SoftBank and Open AI's $500 Billion AI Project Struggles To Get Off Ground (msn.com) 22
The $500 billion Stargate AI project announced by SoftBank and OpenAI at the White House six months ago has failed to complete a single data center deal and sharply scaled back its near-term plans. The venture, which originally pledged to invest $100 billion "immediately," now aims to build one small data center by year-end, likely in Ohio, according to WSJ. SoftBank and OpenAI have disagreed over crucial partnership terms, including site locations.
OpenAI has proceeded independently, signing a deal with Oracle worth more than $30 billion annually starting within three years. That agreement totals 4.5 gigawatts of capacity and would consume power equivalent to more than two Hoover Dams. Combined with a smaller CoreWeave deal, OpenAI has secured nearly as much data center capacity as Stargate promised for this year. SoftBank invested $30 billion in OpenAI earlier this year as part of the infrastructure partnership plans.
OpenAI has proceeded independently, signing a deal with Oracle worth more than $30 billion annually starting within three years. That agreement totals 4.5 gigawatts of capacity and would consume power equivalent to more than two Hoover Dams. Combined with a smaller CoreWeave deal, OpenAI has secured nearly as much data center capacity as Stargate promised for this year. SoftBank invested $30 billion in OpenAI earlier this year as part of the infrastructure partnership plans.
Oh dear (Score:1)
What a shame
Never mind
Is it just me (Score:3, Insightful)
Or do the "AI Failure" stories get more frequent? Well, that is how it goes until a critical mass is reached. And then the house of cards collapses and we get to see how little LLMs can actually do when they need to compete on merit instead on hype.
Kind of boring when you see this happening for the 3rd time or so.
Re: (Score:2)
I'm suspecting this is where we are headed but we don't have the shiny new technology fad to move onto yet. (I'm just waiting for someone to mash up AI and blockchain, imagine how much hype you could muster up over that)
Re: (Score:3)
LLMs can do a lot. The problem is that people try to use them into things that they can't do, instead of using them into things they can do. For example I recently had a 200 page document from which I wanted quickly to get some info, so I just gave link to the document and asked AI and got my answer.
Open AI's problem is that they are fixated on making LLM some kind of AGI that can solve any problem and they simply don't have the required skills for that.
Google Deepmind on the other hand developes AI like it
Re: (Score:2)
I just do not see it. So fare the only somewhat dependable useful LLM applications are "better search" and "better crap" and they could do that 3 years ago. So, yes, searching though large documents that are not very complex is one thing LLMs can do well. You know, "better search". I also talked to somebody recently that uses LLMs to but the finishing touches on formalized funding applications for research projects. That works too and falls under "better crap". (Yes, I have some small experience with writin
Re: (Score:2)
Re: (Score:2)
Or they will have a whole heap of partly built power infrastructure that is just abandoned. Takes years to get that stuff built. Will the bubble burst before they are completed?
Re:Is it just me (Score:4, Insightful)
Yeah, everybody knows that the number of reports on something in popular media is an extremely accurate measure of the underlying statistics. That's also how you reliably determine whether crime is up or down!
This story isn't even about "AI failure"; it is about announcing clearly unrealistic numbers and investments and then meeting reality when actually having to execute.
Meanwhile, AI is still improving, still gaining users, still gaining use cases, still rapidly and significantly changing everything around us.
A bunch of rural communities (Score:2)
I mean sure the data center could recycle that water but that costs money.
They're also incredibly noisy. Like if you live year one there is this constant horrible droning going on.
And of course you're competing for electricity so your power bill goes up. Unlike water so far it's kept up and I'd haven't heard of anyone's lights being shut off but the
Re: (Score:3)
The rich didn't have consumers to worry about during most of human history - like consider the middle ages. The rich owned the land and the people payed them for the privilege of farming it. A special class of artisans served the rich and didn't do much with the peasants. The peasants only existed to labor on the land for the benefit of the rich because farming used to be a lot of work.
Sometimes there wasn't enough food and the peasants rose up to overthrow the rich. How many times did they succeed? 0.
Re: (Score:1)
Re: A bunch of rural communities (Score:2)
Re: (Score:2)
Was that during the middle ages?
Re: (Score:2)
If you make more than $23,380 a year, you belong to the top 50%. If you make more than $122,100 you are in the top 10%. Just in case you are wondering if you are rich or not.
But the reason why people misunderstand the rich and poor is because they think it is money that makes people rich. Money is just a tool which has no value in itself. What makes people rich are the goods they own and the goods they can produce. The rich can own automated factories and trade goods with other rich people and live happily
not gonna happen (Score:2)
few things will not happen harder than this boondoggle is not going to happen
I'll be surprised if even one center ever comes online
Softbank is investing in the wrong things (Score:2)
In a gold rush, do you mine for gold or do you sell the picks? Take the safe route and sell the picks.
4.5 gigawatts (Score:1)
Re: (Score:1)
Re: (Score:1)
It takes time to assemble a 1000 acre computer. (Score:2)
I assure you, this is very real. I am literally in the middle of it.