The current Apple Silicon Mac Pro is the most phoned in product they’ve ever made. A faction in Apple knows it too. I suspect there’s a group in Apple that was heavily involved in making the 2019 Mac Pro good, and there’s another faction that made the Mac Studio good (who previously failed with the Trash Can Mac Pro), that’s the group that wants to make the whole widget and not let anyone tinker around inside at all. That group probably pushed for soldered ram before Apple silicon made it necessary, soldered SSD’s and all that. And there’s legitimacy to both points of view, soldered is reliable. However I’m not buying a Mac Pro from the soldered RAM faction, and yet that’s what they made.
The 2019 Mac Pro also became pointless since there’s no GPU upgrade possibility because Apple was involved in making GPU’s work on their platform and that development will now stop.
Upgradability means good value for the customer. For Apple that means lower profit and higher support costs. It makes sense they don’t want to sell that many Mac Pros if they’re low volume and support costs are higher, and that’s probably factored into the price. But that also sucks, and I don’t want it.
I was ready to buy the new Mac Pro until it came out and was underwhelming. Now that budget is being put to other things.
I hope the next version is actually impressive. That they come out with upgrade cards as was rumored. I hope they put the equivalent of an iPhone on a pci card and have that act as a 1st party replacement for egpus and GPU’s. And there’s enough PCI lanes and all that. I would wager the strikes in Hollywood are also hurting their sales. I want the Mac Pro to live and be amazing. So try again.
Only Apple has the customer data to tell them what's "worth it" or not in terms of R&D and production investment, but with the high-end Mac Studio now taking at least some of what was already a very limited market for the Mac Pro, I'm not surprised to read that its future is on thin ice. It would probably take a big investment of both money and human talent to create a next gen Mac Pro that is significantly faster and more expandable than Mac Studio and I'm not sure the market for such a machine is big enough to justify that investment for a sprawling, $3 trillion company.
This is pretty close to how I feel.
In this case, it doesn't take a genius, a leaky insider, or Gurman's opinion to question the value proposition that motivated Apple to ship the Apple Silicon Mac Pro. Perhaps if Apple had held back a little on the Mac Studio and made it more of a juiced-up Mac mini there would be a little less head scratching. But the Mac Studio Ultra versions seem to have stolen much of the thunder that Apple may have hoped to get from the Apple Silicon Mac Pro. If it comes down to "It has slots!" coupled with "but those slots are not for graphics cards" I sense a lot of potential customers are wondering what's going on.
The Apple Silicon Mac Pro clearly has benefits over the Mac Studio for those who truly need its relatively narrowly scoped range of extensibility. Those folks are out there - somewhere, but they are probably fewer and further between than those who will be totally satisfied with a Mac Studio Max or Ultra, a MacBook Pro with M1/M2 Max, or a PC when it comes time to replace their current Intel Mac Pro.
I'm not an expert on what kinds of non-GPU PCIe boards are available on the market that would make the availability of those expansion slots a must-have. I've used high bandwidth expansion slots on non-Mac platforms to host things like array/vector processor boards used to do things like acoustic target+noise simulation. But truth be told, I don't even know if such boards exist anymore, much less ones that fit into PCIe slots and intended to be used on a Mac platform with available off-the-shelf driver support. Somebody somewhere is going to be absolutely thrilled with the Apple Silicon Mac Pro and sees it as a magnificent improvement over the Mac Studio and an upgrade over the latest Intel Mac Pro. But I have no idea of who those people may be. Does anyone know someone who has jumped on a new Mac Pro?
It seems like Apple would be inclined to team up with a system integrator/builder, whether internal or third party, or an OEM, to sell "custom" Apple Silicon Mac Pros that are preconfigured with whatever PCIe expansion cards are needed to serve specific applications and solve specific problems, rather than being a general purpose computing platform. As a general purpose computing platform the Pro seems to lag the Studio in the bang-for-the-buck metric. But for solving some problems, it may more than earn its keep.
Just checked Apple’s leadership team profiles on https://www.apple.com/leadership/. Did I miss something? I don’t see Mark Gurman’s profile anywhere on that site. Where does he sit in Apple’s leadership team that decides the execution of Apple’s product strategy?
