Skip to content

Comment on Read this but think of Mac GPU universal memory running LLMparent

Comments

Just who is going to be the os or llm provider(s) in the new era I wonder.

We've already seen that question answered. TSMC has two customers for cutting edge nodes; Apple and Nvidia. Apple refuses to address the datacenter market, so they're missing out on the trillions of dollars to be made in high-margin compute.

Keep in mind, Apple could just drop their pride and support Nvidia drivers again. macOS can run CUDA if Apple just signs the installer, they could be selling rackmount Macs as high-performance ARM clusters. But nope! Apple insists on losing.

AboutSource Built by g1lg1l

Hackerly is an independent reader for Hacker News, built on the public HN API. Not affiliated with Y Combinator.