On the other hand, Anthropic exists, so I am not sure.

I can imagine both Apple and Google happily positioning themselves in the same type of relationship as OpenAI-Microsoft.

Google had internal efforts, and Anthropic already. They definitely don’t have the compute to spare to split with another organization.

Apple surely doesn’t have a cluster that at all compares with the big cloud giants.

Oracle and AWS are really the only cloud left, and oracle is already renting to Microsoft for GPU compute.

>Apple surely doesn’t have a cluster that at all compares with the big cloud giants.

Apple has a lot of cash to throw at it. Question would be if Apple is even interested in it.

They should be. The slight improvements in messaging autocomplete in iOS 17 have made a noticeable difference in my texting. To have an iPhone that understands me, and a Siri that doesn’t say, “Here’s what I found on the web” is extremely valuable.

Idk. It looks like Apple is happy to sell you the hardware to do the image recognition etc on your own device and receive the result. They can claim privacy and save on computation.

Then they can charge.

The problem is that a significant amount of the already-made GPUs are in use. Even if they can afford to throw money at it, where is that money going to go?

The best they can do is out-bid their competitors, for the competitors hardware. I'm sure apple doesn't want to pay Google for GCP resource to train an AI. Again, there may not be enough companies renting out GPUs at all.

Apple is burning through more than a million USD per day for its research on Ajax and Co, they definitely have an interest in building big imo.

That's .1% of their annual expenditures