Good grief. 2 points here. 1. Apple has released four Mac Pros in the entire line's existence: 2006, 2013, 2019 and 2023. No idea why anyone ever expected the shift to Apple Silicon was going to mean it adopting some 2 or even 3 year refresh schedule especially when you consider that the Mac Studio and according to some people - the 16" MacBook Pro can replace the (cheaper with default specs) Mac Pro for a ton of users.
2. Intel and AMD aren't done. The 5nm Threadripper PRO 7985WX that AMD will announce in September will max out at 64 cores. And September 2025 Intel is going to announce a Core i9 - a regular desktop chip, not a workstation or server chip - with 40 cores. Apple knows what Intel and AMD are doing. (They also know that a ton of the Mac Pro customer base has since switched to Linux - before you snark "Linux has about 100 users" take a good look at the Mac Pro sales figures - and this is what they will be using.) As this group doesn't care nearly as much about power efficiency as Apple fans think that they do (should) Apple knows that keeping up with Intel and AMD in this space will be a challenge.
3. Why? I have been saying this on this board for some time. And this is why the "Apple bungled the Silicon switch" argument is wrong: Apple's main problem isthey have only one basic CPU core design. Give them credit: they get a lot out of it. They reuse old performance cores as efficiency cores, and for the performance cores they vary the number and power draw from 4 in a low power mode in a thin, inexpensive fanless device like the M1 iPad Air to 24 cores that they pump massive power to in the M2 Ultra Mac Pro. But still: they use an outdated version of this core as efficiency cores and the modern version as performance ones. Intel, for example? Does not have this problem. Intel has an actual efficiency core that is fundamentally different from their performance cores (1 thread instead of 2 and for now doesn't support AVX-512 ISA). They also have Core i3, Core i5, Core i7, Core i9 and Xeon cores for 6 total (they also had Pentium and Celeron but starting with 13th gen they ditched them in favor of the faster e-core). AMD? Same. Athlon/A-Series (which seems to have met the same fate as Intel's Pentium and Celeron), Ryzen 3, Ryzen 5, Ryzen 7, Ryzen 9, Threadripper, Epyc. The Core i9/Ryzen 9/Threadripper/Xeon/Epyc batch uses a ton of power - though notably not so much in the Core i9/Ryzen 9 laptop versions - but Apple simply cannot match the single core performance of Core i9/Ryzen 9 and above. And no, this isn't x86 Windows fanboy nonsense. Apple themselves admitted the same. Yes, the 3nm M3 Ultra will beat anything that Intel and AMD have ... because Intel will be on 7nm (10nm for desktop/workstation/server) and because AMD, though on 5nm, hasn't adopted big.LITTLE on desktop yet (their big.LITTLE chips for laptops and handheld gaming consoles are just now coming out). But when Intel and Apple get to 3nm in 2025?
Yes, Apple could spend billions in R&D on developing cores to compete with Ryzen 9 and above. Or to save time and money they could license the Neoverse ARM server cores and customize them. But why? Remember: Apple only sells 25 million Macs in a good (actually great) year with the vast majority being M2 Pro and below. Were Apple to develop a Mac Pro capable of beating Linux workstations running the AMD Threadripper Pro in single core and multicore performance, how many more people are actually going to buy it? Apple knows the answer to this question already, even if the many Apple fan sites out there aren't going to admit it. Consider that even were Apple to start selling 10 million workstations a year, you will see Nvidia, Qualcomm and the rest react with their own ARM workstation chips. Nvidia would release a scaled down version of their Grace server/cloud SOCs - which use Neoverse V2 cores and their excellent GPUs - for workstations the very day Apple provides evidence of a market for it. There is even an ARM Ubuntu workstation and server stack already in place to take advantage of it.
Sorry guys, while I do buy and use Apple products, I am not an "Apple knows best" sort by any stretch of the imagination. But this is one instance where you are flat out wrong. Making a competitive Mac Pro requires making a competitive workstation CPU, and even if Apple could make one it wouldn't benefit them financially or in terms of market position. So were Apple to abandon the Mac Pro down the line, big deal: Google and Intel both exited the PC market because they were faced with the same set of facts.
