They really only have to be competitive with the lower core consumer chips to accomplish their goal.

I think they should try for the performance crown though and the server business. There's a lot of money in that. In servers power consumption is an important consideration and they have an advantage there as well. ARM based architectures are also a lot more relaxed in that they don't have to present a nearly sequentially consistent view of cache and memory between multiple cores like x86. That may give them an edge scaling the core count. It's horrible to program for if you are coding lock free algorithms, but almost nobody does that and fewer should.

There's a reason Apple don't make Time Capsules any more. Servers are usually commodities, and Apple don't make commodities if they can reasonably avoid it (anything they make that's even close to a commodity still, at least, can be turned into a fashion accessory, like the aluminum keyboards and stuff). Commodity suppliers get put into bidding wars, and operate on razor-thin margins.

Apple have basically opened the market for ARM servers by producing a workstation to go with them. Amazon is probably feeling pretty vindicated for Gravitron right now. Everybody else will catch up with them over the next five years or so, I'm sure.

[1] Excepting those big enough to build their own

I don’t think Apple will do it considering how far behind they are, but I wouldn’t be surprised if they start an AWS competitor using Apple Silicon.

Cloud computing is never 100% a commodity because there’s always business logic to sell.

Sure, S3 and EC2 are by this point commodities, but cloud products like Babelfish for Aurora PostgreSQL are not.

Businesses are plenty interested in single source cloud computing, there are tons of shops that basically can’t leave AWS without a monumental effort.

It's a high margin business, the opposite of a commodity.

Throwing in to that market with an ARM processor would limit you to portable software, putting you into competition with not only Intel and AMD but also Amazon and potentially Qualcomm and others. Which could actually hurt Apple if it gives ARM competitors a wider market for their processors and therefore more resources to spend on development that carries down to PCs and mobiles.

ARM chips are also commodities nowadays. I lost count of how many companies make them. Yet Apple chose to make their own.

Charging circuitry chips are also commodities and Apple bought the division of Dialog semiconductors making such chips to have in house knowledge.

Now Apple bought the modem division of Intel so we're gonna see GSM modems from them in the future.

Not in any non-Apple device, that is. The rest of the world will be stuck with with Qualcomm, Broadcom and some crap from Mediatek.

Also, they can sell you a monthly subscription to iCloud storage instead. I'd rather be in that line of business than selling an overpriced and under-featured NAS. Anyone who cares that much is probably technically savvy enough to prefer a proper NAS anyway.

Their OS is pretty unsuitable for server tasks[1], and they've abandoned most of their server tools, so I think they don't care about that market anymore.

[1] No synflood mitigations, easy to panic by spawning threads that sleep.

Replacing the Intel machines with something that runs (a) faster and (b) with less power, is a double-win. I'd frankly be amazed if there wasn't a N-core or N-module server setup somewhere within the spaceship undergoing tests. And it's not that far from there to selling a Mac-based server. Which they've done before, after all...

First of all, that's ludicrously expensive. You need an active interposer that's larger than the chip to do something like that. 4x HBM stacks means 4096 wires to run between your compute-device and the RAM, that's a lot of wires in a tiny area, and only possible with advanced packaging (expensive) methodologies.

Second: HBM2 based designs have poor memory-capacity: 8GB per stack means 4x HBM2 caps out at 32GBs under most reasonable configurations.

Going into 2021 with a ludicrously expensive 32GB machine seems a bit weak. I don't think anyone in the high-end market is targeting that level of memory capacity. Even if we're generous and HBM capacity doubles between now and then, I'm not sure if 64GBs is enough for workstation tasks.

HBM works for supercomputers, because when 32GB isn't enough, you just buy another node. A64FX, the current CPU-based supercomputer that uses HBM, only has 32GB/node. Then Fugaku bought 150,000 nodes.

So you have a high-bandwidth, but no better latency, L4 chip. And that's kind of hard to deal with and optimized against.

Xeon Phi had a DDR4 + HMC (where HMC was a early competitor to HBM), and getting good performance out of that was unusual. DDR4 ended up having slightly better latency, so the HMC cache would hurt some workloads.

I think Apple would want to release a 64GB or 128GB Mac Pro... but only to upsell its customers to a 256GB or higher Mac Pro. The ability to configure more RAM into a system is key for marketing (even if this RAM is soldered onto the motherboard: like LPDDR4x, it wouldn't be too expensive to make different motherboards especially in a niche $10,000 computer like the Mac Pro).

But if Apple did go with HBM2 itself... well... then its physically unable to go above 32GB (today), or maybe ~64GB (if bigger HBM2 modules come out next year).

--------

Sticking with LPDDR4x just seems like the most obvious, and best, decision for Apple. So that's my expectation. But then again: making a 8-way decoder + fat-as-heck execution pipeline wasn't part of my expectation either. So maybe Apple will surprise us again.

Or, in Elon Musk speak, a "hardcore smackdown", with x86 playing the role of the gas engine.