Comment by paxys
6 hours ago
"Largely" is doing a lot of heavy lifting here. Yes Google and Amazon are making their own GPU chips, but they are also buying as many Nvidia chips as they can get their hands on. As are Microsoft, Meta, xAI, Tesla, Oracle and everyone else.
But is Google buying those GPU chips for their own use, or to have them on their data centers for their cloud customers?
google buys nvidia GPUs for cloud, I don't think they use them much or at all internally. The TPUs are both used internally, and in cloud, and now it looks like they are delivering them to customers in their own data centers.
When I was there a few years ago, we only got CPUs and GPUs for training. TPUs were in too high of demand.
I can see them being used for training if they're vacant.
Both. Internal are customers too.
How about Apple? How is Apple training its next foundation models?
To use the parlance of this thread: "next" foundation models is doing a lot of heavy lifting here. Am I doing this right?
My point is, does Apple have any useful foundation models? Last I checked they made a deal with OpenAI, no wait, now with Google.
Apple does have their own small foundation models but it's not clear they require a lot of GPUs to train.
3 replies →
I think Apple is waiting for the bubble to deflate, then do something different. And they have the ready to use user base to provide what they can make money from.
5 replies →
Apple is sitting this whole thing out. Bizarre.
I mean, they tried. They just tried and failed. It may work out for them, though — two years ago it looked like lift-off was likely, or at least possible, so having a frontier model was existential. Today it looks like you might be able to save many billions by being a fast follower. I wouldn’t be surprised if the lift-off narrative comes back around though; we still have maybe a decade until we really understand the best business model for LLMs and their siblings.
1 reply →
This whole thread is about whether the most valuable startup of all time will be able to raise enough money to see the next calendar year.
It's definitely rational to decide to pay wholesale for LLMs given:
- consumer adoption is unclear. The "killer app" for OS integration has yet to ship by any vendor.
- owning SOTA foundation models can put you into a situation where you need to spend $100B with no clear return. This money gets spent up front regardless of how much value consumers derive from the product, or if they even use it at all. This is a lot of money!
- as apple has "missed" the last couple of years of the AI craze, there has been no meaningful ill effects to their business. Beyond the tech press, nobody cares yet.
From a technology standpoint I don’t feel Apple’s core competency is in AI model foundations
They might know something?
1 reply →
They are in housing their AI to sell it as a secure way to AI, which 100% puts them in the lead for the foreseeable future.