OpenAI vs. Microsoft – Soap Opera

Microsoft holds most of the cards in this fight.

• OpenAI and Microsoft are increasingly at odds over what position Microsoft will have in OpenAI once it has restructured into a for-profit company, but OpenAI’s need for cash from SoftBank and the contracts it has already signed mean that Microsoft has the upper hand.
• OpenAI’s $40bn financing from SoftBank requires that OpenAI completes its restructuring into a for-profit company, and to do that it has to get Microsoft to sign off on the deal.
• To complicate matters, Open AI has to get this done by the end of 2025 or half of the money promised may never materialise.
• This would be a serious problem given how much cash OpenAI is burning, both in its pursuit of super-intelligent machines and in supporting 500m users, most of whom don’t pay for the service.
• Over the last 6 years, Microsoft has poured $13bn into OpenAI and in return, it gets preferential rights to profit streams and access to OpenAI’s technology until 2030.
• OpenAI also exclusively uses Azure as its compute provider, meaning, that almost all of the $13bn that Microsoft spent came right back in the form of revenues to Azure.
• At the heart of the discussion is just what stake Microsoft will have once the restructuring is complete in return for giving up what looks like a preference shareholder position as well as its exclusive supply arrangement.
• The range is wide with market chatter ranging from 20% to 49%, and I suspect that at the end of the day, the number will be closer to the top end of the range.
• This is because all Microsoft has to do to spike OpenAI’s guns is walk away or agree to nothing.
• Microsoft will be relatively unaffected if the current contract continues as it has been written, but OpenAI will be forced to go and look for more money, which will be hard to get, seeing as any return on it will be very unclear.
• This is because OpenAI’s current corporate structure is a conflicted mess of a for-profit subsidiary of a non-profit company, which has caused the company to almost implode once already.
• I have long viewed Microsoft’s sole-source relationship with OpenAI as a risk not worth taking, and in the last 12 months, Microsoft has clearly moved to address that risk.
• Its AI services now support the models of OpenAI’s rivals, as Microsoft sees (correctly in my view) the real value being in the services that are built on top of the models rather than the models themselves.
• Furthermore, with its acquisition of Windsurf and the launch of new products, OpenAI increasingly looks like becoming a competitor rather than a partner.
• Consequently, I think that Microsoft has the upper hand and will end up with 30% or more of OpenAI if a deal is reached which would probably go into its books at around $100bn.
• This would be a tidy resolution to a messy problem and would deliver Microsoft a return of 7.7x, which I suspect is where there may be some desire on Microsoft’s part to get a deal done.
• Either way, Microsoft emerges as the winner from this tussle.