Microsoft has decided its strategic OpenAI partnership isn’t as strategic as it was previously.  According to comments made to the Financial Times (archived):

“We have to develop our own foundation models, which are at the absolute frontier, with gigawatt-scale compute and some of the very best AI training teams in the world,” said the Google DeepMind co-founder who joined Microsoft in 2024.

The software giant is investing heavily in assembling and organising the vast datasets needed to train advanced systems. “That’s our true self-sufficiency mission,” he added.

When OpenAI burst onto the scene in November 2022, it seemed like Microsoft was in the right place with the right investment.  They owned a big chunk of the foundation and hence a seat at the table of frontier research.

Fast forward three and a quarter years, and that seat at the table doesn’t seem quite as valuable.

First, the tech is no longer some secret OpenAI sauce.  Google, Anthropic, and others have created systems with capabilities matching or exceeding ChatGPT’s.  In other words, any large hyperscalar can recreate what OpenAI has created.  So who needs OpenAI?

Second, OpenAI is not making money.  It’s burning cash.  It’s a hyper-incinerator.  OpenAI has no clear path to profitability any time soon, and will continue to need large volumes of investment in order to just keep the doors open.  Every query run against ChatGPT loses money.  OpenAI is not going to provide any return on investment in the near future.  So who needs OpenAI?

And finally, Microsoft probably sees the need for control of a core technology.  Microsoft believes that

“White-collar work, where you’re sitting down at a computer, either being a lawyer or an accountant or a project manager or a marketing person — most of those tasks will be fully automated by an AI within the next 12 to 18 months.”

While that might be a bit pie-in-the-sky, if you believe that, then you want to have as much control over the technology powering that because it will likely make all of your other products obsolete.  If all white collar work is “fully automated by an AI,” then who needs to buy Office licenses?  Microsoft needs to strategically position itself to replace that lost revenue, and it can best do that if it owns the AI service end-to-end.   So who needs…you get the point.

The Microsoft-OpenAI partnership to date has been Microsoft providing the infrastructure (aka credits on Azure) and OpenAI providing the expertise.  But that expertise is no longer needed, and OpenAI famously has over a trillion in infrastructure spending commitments it has no realistic way of paying for.  Microsoft is probably looking ahead and realizing that it’s not going to be paid in the long run to provide services to OpenAI, either because OpenAI can’t pay or because Sam Altman has convinced other competitors (e.g., Oracle) to provide services.

So if it’s not making money off OpenAI, investing in building its own models makes sense.

The loser here is OpenAI, which now will have yet another well-funded competitor.  It can’t hope to compete against Google and Microsoft – two of the richest companies on Earth – as well as Anthropic, given that OpenAI has been in the red since its foundation and VC money is not limitless.  Indeed, one of its biggest VCs – Microsoft – just turned off the spigot.

(Oh, and the artwork for this article was created with the prompt “create a symbolic icon image of a Microsoft AI”…on ChatGPT.  I’m sure they appreciate that).

raindog308

Raindog308 is a longtime LowEndTalk community administrator, technical writer, and self-described techno polymath. With deep roots in the *nix world, he has a passion for systems both modern and vintage, ranging from Unix, Perl, Python, and Golang to shell scripting and mainframe-era operating systems like MVS. He’s equally comfortable with relational database systems, having spent years working with Oracle, PostgreSQL, and MySQL.

As an avid user of LowEndBox providers, Raindog runs an empire of LEBs, from tiny boxes for VPNs, to mid-sized instances for application hosting, and heavyweight servers for data storage and complex databases. He brings both technical rigor and real-world experience to every piece he writes.

Beyond the command line, Raindog is a lover of German Shepherds, high-quality knives, target shooting, theology, tabletop RPGs, and hiking in deep, quiet forests.

His goal with every article is to help users, from beginners to seasoned sysadmins, get more value, performance, and enjoyment out of their infrastructure.

You can find him daily in the forums at LowEndTalk under the handle @raindog308.