Just checked Apple’s leadership team profiles on https://www.apple.com/leadership/. Did I miss something? I don’t see Mark Gurman’s profile anywhere on that site. Where does he sit in Apple’s leadership team that decides the execution of Apple’s product strategy?
Good grief. 2 points here. 1. Apple has released four Mac Pros in the entire line's existence: 2006, 2013, 2019 and 2023. No idea why anyone ever expected the shift to Apple Silicon was going to mean it adopting some 2 or even 3 year refresh schedule especially when you consider that the Mac Studio and according to some people - the 16" MacBook Pro can replace the (cheaper with default specs) Mac Pro for a ton of users.
2. Intel and AMD aren't done. The 5nm Threadripper PRO 7985WX that AMD will announce in September will max out at 64 cores. And September 2025 Intel is going to announce a Core i9 - a regular desktop chip, not a workstation or server chip - with 40 cores. Apple knows what Intel and AMD are doing. (They also know that a ton of the Mac Pro customer base has since switched to Linux - before you snark "Linux has about 100 users" take a good look at the Mac Pro sales figures - and this is what they will be using.) As this group doesn't care nearly as much about power efficiency as Apple fans think that they do (should) Apple knows that keeping up with Intel and AMD in this space will be a challenge.
3. Why? I have been saying this on this board for some time. And this is why the "Apple bungled the Silicon switch" argument is wrong: Apple's main problem isthey have only one basic CPU core design. Give them credit: they get a lot out of it. They reuse old performance cores as efficiency cores, and for the performance cores they vary the number and power draw from 4 in a low power mode in a thin, inexpensive fanless device like the M1 iPad Air to 24 cores that they pump massive power to in the M2 Ultra Mac Pro. But still: they use an outdated version of this core as efficiency cores and the modern version as performance ones. Intel, for example? Does not have this problem. Intel has an actual efficiency core that is fundamentally different from their performance cores (1 thread instead of 2 and for now doesn't support AVX-512 ISA). They also have Core i3, Core i5, Core i7, Core i9 and Xeon cores for 6 total (they also had Pentium and Celeron but starting with 13th gen they ditched them in favor of the faster e-core). AMD? Same. Athlon/A-Series (which seems to have met the same fate as Intel's Pentium and Celeron), Ryzen 3, Ryzen 5, Ryzen 7, Ryzen 9, Threadripper, Epyc. The Core i9/Ryzen 9/Threadripper/Xeon/Epyc batch uses a ton of power - though notably not so much in the Core i9/Ryzen 9 laptop versions - but Apple simply cannot match the single core performance of Core i9/Ryzen 9 and above. And no, this isn't x86 Windows fanboy nonsense. Apple themselves admitted the same. Yes, the 3nm M3 Ultra will beat anything that Intel and AMD have ... because Intel will be on 7nm (10nm for desktop/workstation/server) and because AMD, though on 5nm, hasn't adopted big.LITTLE on desktop yet (their big.LITTLE chips for laptops and handheld gaming consoles are just now coming out). But when Intel and Apple get to 3nm in 2025?
Yes, Apple could spend billions in R&D on developing cores to compete with Ryzen 9 and above. Or to save time and money they could license the Neoverse ARM server cores and customize them. But why? Remember: Apple only sells 25 million Macs in a good (actually great) year with the vast majority being M2 Pro and below. Were Apple to develop a Mac Pro capable of beating Linux workstations running the AMD Threadripper Pro in single core and multicore performance, how many more people are actually going to buy it? Apple knows the answer to this question already, even if the many Apple fan sites out there aren't going to admit it. Consider that even were Apple to start selling 10 million workstations a year, you will see Nvidia, Qualcomm and the rest react with their own ARM workstation chips. Nvidia would release a scaled down version of their Grace server/cloud SOCs - which use Neoverse V2 cores and their excellent GPUs - for workstations the very day Apple provides evidence of a market for it. There is even an ARM Ubuntu workstation and server stack already in place to take advantage of it.
