectacular. The capital requirements are unprecedented. The burn rate is staggering.
All of these things are true simultaneously.
This is the new reality of frontier AI. Not every hot AI deal will look like this, but many infrastructure plays will require similar capital intensity. The question isn’t whether you can raise at massive valuations—OpenAI proved you can. The question is whether the dilution, complexity, and burn make sense for your business model and your exit strategy.
For most AI companies, the answer is no. For foundation model companies competing for AGI, it might be the only path forward.
