OpenAI Has "Run Out of GPUs" - Sam Altman Mentions Incoming Delivery of "Tens of Thousands"

[T0@st]

Yesterday, OpenAI introduced its "strongest" GPT-4.5 model. A research preview build is only available to paying customers—Pro-tier subscribers fork out $200 a month for early access privileges. The non-profit organization's CEO shared an update via social media post; complete with a "necessary" hyping up of version 4.5: "it is the first model that feels like talking to a thoughtful person to me. I have had several moments where I've sat back in my chair and been astonished at getting actual good advice from an AI." There are apparent performance caveats—Sam Altman proceeded to add a short addendum: "this isn't a reasoning model and won't crush benchmarks. It's a different kind of intelligence, and there's a magic to it (that) I haven't felt before. Really excited for people to try it!" OpenAI had plans to make GPT-4.5 available to its audience of "Plus" subscribers, but major hardware shortages have delayed a roll-out to the $20 per month tier.

Altman disclosed his personal disappointment: "bad news: it is a giant, expensive model. We really wanted to launch it to Plus and Pro (customers) at the same time, but we've been growing a lot and are out of GPUs. We will add tens of thousands of GPUs next week, and roll it out to the plus tier then...Hundreds of thousands coming soon, and I'm pretty sure y'all will use every one we can rack up." Insiders believe that OpenAI is finalizing a proprietary AI-crunching solution, but a rumored mass production phase is not expected to kick-off until 2026. In the meantime, Altman & Co. are still reliant on NVIDIA for new shipments of AI GPUs. Despite being a very important customer, OpenAI is reportedly not satisfied about the "slow" flow of Team Green's latest DGX B200 and DGX H200 platforms into server facilities. Several big players are developing in-house designs, in an attempt to ween themselves off prevalent NVIDIA technologies.



View at TechPowerUp Main Site | Source

 

[Jtuck9]

I don't think Gary Marcus is particularly impressed

Hot take: GPT 4.5 is a nothing burger

Pure scaling in shambles

garymarcus.substack.com

 

[Veseleil]

Cool story bro.

 

[ZoneDymo]

gotta keep people believing otherwise it falls apart too soon.

 

[Vayra86]

Sam Altman announces new ICO... erm I mean... yeah

 

[MentalAcetylide]

My only question is what in the hell are they doing with all of this AI stuff? I would say concocting novel ways of bending people over while making them smile & milking them for as much money as possible, but I don't think they need AI for that given the general lack of prudence among the masses.

 

[mb194dc]

Imagine how much money they're burning with all the pointless hardware purchases. They'll be lucky to keep going another 12 months.

 

[Tartaros]

Then next week another chinese comes up with an AI that runs on a MSX and Jensen strangulates Sam with his leather jacket.

 

[Steevo]

Just needs more money, more time, they swear they will achieve their goal...... maybe they and Fusion will be ready in 50 years?

 

[trsttte]

Imagine how much money they're burning with all the pointless hardware purchases. They'll be lucky to keep going another 12 months.

As long as M$ is willing to keep paying the bills they can last a long long time. That won't be forever though... Can't wait for the headlines trashing them in a couple weeks when Deepseek or any other upstart launches a model that's able to achieve 70, 80 or even 90% of what this does for a fraction of the cost

 

[john_]

In the meantime, Altman & Co. are still reliant on NVIDIA for new shipments of AI GPUs. Despite being a very important customer, OpenAI is reportedly not satisfied about the "slow" flow of Team Green's latest DGX B200 and DGX H200 platforms into server facilities.

There is AMD out there, but I guess they don't have the software skills to incorporate AMD GPUs in their systems.

 

[freeagent]

There is AMD out there, but I guess they don't have the software skills to incorporate AMD GPUs in their systems.

You sure about that?

Maybe AMD doesn't have the skills to incorporate the hardware needed.. just throwing that out there.

 

[lexluthermiester]

OpenAI Has "Run Out of GPUs"

So sad... I can hear tiny violins playing...

 

[TheinsanegamerN]

There is AMD out there, but I guess they don't have the software skills to incorporate AMD GPUs in their systems.

You sure about that?

Maybe AMD doesn't have the skills to incorporate the hardware needed.. just throwing that out there.

It could also be both. Grifters are, by nature, lazy people, so they'll gravitate to nVidia since CUDA is already well documented and tools already exist. Deepwhale or whatever it was called showed AMD hardware can be very competitive if you build from the ground up. Of course, because only the most driven would ever do that, not having tools ready for clients puts you at a massive disadvantage.

 

[freeagent]

CUDA is almost 20 years old. Quite the head start.

 

[Darmok N Jalad]

"Sorry, Sam, we seem to have misplaced some ROPs somewhere. Once we find them, we'll be in touch."

 

[john_]

You sure about that?

Maybe AMD doesn't have the skills to incorporate the hardware needed.. just throwing that out there.

I see them begging Nvidia for GPUs, but probably Nvidia has given priority to Musk, because, well, he is in the government, and Microsoft and Google and Meta. Add Deepseek in the mixture and OpenAI from leader in AI could drop many places.
Also OpenAI thinking of building it's own chips and Altman being in bad relations with Musk, who let me repeat is in the government, well OpenAI probably gets the last place in the priority list of Nvidia's top customers. They either differenciate themselves, or keep begging.

 

[MentalAcetylide]

Then next week another chinese comes up with an AI that runs on a MSX and Jensen strangulates Sam with his leather jacket.

The Bard's Tale III has a weapon called Kali's Garrote; Nvidia has Huang's Garrote, equip to screw customers & multiply quarterly earnings.

 

[Rover4444]

They know they can use Radeon Instincts instead, right? LLMs don't need the image generation power of NVIDIA hardware, they could just put all the stuff doing inference on Instincts and reserve the NVIDIA hardware for training. Unless they need more density, then like, what? Are you training a 1T+ model or something?

OpenAI Has "Run Out of GPUs"

... And that's a GOOD thing!

 

[DaemonForce]

"Sorry, Sam, we seem to have misplaced some ROPs somewhere. Once we find them, we'll be in touch."

OMFG what a nightmare. If this ROP issue is way bigger than being communicated, that's a LOT of defects.
Around every two dozen cards missing the 8 ROPs or whatever is a whole ass other flagship card missing.
That's actually expensive.