OpenAI Plans To Shift Compute Needs From Microsoft To SoftBank

According to The Information (paywalled), OpenAI plans to shift most of its computing power from Microsoft to SoftBank-backed Stargate by 2030. TechCrunch reports: That represents a major shift away from Microsoft, OpenAI's biggest shareholder, who fulfills most of the startup's power needs today. The change won't happen overnight. OpenAI still plans to increase its spending on Microsoft-owned data centers in the next few years.

During that time, OpenAI's overall costs are set to grow dramatically. The Information reports that OpenAI projects to burn $20 billion in cash during 2027, far more than the $5 billion it reportedly burned through in 2024. By 2030, OpenAI reportedly forecasts that its costs around running AI models, also known as inference, will outpace what the startup spends on training AI models.

Sure...

[locater16]

I plan to shift all my transportation needs to flying unicorns by 2040, all that stands in my way is sourcing some flying unicorns.

Re:

[Tough Love]

Spotted the microsoft employee.

Who cares?

[rlwinm]

By 2030 the AI fad will be bust. There will probably be some equally useless pie-in-the-sky fad in full swing by then.

In the mean time it would be nice if these guys would solve some real world problems that affect millions: Such as removing all the AI buttons all over GMail's composer. I think doing that would be a huge win for Humanity.

Re:Who cares?

[evanh]

Sadly, it's damn good at identifying and tracking individuals. It's the ideal big-brother tool so it's going to continue to have big Pentagon spending.

OpenAI (Mr Altman) wants to fully compete with Palantir and the likes. M$ likely is not entirely on board with that direction.

Re:

[fuzzyfuzzyfungus]

Microsoft wanted out of that unprofitable and silly looking thing with the VR headsets; but they've been trying to sell AI to the feds since before the current hype phase. Their PR fluff for the Air Force ABMS and broader enthusiasm for JADC2 specifically mention using 'Azure Cognitive Services' for "integrated, intelligent apps enabling rapid human decision–making, even without having direct AI or data science skills or knowledge."

They aren't the sorts that just outright name the company after a c

Ok, but here me out, from a marketing standpoint

[ebunga]

We can use our AI data to learn about consumer preferences and then use that to offer up ads when they sign up to our free wifi service advertised at the gas pump so they'll be encouraged to come into the store and buy some fried chicken because they're ignoring the the signs on the pump, the signs out front, the signs on the front door and windows of the store and the smell of that Fried Gold wafting through the air, and all the signs and sights and smells when they go into the store to use the bathroom before leaving and not buying our extremely profitable fried chicken. This should only cost about $120,000 per store and can give us upwards of seven or eight new customers. I mean, that's $120,000 per store per month and eight new customers over the expected life of the store, but, just think of what else we could do if this pans out?

Re: Ok, but here me out, from a marketing standpoi

[Bodrius]

I see where you are going wrong... the way you spin... I mean, calculate this is you are spending 120k for customer acquisition of 6-8 new customers, *for the life of the customer*.

I mean this is 30-40 years... ok, 20 years because they are clearly eating fried chicken every day. But thats fried chicken profits 365 days * 20 years, it probably pays for itself with just 3 customers and thats without accounting for synergies and other product placement.

Can I has money now?

Re:

[ebunga]

Brilliant. I'll pay our IT consultants overtime to hop on a conference call with you and your team.

Re: Ok, but here me out, from a marketing standpoi

[Fortnite_Beast]

Look, I love Royal Farms as nuch as the next guy, but I don't want them tracking me while I pump gas and buy fried chicken.