Sorry guys, while I do buy and use Apple products, I am not an "Apple knows best" sort by any stretch of the imagination. But this is one instance where you are flat out wrong. Making a competitive Mac Pro requires making a competitive workstation CPU, and even if Apple could make one it wouldn't benefit them financially or in terms of market position. So were Apple to abandon the Mac Pro down the line, big deal: Google and Intel both exited the PC market because they were faced with the same set of facts.
I think you read way way way too much into my comment. The only thing I found to be somewhat off-putting was the statement:
"In explaining which models should receive the M3 Ultra chip, Gurman says the list is the Mac Studio and the Mac Pro ..."
I happen to believe that Apple and its leadership team should be in full control of deciding which products receive the M3. No more, no less. If Apple decides to put the M3 in MacBook Pro or iPad Pro I don't think anyone is going to complain and they aren't going ask for permission from Gurman.
To your points:
1. Totally agree. That's why I own a Mac Studio.
2. Totally agree. I would also add that those who thought that Apple's M1 was the beginning of the end for Intel or AMD are sadly mistaken. Intel was wounded when it lost Apple's business, but it was far from being a fatal wound. They are recovering quite well and have the benefit of seeing and learning from what Apple has already done. They are simply not worried about efficiency to the extent Apple is because a lot of their stuff is tethered to a wall outlet. I use Linux, so the number of Linux users is at least 101.
3. That's a lot to digest. But I think the new Mac Pro indicates that Apple has conceded a certain segment of the high end workstation market to the competition. Some but not all. Yes, they are leaving some money on the table, but they are making plenty of money in other areas that more than makes up for the loss. They don't have to win everything everywhere. Apple can still sell some Mac Pros into the niches where they are a good fit. As you've alluded to, these niches don't require yearly updates but just enough to keep that segment happy. How long they can sustain this is really up to Apple and the ROI numbers. They always try to have other products ramping up to replace (and usually surpass) the revenue from the ones that are fading away.
Overall, the Mac Pro just doesn't make a lot of sense these days.
A hefty workstation like this would normally support multiple CPUs, ridiculous amounts of memory, and multiple powerful GPUs. The new Mac Pro doesn't support any of this, so the only thing you can use all that space for are very specialized IO cards - fibre channel, infiniband, and the like.
That's a pretty subniche bit of an already niche market, so I wouldn't be surprise to see it go and Apple just quietly abandoning it.
Just checked Apple’s leadership team profiles on https://www.apple.com/leadership/. Did I miss something? I don’t see Mark Gurman’s profile anywhere on that site. Where does he sit in Apple’s leadership team that decides the execution of Apple’s product strategy?
Word is that Gurman, and other Bloomberg writers, get paid a bonus for articles that move a company's stock price. He and others there have a financial incentive to make provocative statements (remember the now entirely debunked Big Hack article that Bloomberg never retracted?) to try and affect the stock price. By couching them as hypotheticals they avoid the hits their credibility ought to take.
The bottom line is that anything published by Bloomberg must be taken with a very large grain of salt.
So in other words, Apple Insider is spreading misinformation.
Don't feel too bad about it. If you have any form of Retail general investment product per of that will be shares in a fund that specifically makes its money on so-called "sentiment trading".
The biggest weird thing about the current Mac Pro is they kept the design and only upgraded it with a chip they effectively had 2 years ago.
Also the PCI expansion has long been redundant with Thunderbolt and Apple has actually surprisingly killed the need for the Mac Pro with the Mac Studio being the same thing for $3k less. If they had kept higher end desktop chips tied to the iMac, the Mac Pro would have still been relevant compared to the alternative of more expensive iMac Pro models.
As for them updating the Mac Pro and money, I assume they kept the design so moving forward they can annually or semi-annualy update the Mac Pro alongside the Studio with a current gen Ultra chip. I also assume they currently have a team working on a redesigned Mac Pro chassis aimed for the late 2020s, but they might use these metrics to abandon that redesign. But unless that redesign brings back upgradability, it's a moot point anyways.
The biggest apparent technical issue seems to be the failure of the quad-die Apple Silicon chip. Perhaps they could make up for this with a dual socket system?
