I would be so interested to learn why at all their Mac Pro whole thing went so lame. The idea was cool to me (cool as in general sense, I certainly don't need that firepower), but it always felt like it was underfunded or undermanaged, updated rarely and finally died probably for lack of ideas what to do else.
This is the viewpoint of wanting to champion the classic 'box with slots' product. The bigger picture there is larger demographic impacts/trends than that changed and Apple didn't follow a 'stuck in time bubble' strategy.
So what users were doing/behaving had as much impact on Apple as Apple had impact on the Mac Pro.
Changing Demographics.
i. The percentage of laptops sold from the era PowerMacG4/G5 and MacPro 2016 to now have changed substantively.
Back when Apple was handing out numbers. (portables is blue and desktops is grey )
Everyone knows the PC market has been in decline for the past few years. But one segment of that market is doing spectacularly well, and one company has managed to carve out enviable sales and profits by dominating that niche. Guess who?
www.zdnet.com
Since Apple has dropped explicit statistics it has only gotten worse. And it isn't an 'Apple only' thing either. Same basic general shift in overall PC market windows/macOS/Linux all primarily showing the same shift.
To a large extent, it s linked to the comment you made , 'I don't certainly need that firepower'. Back when laptops were substantively weaker larger number of users for 'forced'/'herded' into buying a desktop because that was the only performance option that met their workload demands. Even in Apple Desktops as the iMac shifted from 'mobile CPU/GPU" to Desktop CPU/GPU it also took more share from other desktops. ( Mini was kneecapped due to self imposed constraints and the Mac Pro price crept higher. Along with Intel/AMD CPU/GPU costs. )
Recently, Apple rolling out the Mac Neo while at same time dropping the Mac Pro was not going to lead to a net drop in unit sales at all.
I think it was around 2017 where Apple said that Mac Pro was somewhere around 'single digit' of mac sales. Some folks hand-waved and said this might it was around 6-8% when far likely this was hovering more close to 1%. Once not even a single digit percentage are they are even going to do a product? Apple does not have an objective to make everything for everybody.
Apple shifting to more of a laptop focus is something that the market is doing at least as much as a 'directive' being pushed by Apple onto users. Apple main focus over last decade has been putting 'workstation' like performance into a 14/16" laptop form factor for the MBP. Some of that was putting extremely capable focused fixed function units into the SoC. Mac Pro 2019 Afterburner card to run ProRes RAW.... 3 years later iPad Pro does even more ProRes RAW at a far lower cost. Don't always need 'brute force' power to get things done.
ii. So the pool of Mac Pro users is relatively smaller. That is made even worse by substantive fractions of that pool upgrading on much longer cycles. Some numbers to put into perspective, so there is a user base of around 800,000 users total. If vast majoirty only upgrade every 10 years then
800,000 / 10 --> 80,000 upgraders a year.
800,000 / 9 --> ~88K
800K / 8 ---> 100K
800K / 7 ---> ~114K
800K / 6 ---> ~133k
800K / 5 ---> 160K
800K / 4 ---> 200K
Note how the Vision Pro was a huge 'fail' in most of these forums and tech press analysis at a run rate of about 300K/year. A number higher than any of these.
If upgrade buys are slower , then more than a good chance that Apple is going to upgrade slower. If sell every two years then on charge above fro 5-7 range get about 220-300K pool to sell into if that is dominate range.
Is Apple going to do massive draws from the Windows side in this space? Probably not. Folks who primarily want the an inexpensive containers to put the most commodity priced parts into are not coming to Apple in droves on the laptop side either.
Workstations in the power Xeon-W class space shrank on the windows side also. Apple eschewed the xMac for various reasons, but post the Intel-CPU transition, the Mac Pro never was suppose to be the old sub $2K PowerMac .
Apple stopped reported unit sales for Mac in 2018 but last years (pulled from Google AI overview but look about right).
"...
Final Years of Apple-Reported Mac Unit Sales
• Fiscal Year 2018: 18.2 million units
- Fiscal Year 2017: 19.27 million units
- Fiscal Year 2016: 18.52 million units
- Fiscal Year 2015: 20.57 million units (the peak year for Mac unit sales)
- Fiscal Year 2014: 18.92 million units
- Fiscal Year 2013: 16.33 million units
..."
Even if take the 800K pool and compare to those yearly rates, it is mainly a smaller and smaller percentage. Estimates now are that Apple is in the 20-30M run rate per year range. Apple Silicon transition only pushed the unit numbers higher still.
Even going back a few more years.
