Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Stargate Project: SoftBank, OpenAI, Oracle, MGX to build data centers

apnews.com

Post reply on HN
Print view View original

551–560 of 1001 posts

Re: Stargate Project: SoftBank, OpenAI, Oracle, MGX to build data centers

#551

Post by pixelmonkey »

Here is what I think is going on in this announcement. Take the 4 major commodity cloud companies (Google, Microsoft, Amazon, Oracle) and determine: do they have big data centers and do they have their own AI product organization?

- Google has a massive data center division (Google Cloud / GCP) and a massive AI product division (Deep Mind / Gemini).

- Microsoft has a massive data center division (Azure) but no significant AI product division; for the most part, they build their "Copilot" functionality atop their partner version of the OpenAI APIs.

- Amazon has a massive data center division (Amazon Web Services / AWS) but no significant AI product division; for the most part, they are hedging their bets here with an investment in Anthropic and support for running models inside AWS (e.g. Bedrock).

- Oracle has a massive data center division (Oracle Cloud / OCI) but no significant AI product division.

Now look at OpenAI by comparison. OpenAI has no data center division, as the whole company is basically the AI product division and related R&D. But, at the moment, their data centers come exclusively from their partnership with Microsoft.

This announcement is OpenAI succeeding in a multi-party negotiation with Microsoft, Oracle, and the new administration of the US Gov't. Oracle will build the new data centers, which it knows how to do. OpenAI will use the compute in these new data centers, which it knows how to do. Microsoft granted OpenAI an exception to their exclusive cloud compute licensing arrangement, due to this special circumstance. Masa helps raise the money for the joint venture, which he knows how to do. US Gov't puts its seal on it to make it a more valuable joint venture and to clear regulatory roadblocks for big parallel data center build-outs. The current administration gets to take credit as "doing something in the AI space," while also framing it in national industrial policy terms ("data centers built in the USA").

The clear winner in all of this is OpenAI, which has politically and economically navigated its way to a multi-cloud arrangement, while still outsourcing physical data center management to Microsoft and Oracle. Probably their deal with Oracle will end up looking like their deal with Microsoft, where the trade is compute capacity for API credits that Oracle can use in its higher level database products.

OpenAI probably only needs two well-capitalized hardware providers competing for their CPU+GPU business in order to have a "good enough" commodity market to carry them to the next level of scaling, and now they have it.

Google increasingly has a strategic reason not to sell OpenAI any of its cloud compute, and Amazon could be headed in that direction too. So this was more strategically (and existentially) important to OpenAI than one might have imagined.

Re: Stargate Project: SoftBank, OpenAI, Oracle, MGX to build data centers

#552

Post by alpb »

pinot wrote: post #504
wujerry2000 wrote: post #492

For fun, I calculated how this stacks up against other humanity-scale mega projects. Mega Project Rankings (USD Inflation Adjusted) The New Deal: $1T, Interstate Highway System: $618B, OpenAI Stargate: $500B, The Apollo Project: $278B, International Space Station: $180B, South-North Water Transfer: $106B, The Channel Tunnel: $31B, Manhattan Project: $30B Insane Stuff.

Those are all public projects except for one..

Yeah, I'm not sure why we're pretending this will benefit the public. The only benefit is that it will create employment, and datacenter jobs are among the lowest paid tech workers in the industry.

Re: Stargate Project: SoftBank, OpenAI, Oracle, MGX to build data centers

#554

Post by achierius »

cpursley wrote: post #378
DoubleGlazing wrote: post #24

That's a ridiculous sum of money that could be better spent on much more worthy things.

So was getting a man to the moon. Do you want to lose the AI race to the Chinese?

Why would I care? Do you really want Masayoshi Son in charge of a theoretical superhuman AI?

Re: Stargate Project: SoftBank, OpenAI, Oracle, MGX to build data centers

#555

Post by ripped_britches »

throw310822 wrote: post #313
goatlover wrote: post #216

Earlier quoted context omitted.

Weird definition of close you have there. If I asked for $700, and you gave me $50, would that be close?

Depends. If I fart in a glass jar and then I try to sell it to you for $700, but you end up buying it for $50, I'd say it's pretty close.

This is my signal that it’s time to put up HN and go to bed for the night

Re: Stargate Project: SoftBank, OpenAI, Oracle, MGX to build data centers

#557

Post by Dalewyn »

krick wrote: post #544
wujerry2000 wrote: post #492

For fun, I calculated how this stacks up against other humanity-scale mega projects. Mega Project Rankings (USD Inflation Adjusted) The New Deal: $1T, Interstate Highway System: $618B, OpenAI Stargate: $500B, The Apollo Project: $278B, International Space Station: $180B, South-North Water Transfer: $106B, The Channel Tunnel: $31B, Manhattan Project: $30B Insane Stuff.

It's unfair, because we are talking in the hindsight about everything but Project Stargate, and it's also just your list (and I don't know what others could add to it) but it got me thinking. Manhattan Project goal is to make a powerful bomb. Apollo is to get to the Moon before soviets do (so, because of hubris, but still there is a concrete goal). South-North Water Transfer is pretty much terraforming, and others ar…

>What is the goal?

Be the definitive first past the post in the budding "AI" industry.

Why? He who wins first writes the rules.

For an obvious example: The aviation industry uses feets and knots instead of metres because the US invented and commercialized aviation.

Another obvious example: Computers all speak ASCII (read: English) and even Unicode is based on ASCII because the US and UK commercialized computers.

If you want to write the rules you must win first, it is an absolute requirement. Runner-ups and below only get to obey the rules.

Re: Stargate Project: SoftBank, OpenAI, Oracle, MGX to build data centers

#559

Post by rhubarbtree »

Davidzheng wrote: post #373
TheAceOfHearts wrote: post #300

I'm confused and a bit disturbed; honestly having a very difficult time internalizing and processing this information. This announcement is making me wonder if I'm poorly calibrated on the current progress of AI development and the potential path forward. Is the key idea here that current AI development has figured out enough to brute force a path towards AGI? Or I guess the alternative is that they expect to figure…

Let me avoid the use of the word AGI here because the term is a little too loaded for me these days. 1) reasoning capabilities in latest models are rapidly approaching superhuman levels and continue to scale with compute. 2) intelligence at a certain level is easier to achieve algorithmically when the hardware improves. There's also a larger path to intelligence and often simpler mechanisms 3) most current generation…

> 1) reasoning capabilities in latest models are rapidly approaching superhuman levels and continue to scale with compute.

What would you say is the strongest evidence for this statement?

Re: Stargate Project: SoftBank, OpenAI, Oracle, MGX to build data centers

#560

Post by thepace »

heydenberk wrote: post #29

~$125B per year would be 2-3% of all domestic investment. It's similar in scale to the GDP of a small middle income country. If the electric grid — particularly the interconnection queue — is already the bottleneck to data center deployment, is something on this scale even close to possible? If it's a rationalized policy framework (big if!), I would guess there's some major permitting reform announcement coming soon.

It is not the just queue that is the bottleneck. If the new power plants designed specifically for powering these new AI data centers are connected to the existing electric grid, the energy prices for regular customers will also get affected - most likely in an upwardly fashion. That means, the cost of the transmission upgrades required by these new datacenters will be socialized which is a big problem. There does not seem to be a solution in sight for this challenge.
Post reply on HN