While there were rumors, I don't think Apple ever had a four die system. The two die produces are two die, edge to edge on a multi-chip module (MCM), with a huge and ridiculously fast "UltraFusion" interconnect that make them essentially one chip. So far, there is only one such interconnect per chip. AMD does their "chiplet" thing with multiple on chip links on an interconnect called Infinity Fabric, much more flexible but not as fast as Apple's. To make a four chip solution using what Apple has today, you'd need either two Ultrafusion interconnects per chip or a central four!]-way switch.
Some more conventional "dual socket" architecture would have two SoCs connected via... PCIe? Some other interconnect? That leaves very few PCie lanes for anything else today, and it's no longer a unified architecture, but some kind of cluster. Very different.
It'd also be great if they found a way to do a two tiered memory system, so you could add terabytes of DRAM to the embedded system. It sounds difficult, but Apple already pulled something similar off with the old two tiered SSD/HDD systems. You'd let the 192GB of on-die RAM page out to the 8 TB of DIMMs.
The DRAM isn't on-die, but it's on the MCM. It's conventional DDR5, but tighter timing on the MCM lets it run faster. They could support another two or four DDR5 channels for off-module DRAM, but tjat is a ton of extra signals to deal with.
Also, you're not using off-module DRAM as paged memory, it's not all much slower, but the overhead would be killer. Rather, the OS would have to manage tiers of memory, like most NUMA architrctures impose. All critical OS memory gets allocated in on-module RAM.
It's a relic from another era. Mac Studio covers 98% of the use-cases and sorry but the remaining folks who "need" these really don't- they just need to update how they handle I/O and seek more efficient processing.
It's a relic from another era. Mac Studio covers 98% of the use-cases and sorry but the remaining folks who "need" these really don't- they just need to update how they handle I/O and seek more efficient processing.
Agreed. The antiquated Pro is an inevitable victim of the switch to Apple silicon. The revolution of Apple silicon is all about bringing as much possible on-chip for speed gains and power savings. This is the antithesis of a 'tower' desktop.
I loved my 2008 Mac Pro. I could not fail to admire the beautiful engineering every time I had reason to switch a graphics card, add a hard drive, then latterly an SSD, or add more bullet-proof ECC RAM. It used so much power it kept my office warm and, when it was working hard, I couldn't hear the rain outside. Those days are gone. My modest M1 MacBook Air that I replaced it with is faster and silent. It drives a huge external display and any other functionality is provided care of Thunderbolt rather than huge, cumbersome PCI cards.
The Mac Pro is a dead parrot. Deceased. It has outlived its evolutionary purpose.
In the Mac Pro's favour I will say that I now require heating in my office and I find the sound of rain, and other things I couldn't previously hear, quite distracting.
Apple cheaped out and chose to not innovate further. They could have launched a Mac Pro with an m extreme chip or built it to allow multiple SOCs. But Ternus couldn’t be bothered I guess.
If there is to be a future, they must build it to accommodate multiple SOCs and replacement/upgrade SOC MODULES.
IF THE MAC STUDIO IS AS GOOD AS IT GRTS, that’s not that hot. It’s capable and powerful, but not the world destroyer the g5 power Mac / cheese grater Mac Pro was or that the 2019 Mac Pro was.
Apple could dominate the performance arena, but apparently the Dell motto of “good enough” has carried over.
It's a relic from another era. Mac Studio covers 98% of the use-cases and sorry but the remaining folks who "need" these really don't- they just need to update how they handle I/O and seek more efficient processing.
Um no, Thunderbolt does not have enough bandwidth for 8K video it has to be done via PCI-E.
All the people panning these machines expected their idea of a workstation. This was never on the cards what was on the cards is what we got a very niche machine to host AVID HDX cards for ProTools and SDI video I/O cards for Resolve/Premier/Final Cut.
I think this is the response Apple was hoping for so they have a reason to kill off the Mac Pro. Apple SoC isn't designed to be used in a workstation type computer, workstations need expandability for RAM and specialized boards. The Mac Studio is the new Mac Pro for Apple SoC era.
I had a Mac Pro "Cheese grater" and they were great. Cheese grater's were expensive, but not so much that working people couldn't afford them like the "Trash Bin" and New Cheese grater Mac Pros. The original Cheese Grater was expandable the CPU, RAM, Storage so you could buy a base model and upgrade the components as you needed more power. The new SoC Mac Pro lost all that and only something film studios can afford and they have issues with the restriction of the PCIe slots of the new Mac Pro.