The PC market has declined about 10% this year, yet Apple sold more Macs than ever in its 4th fiscal quarter (July-Sept. 2015). iPhone sales didn’t set a record – that usually happens in the holiday quarter – but they are up over the same quarter in 2014. Pity the poor iPad, which continues to […]
lowendmac.com
The 'box with slots' were mainly a dominate category for Mac back when Apple relatively didn't sell that many Macs. They were a 'big fish' in 'small pond'. The general Mac trendline though is methodical forward progress. The hype that Apple had abandoned/under-invested the Mac really didn't match what the results were. The Mac never showed the growth rates of the others , but Apple didn't really abandon it. ( The Mac is selling into a vastly more mature general PC market. Just not going to see hypergrowth at that maturity state. ). In the bigger pond, the Mac Pro wasn't a 'big fish'. The 'big' was more so an illusion of the small pond than of being actually 'big'.
Apple is also not afraid in the graph above for iPhones to kill off iPods. ( and iPads have corrected the slide at this point. It isn't a declining category. No the most expensive iPad Pro didn't form the core of that turn-around. ). Apple also stopped the Mini from going after taking share away from the remaining iMac.
After the Mac Pro 2013 growth went up. If the Mac Pro was a super critical factor to overall Mac unit sales growth ...how could that happen? It is not. Apple focused on the systems that were critical to drive growth and did a decent job.
3. Even during the 2006=2012 the market dynamics of the GPU cards was substantively broken. Apple would commission their own standard config cards and a few other optional config cards. Then more than a few folks would wait for the hacks to copy the bootrom and drivers over to off the shelf windows card so as to not pay for the additional work. ( at the low levels macOS is not Windows. more work is required to make it work well on macOS).
Apple tolerated hackintosh but doesn't particularly enhance health of Mac Pro product if again folks just by pass paying for the addition work to do a Mac Pro in soft/firmware.
There is a group of folks in the Mac Pro user pool that think GPU drivers will just fall out of the sky because the Mac Pro exists. When Apple move the entire rest of the Mac product line off of 3rd party GPUs, that has impact on Mac Pro. The volume of those other units were a major reason the GPU drivers got done for each generation class.
AMD and Nvidia out to make the higher end GPUs as hot as possible didn't really help either. By 2015-2017 Apple was doing very serious work on their own implement and really did not buy into the 'infinity and beyond' on power consumption as being necessary. Also didn't help that Intel and AMD bumbled a but in this era on implementations of CPU and GPU respectively. And 'not a good partner' behavior put Nvidia in the doghouse; so pragmatically not an option.
All the Mac product 2016 and after were mainly transition prep for the transition. Apple may not have know in 2014-15 when the transition would happen, but probably did know there would be a transition eventually.
But seriously, why? Was it not enough profitable to really keep up with it?
Probably more a mismatch in opportunity costs than profits.
a. Less money spent on optimizing Apple GPU drivers and code would mean higher risk for the who Mac (and most of the iPad ) ecosystem .
[ e.g., Intel's foray into higher end dGPU was limited by drivers that didn't make the iGPU to dGPU focus switch well and also covering all the legacy cruft that is the PC gaming market. Apple had a much simpler and more consistent target that spanned watch-phone-ipad-macs. ]
b. keep macOS on Intel around just for Mac Pro? Versus put that effort into a better macOS on the vast majority of systems in the ecosystem.
c. Apple's standard modus operandi for internally designed silicon is to put it into multiple products. Each product doesn't get an individual die . The same die is shared across multiple products. Apple concentrates on making fewer, but better than competition dies rather than making individual dies for 'everyone'.
By several accounts Apple attempted to make an "bigger than Ultra , Extreme" part , but even if came to market would have been too expensive to drive any 'volume' that might 'save' the Mac Pro. Some kind of 'vanity' product would be on very thin ice even if did make it out the door. [ The M5 Ultra design when revealed may show that not all of that work was a 'bust' . 4 'chiplets' but only getting an 'ultra' out of that. ]
[ for a long while AMD tried to match Intel toe-to-toe with largest product catalog possible. If dealing with non-monopoly user pools have to pick you battles. Do some products and let others go. Intel is in basically almost same boat now. There was no way Apple was going to have SoC catalog that was going to be as wide and broad as Intel after moving away. ]
Especially strange to see how they now do Ultra models of everything or planning to do this year. So the high power stuff IS the margin, obviously and interests them.
If there is enough LPDDR5 RAM to go around, then there will be a Mac Studio Ultra version. Apple doesn't need a Mac Pro to ship something with an 'Ultra' tag on it.
The MacBook Ulra maybe appearing because perhaps Mn Max chips won't appear every generation just like Mn Ultra chip didn't. Other vendors don't throw their largest chips into the trash can every 12 months and Apple isn't doing any special here to keep doing that long term either. The Mac Pro with a trailing edge M2 Ultra because might need to extend the life is a problem for the Mac Pro over the longer term.
