That's not my interpretation. It sounds like the restructuring is part of the deal that values OpenAI at $100-150 billion [1]. They're not "gifting" away anything any more than a Series A investor gifts something to a Series B investor. They're restructuring so that their stake will be worth more afterwards than it is now, regardless of the percentages. That's what every company owner goes through when the company raise a VC round, goes public, or even just offers employees stock options. That doesn't change because the owner is a nonprofit and it sounds like until they restructure, the for-profit is worth nowhere near that crazy 12 figure number.
"Minority stake" just means that they won't have enough to control the corp out right with 50%+1 which is probably what everyone wants to justify the investment. Reuters TFA says "The plan is still being hashed out with lawyers and shareholders and the timeline for completing the restructuring remains uncertain" so we don't really know what the post valuation numbers look like or who is getting what. We also don't know how the voting vs non-voting shares will split. Losing majority control after multiple multi-billion dollar rounds is the norm so if Microsoft's previous $10 bil investment converts and 10-20 pts go to the employee pool, it's a perfectly fair deal.
> Even putting aside the core issue above (they won't have many shares to sell), the second part of my comment comes back here: what would they buy with all that money? Anthropic? Their explicit mission is to beat for-profit firms in the race to AGI so convincingly that an arms race is avoided. How could they possibly accomplish this after selling control of the most capable AI system on the planet?
As far as I can tell, without this deal OpenAI LLC goes bankrupt under the rumored $5b/yr losses and the charity loses all relevance when the bankruptcy court fire sales the IP. With this deal, it can create a secondary market and use the funds to focus on its actual mission. The only way that his deal doesn't make sense (in my mind) is if you believe that GPT4/o1/whatever are the keys to fulfilling OpenAI's mission and it becomes impossible if it loses control. Personally I find that very hard to believe, which might be the actual disconnect we're having here.
What would they buy? They'd hire people to do research under their umbrella instead of a for-profit one, fund compute infrastructure for researchers who don't have billions for H100s, give grants to organizations, or spin off more startups. Even if it's a 20% stake of a $100 billion, that's enough for a ivy league sized endowment that can fund AI research for generations. If it's a 49% stake of $150 billion, that makes it the second wealthiest charity after Novo Nordisk Foundation - which continues to do tons of biomedical research even though it doesn't have total control over the public Novo Nordisk or the semaglutide IP.
OpenAI would become one of the largest grant giving organizations in the world overnight using just the interest from the endowment. Imagine the equivalent to 50-100% of the NSF's annual grant budget going just to AI research!
> Normal charities have no issues with unrelated business income, because they don't run "trades or businesses", they just spend their money.
See my other reply [2] for a sample of some nonprofits that have for-profit arms. A significant fraction of them do, especially those that offer some sort of product or service (I don't have hard stats but I'd venture it's the majority of the major charities that do the latter). "Normal" charities that just disburse grant money and spends donations is only one type among many.
[1] https://www.reuters.com/technology/artificial-intelligence/o...