So I think Apple know with there silicon the old idea of a Mac Pro is over so just need an excuse to EOL Mac Pro.
Apple hasn't been shy about EOL designs they don't see a path for. That said the Mac Pro has bounced around and seems to have sentimental value something we rarely see. The current design was done at a time that they would have have ASi in mind.
Plus there are interesting things happening that Apple would seem to want to be involved with as they develop and open up interesting options to be the developers/researchers consumer hardware of choice.
It seems they could go either way. The studio is the successor to the trash can which was the successor to the cubeMac. and they acknowledged that is great for a lot of people but not that top niche they want to be in.
I see no fumble. It was clear after the launch of Apple silicone that soc was what they were doing and the macpro wasn’t going to be the macpro of yesteryear.
… I'm not sure the market for such a machine is big enough to justify that investment for a sprawling, $3 trillion company.
Couldn't have said it better myself. That Apple behaved like a lean, mean startup for almost two-and-a-half decades after its brush with bankruptcy and Steve Jobs's return, I could perfectly understand and accept. Except they've been an absolute juggernaut ever since the iPhone became a thing, nearly fifteen years ago. Heck, they even design their own chips…
Apple is at this point where not only they can, but also MUST produce some inspirational, flagship, loss-leader products. Even if they had to sell a $5000 Mac Pro at a loss (yes, you've read it correctly; they might be taking a hit just due to how expensive designing a specific chip for it might be), if that's the only way they can keep the high-end markets which need obscene amounts of memory and graphics capability and make them them look cool (and actually be useful to society; scientific applications which deal with massive datasets come to mind), so be it.
Consider it a form of a still somewhat self-serving form of hardware philanthropy. They did it before, at a time when it made less sense from a financial or even strategic standpoint (remember the XServe, anyone?), so there's no excuse now. It only makes them look even more like bean counters.
I think they should abandon the Mac Pro and revive the Mac Server with drive trays, slots for IO and redundancy. That would sell considering the power efficiency of the M series chips.
Who would want to own an expensive Mac Server when AWS Graviton is *right there*? Apple's product is distinguished as a user device, and that's where they can justify a price premium.
Um, what?
That's not even a product. Small businesses don't want "cloud" (somebody else's computer) garbage, they want servers.
Only Apple has the customer data to tell them what's "worth it" or not in terms of R&D and production investment, but with the high-end Mac Studio now taking at least some of what was already a very limited market for the Mac Pro, I'm not surprised to read that its future is on thin ice. It would probably take a big investment of both money and human talent to create a next gen Mac Pro that is significantly faster and more expandable than Mac Studio and I'm not sure the market for such a machine is big enough to justify that investment for a sprawling, $3 trillion company.
That's already reflected in its price. As long as there's a market and Apple can continue to make a profit from it, they'll keep it around. They updated the Mac Pro knowing the size of the targeted user base, especially after the Ultra Studio was released.
I do have a feeling the next iteration will be a smaller chassis with half the number of PCI expansion slots though. I can't imagine there's any significant share of users that need all of those slots, especially since none can use a GPU card.
The Mac Pro case, at this point, is just that, a case. That's like Stanley Tools, saying they are discontinuing their Hammer, cause they aren't selling as much. I mean it's a piece of metal with a wooden handle.
The Mac Pro case is a machined case, motherboard, with SoC. DONE... (oh and yes the wheels, for the smart a$$es out there, but...)
I think they can say, ONE MAC PRO, coming up... zzrt zzrt rrr, there's the case, put a MOBO, in done... I mean it's pretty much mastered! (2000, 3000, 5,000, how ever many they make)
Just checked Apple’s leadership team profiles on https://www.apple.com/leadership/. Did I miss something? I don’t see Mark Gurman’s profile anywhere on that site. Where does he sit in Apple’s leadership team that decides the execution of Apple’s product strategy?
Word is that Gurman, and other Bloomberg writers, get paid a bonus for articles that move a company's stock price. He and others there have a financial incentive to make provocative statements (remember the now entirely debunked Big Hack article that Bloomberg never retracted?) to try and affect the stock price. By couching them as hypotheticals they avoid the hits their credibility ought to take.
