Source: OpenAI will get all of the equity back once $92B in profit plus $13B in initial investment are paid to Microsoft and once other investors earn $150B
I don't usually write about business deals, much less about rumors about business deals, but this one has me scratching my head …
The Road to AI We Can TrustGary Marcus
Context & Ripple Effects
This rumored term sheet puts a price tag on unwinding the structure Microsoft helped build: what began with Microsoft's $1B bet on OpenAI in 2019 grew into a ~$10B commitment that was largely delivered as Azure cloud credits, giving Microsoft both leverage and exposure rather than clean equity.
Microsoft's negotiating position changes immediately: instead of an unspecified equity claim, it now has a sequenced cash-out — $92B in profit plus its $13B back — which it can accept, contest, or trade against continued model access past 2030.
OpenAI gains a concrete path to full ownership of its commercial arm, but only after clearing a combined ~$255B payout hurdle to Microsoft and other investors.
Second-order effects
Because much of Microsoft's original commitment arrived as cloud purchases, part of any settlement likely recirculates as Azure spend — meaning the 'payment' partially flows back to Microsoft's own infrastructure business regardless of the equity outcome.
Other investors' $150B earn-out sets a cap-shaped template that will shape every future funding round: late entrants are effectively buying senior claims on a fixed waterfall rather than open-ended upside.
Third-order effects
If this structure holds, frontier-lab financing formalizes into something closer to structured debt than venture equity — capped returns, sequenced payouts, and compute credits as currency — with the 2019 critics' question about whether OpenAI can still democratize AI resurfacing under a fully for-profit owner.
A cleared buyback path also removes the main obstacle to an OpenAI IPO, which would make public-market disclosure requirements apply to a frontier lab for the first time.
The trend: Frontier labs are converting philanthropic-era capital structures into capped, sequenced payout waterfalls — trading investor equity for eventual independence while compute providers double-dip as creditors.
Simply remarkable: @sama, who cofounded and has steered the hottest AI startup of the last decade, chose to hold ~zero~ equity in the for-profit arm he has built. https://www.theinformation.com/ ... @OpenAI https://twitter.com/...
@GaryMarcus ... Earn $150B ? How much did they earn to date? Seems a bit “aspirational” to put it mildly. Very mildly. Just for perspective: https://twitter.com/...
At that point, do you still get to call it a non-profit? You know, when you're generating more than $10B lifetime and 98% owned by companies and investors? https://twitter.com/...
nice visual + reporting from fortune that shows why every story about OpenAI's fundraising includes a caveat about a complex structure that makes the valuation less relevant than other deals https://fortune.com/... https://twitter.com/...
@GaryMarcus ... Seems to me it could be both a firesale and hedge. If Altman has no idea how to make money, but everyone is telling him this could be huge, then I suppose you'd want a deal like this if you can get it. Personally, I think Microsoft is badly overpaying.
Update on MSFT and OpenAI deal - proposing nonprofit status again after profit paid back - potentially increasing investor cap. Instead of hard cap, 20% increase per year - Altman hasn't got equity in the for profit arm. Unclear if this deal changes that https://twitter.com/...
Lots of great, new details about the $10 billion Microsoft/OpenAI deal — including the fact that $MSFT previously invested $3 billion (not just $1 billion) in OpenAI. Inside the structure of OpenAI's looming new investment from Microsoft and VCs https://fortune.com/...