The AirPod Ultra pretty likely is mainly just some cameras attached to the same fundamental AirPod (maybe the next gen foundation but the unit numbers would come in futures) . They are not decoupling from AirPod product.
The Xeon-Workstation chips were really side effects of Intel doing Xeon chips for larger servers. There were far more repurposed silicon than design for dies. So the server customers are paying for the bulk of the R&D. Decouple the server market and it isn't a viable a product. Decoupling from Intel meant decoupling from multiple factors. consumer CPU cores. iGPU cores , and server CPU cores (also several levels of dGPUs). Not going to get replacements for all of those with dramatically smaller unit numbers.
Can't make sense of it all, or may be it was the idea but just poor execution killed it.
Apple made some bad moves.
1. The Mac Pro 2013 was somewhat conceived as a OpenCL at about the point Apple pivoted to Metal.
It has other major design problems but that could have been a counterbalance to problems that fell through
the cracks.
2. Apple either should have done the iMac Pro ( and perhaps T2 chip) sooner ( 2016 ) or refactored the MP design in 2015-6 to fill the gap better. ( either strip out the 2nd GPU + non-triangular block in the smaller enclosure or backslide toward older Mac Pro form factor) The Rip van Winkle mode is at deep odds with their rigid no discussion of future products policy. The assumption built into the policy is that will "do things" on a somewhat frequent basis. 3 years is too long. (deliver future products more than just talk about them) I think at one point there was a blog that highlighted that the iMac Pro and Mac Pro was some product management so wouldn't get both at the same time.
The whole butterfly keyboard clean-up on 'aisle 12' was largely decoupled from the Mac Pro because but a distraction also They tried to fix-it. Still didn't work as well, but at least there was moment over a couple of years.
By April 2017 Apple had to partially bail on the 'no future product discussion' policy because was getting to ridiculous point on "days since refresh" time clocks (e.g., mac rumors Buyers Guid 'clock'. ). The 'good product management' would have meant never would have needed that crisis meeting.
3. Although Apple was very unlikely to grab a gallon of CUDA kool-aid and drink until intoxicated, Apple also took all the wrong lessons from CUDA has primarily a moat building mechanism as they applied it to Metal. Apple was unhappy with the Khronos Group ( OpenCL/GL/Vulkan ). They missed the 'add an AI accelerator' compute card wave. Intel was pushing OneAPI ( SyCl and some other frameworks). AMD had RocM (HIP, OpenCL, OpenMP ) . Apple deprecated the major open standards (OpenCL/OpenGL) and replaced with largely nothing as open. They made porting to Apple Silicon as expensive as possible for software that tried to be multiplatform at the more POSIX/UNIX level. In that context, once again Apple painted themselves into a corner with largely purely self inflicted constraints.
Similarly Apple was slow as molasses adopting RDMA frameworks with core OS support. All of that should have been in place before rolled out Mac Pro on Apple Silicon.
Apple did extremely little to growth in the PCI-e card ecosystem around macOS. Most thunderbolt Mac could support a Thunderbolt external enclosure ( there are solutions that Apple seems to indirectly leverage like xMac Mini/Studio from Sonnet Technologies. ). There is a variety of stuff but most all seems to be coasting pre MP 2019 deployments for volume. ( still a fair number of kext only drivers which is deprecated and probably on thin ice after macOS on Intel completely fades away. )
The Mac Pro 2023 launch and Apple mainly talked about the expensive (high sunk cost ) legacy cards you put into the machine. Almost nothing forward looking there at all. [ The Mac Pro 2019 has some backwards looking stuff. the R4 HDD focused subsystem. But there was far more forward looking than backward looking cards. ]
I don't think the Mac Pro product management should have been in charge of all PCI-e add-in cards development for the whole ecosystem. I suspect defacto it was there as a side-effect when Mac Pro had bigger footprint. But declined as MP 2013 came to live and sagged over time.
4. Thunderbolt kool-aid. April 2017 Apple explicity mentions that over reliance on Thunderbolt was a mistake. Less than 10 years later back to drinking same kool-aid was drinking when painted themselves into a corner in 2016-2017.
5. Splitting the Rack and Desktop case. Increasing the Mac Pro entry price by 100% was going to prune off a substantive fraction of the classic MacPro user base. So going to be lower volume product. Now make it an even
LOWER volume chassis by splitting the rack and tower versions. There could have been a central core that attached with rail or a stand too. Neither one of them would have been as individually 'pretty', but the Mac Pro didn't need excessively 'pretty'.