The bottom line is that anything published by Bloomberg must be taken with a very large grain of salt.
So in other words, Apple Insider is spreading misinformation.
Are They? anonymouse hasn't provided any links or evidence. Or more succinctly: "By couching them as hypotheticals they avoid the hits their credibility ought to take."
Comments
In this case, it doesn't take a genius, a leaky insider, or Gurman's opinion to question the value proposition that motivated Apple to ship the Apple Silicon Mac Pro. Perhaps if Apple had held back a little on the Mac Studio and made it more of a juiced-up Mac mini there would be a little less head scratching. But the Mac Studio Ultra versions seem to have stolen much of the thunder that Apple may have hoped to get from the Apple Silicon Mac Pro. If it comes down to "It has slots!" coupled with "but those slots are not for graphics cards" I sense a lot of potential customers are wondering what's going on.
The Apple Silicon Mac Pro clearly has benefits over the Mac Studio for those who truly need its relatively narrowly scoped range of extensibility. Those folks are out there - somewhere, but they are probably fewer and further between than those who will be totally satisfied with a Mac Studio Max or Ultra, a MacBook Pro with M1/M2 Max, or a PC when it comes time to replace their current Intel Mac Pro.
I'm not an expert on what kinds of non-GPU PCIe boards are available on the market that would make the availability of those expansion slots a must-have. I've used high bandwidth expansion slots on non-Mac platforms to host things like array/vector processor boards used to do things like acoustic target+noise simulation. But truth be told, I don't even know if such boards exist anymore, much less ones that fit into PCIe slots and intended to be used on a Mac platform with available off-the-shelf driver support. Somebody somewhere is going to be absolutely thrilled with the Apple Silicon Mac Pro and sees it as a magnificent improvement over the Mac Studio and an upgrade over the latest Intel Mac Pro. But I have no idea of who those people may be. Does anyone know someone who has jumped on a new Mac Pro?
It seems like Apple would be inclined to team up with a system integrator/builder, whether internal or third party, or an OEM, to sell "custom" Apple Silicon Mac Pros that are preconfigured with whatever PCIe expansion cards are needed to serve specific applications and solve specific problems, rather than being a general purpose computing platform. As a general purpose computing platform the Pro seems to lag the Studio in the bang-for-the-buck metric. But for solving some problems, it may more than earn its keep.
1. Apple has released four Mac Pros in the entire line's existence: 2006, 2013, 2019 and 2023. No idea why anyone ever expected the shift to Apple Silicon was going to mean it adopting some 2 or even 3 year refresh schedule especially when you consider that the Mac Studio and according to some people - the 16" MacBook Pro can replace the (cheaper with default specs) Mac Pro for a ton of users.
2. Intel and AMD aren't done. The 5nm Threadripper PRO 7985WX that AMD will announce in September will max out at 64 cores. And September 2025 Intel is going to announce a Core i9 - a regular desktop chip, not a workstation or server chip - with 40 cores. Apple knows what Intel and AMD are doing. (They also know that a ton of the Mac Pro customer base has since switched to Linux - before you snark "Linux has about 100 users" take a good look at the Mac Pro sales figures - and this is what they will be using.) As this group doesn't care nearly as much about power efficiency as Apple fans think that they do (should) Apple knows that keeping up with Intel and AMD in this space will be a challenge.
3. Why? I have been saying this on this board for some time. And this is why the "Apple bungled the Silicon switch" argument is wrong: Apple's main problem is they have only one basic CPU core design. Give them credit: they get a lot out of it. They reuse old performance cores as efficiency cores, and for the performance cores they vary the number and power draw from 4 in a low power mode in a thin, inexpensive fanless device like the M1 iPad Air to 24 cores that they pump massive power to in the M2 Ultra Mac Pro. But still: they use an outdated version of this core as efficiency cores and the modern version as performance ones. Intel, for example? Does not have this problem.
