It looks like Apple’s take on “modular” is not going to be anything like what the people here are expecting. Instead of replaceable parts, it sounds more like a retread of the Mac Pro where you add extra functionality via thunderbolt functionality.
Honestly, I think the whole point of the article is that we have no idea what "it looks like".
I have no idea what they'll do, but there are a few things we know: 1) they want to make something modular, 2) they're consulting with working professionals to find out where the pain points are, 3) they're planning to release something in 2019. Frustrating as it may be to know it's another year wait, it's better than what we knew before.
Also, I see a fundamental inconsistency in a lot of people's complaints here. First of all, people keep bringing out the old "Apple doesn't care about the Pro market, because it's too small" argument, but if that were the case why would they break with decades of customary silence and start telling us what they're up to? Why would they spend 2 years developing a new Mac Pro? And why would they make a point of assuring us of the fact that they are actually looking at professional workflows and identifying pain points? That's a lot of hand-holding for a market they don't care about. As to the hilariously paranoid suggestions that they're lying to us—well, again, why would they bother?
Secondly, we have to consider that the benefits of modularity and economy don't only benefit
us. If Apple can make a machine that's easier for
them to keep up-to-date, to repair, etc., then that helps their bottom-line as well. Since the rate of pro product updates has been a huge pain point, my guess is that they want to build a basis for a machine that enables them to plan a roadmap and stick to it. Could this just be a new basic tower? Sure. But it could also be something more friendly for Pro users, like myself, who will always do a good chunk of their work on a MacBook Pro. (As to anyone who suggests that "Pro" users don't work on laptops? Well, you're just wrong. Wake the **** up.) An external GPU, for example, might rid me of the irritation of having to disconnect/reconnect my display to/from my Ubuntu box. If the new Mac Pro system used an external GPU, I could probably just plug in a single TB3/4 connection and be ready to go. Or maybe they design some method for clustering, so that Mac Pro + MacBook Pro become a single, multi-core machine? A pipe-dream, maybe, but possible.
Thirdly, Apple has taken a ton of heat—particularly since releasing the latest MacBook Pro—for soldering in upgradable components like SSD and RAM. I would be
very surprised if they took that route with the Mac Pro (after all, they didn't even do that with the trashcan). Considering the rate of industry development, upgradeable GPUs would be very smart, as well. Will they enable this? I really don't know. But it would make sense, particularly given that GPUs are regularly being used for more than just gaming these days (e.g., machine learning/high-performance computing).
One last thing, which was almost hidden in that article, is that it sounds like they're looking at bottlenecks
right down to system level. I find this part very encouraging, because it suggests that they've realized that
macOS design and performance is part of the equation, as well. Also, there's virtually no improvement they could make through such an understanding that wouldn't vicariously benefit consumer's in the process, making any improvements proposed a pro/consumer win/win—i.e., no excuse to avoid implementation.