What are some roadblocks before Apple may play a role in the server market?

Sorry if this is not the place to post such a question; if there’s other forums better suited for this kind of questions, I’m more than happy to post there.

Apple may or may not want this, but let’s for sake of the argument say they might want to enter the most lucrative part of the x86-64 ecosystem, the servers. With the M4 chip reviews out, it’s crazy that it might just accomplish traditional integer tasks e.g. code compilation just as fast as those with similar core counts with like 1/10 of total energy consumption. Deep integration with its other peripheral and OS itself and being an SoC rather than just a CPU definitely plays a role here, but there must be more to it.

Yet there’s few if any rumors Apple were to release its own counterpart to the EPYC/XEON class platforms. I mean Nvidia is doing this with their Grace CPU, boasting the company’s well-known hyper fast interconnection between chips and of course most importantly the CUDA cores.

What are some major roadblocks out there? Maybe in fact so much of the energy efficiency and performance advantages boils down to being an SoC its memory access is both low latency and low power consumption, and a server with huge memory requirements means either Apple has to deal with heterogeneous memory (SoC + lots of generic DDR5 DIMM) or just give up on the SoC idea, that Apple would be holding little advantage against x86-64 in the server space? Or maybe the scaling of peripherals isn’t a trivial task, let alone whole system verification and reliability concerns? The software ecosystem relies heavily on x86-64 and not every enterprise wants to move over to ARM? Or it’s simply because ARM might charge more than Apple want to pay if Apple were to release its own server line of platforms?

I’m EE background but more of a software guy, so I’m not even sure what are the major roadblocks before one company might be selling server platforms, but based solely on the figures we have now, I’d imagine it’ll be a really interesting alternative if Apple were to release its own EPYC/XEON counterpart.

This is purely hypothetical, because if Apple wanted that, they would have by now. :roll_eyes:

Apple isn’t interested in the server market b/c they can’t deploy their usual strategy, which is locking in people into their eco-system and make them invest soo much they can never leave w/o significant financial and/or data losses.

Don’t bother coming up with academic roadblocks for deployment of Apple servers, it’s not gonna happen. Period. :man_mage:

5 Likes