Intel has an actual efficiency core that is fundamentally different from their performance cores (1 thread instead of 2 and for now doesn't support AVX-512 ISA). They also have Core i3, Core i5, Core i7, Core i9 and Xeon cores for 6 total (they also had Pentium and Celeron but starting with 13th gen they ditched them in favor of the faster e-core). AMD? Same. Athlon/A-Series (which seems to have met the same fate as Intel's Pentium and Celeron), Ryzen 3, Ryzen 5, Ryzen 7, Ryzen 9, Threadripper, Epyc. The Core i9/Ryzen 9/Threadripper/Xeon/Epyc batch uses a ton of power - though notably not so much in the Core i9/Ryzen 9 laptop versions - but Apple simply cannot match the single core performance of Core i9/Ryzen 9 and above. And no, this isn't x86 Windows fanboy nonsense. Apple themselves admitted the same. Yes, the 3nm M3 Ultra will beat anything that Intel and AMD have ... because Intel will be on 7nm (10nm for desktop/workstation/server) and because AMD, though on 5nm, hasn't adopted big.LITTLE on desktop yet (their big.LITTLE chips for laptops and handheld gaming consoles are just now coming out). But when Intel and Apple get to 3nm in 2025?
Yes, Apple could spend billions in R&D on developing cores to compete with Ryzen 9 and above. Or to save time and money they could license the Neoverse ARM server cores and customize them. But why? Remember: Apple only sells 25 million Macs in a good (actually great) year with the vast majority being M2 Pro and below. Were Apple to develop a Mac Pro capable of beating Linux workstations running the AMD Threadripper Pro in single core and multicore performance, how many more people are actually going to buy it? Apple knows the answer to this question already, even if the many Apple fan sites out there aren't going to admit it. Consider that even were Apple to start selling 10 million workstations a year, you will see Nvidia, Qualcomm and the rest react with their own ARM workstation chips. Nvidia would release a scaled down version of their Grace server/cloud SOCs - which use Neoverse V2 cores and their excellent GPUs - for workstations the very day Apple provides evidence of a market for it. There is even an ARM Ubuntu workstation and server stack already in place to take advantage of it.
Sorry guys, while I do buy and use Apple products, I am not an "Apple knows best" sort by any stretch of the imagination. But this is one instance where you are flat out wrong. Making a competitive Mac Pro requires making a competitive workstation CPU, and even if Apple could make one it wouldn't benefit them financially or in terms of market position. So were Apple to abandon the Mac Pro down the line, big deal: Google and Intel both exited the PC market because they were faced with the same set of facts.
"In explaining which models should receive the M3 Ultra chip, Gurman says the list is the Mac Studio and the Mac Pro ..."
I happen to believe that Apple and its leadership team should be in full control of deciding which products receive the M3. No more, no less. If Apple decides to put the M3 in MacBook Pro or iPad Pro I don't think anyone is going to complain and they aren't going ask for permission from Gurman.
To your points:
1. Totally agree. That's why I own a Mac Studio.
2. Totally agree. I would also add that those who thought that Apple's M1 was the beginning of the end for Intel or AMD are sadly mistaken. Intel was wounded when it lost Apple's business, but it was far from being a fatal wound. They are recovering quite well and have the benefit of seeing and learning from what Apple has already done. They are simply not worried about efficiency to the extent Apple is because a lot of their stuff is tethered to a wall outlet. I use Linux, so the number of Linux users is at least 101.
3. That's a lot to digest. But I think the new Mac Pro indicates that Apple has conceded a certain segment of the high end workstation market to the competition. Some but not all. Yes, they are leaving some money on the table, but they are making plenty of money in other areas that more than makes up for the loss. They don't have to win everything everywhere. Apple can still sell some Mac Pros into the niches where they are a good fit. As you've alluded to, these niches don't require yearly updates but just enough to keep that segment happy. How long they can sustain this is really up to Apple and the ROI numbers. They always try to have other products ramping up to replace (and usually surpass) the revenue from the ones that are fading away.
Also the PCI expansion has long been redundant with Thunderbolt and Apple has actually surprisingly killed the need for the Mac Pro with the Mac Studio being the same thing for $3k less. If they had kept higher end desktop chips tied to the iMac, the Mac Pro would have still been relevant compared to the alternative of more expensive iMac Pro models.
As for them updating the Mac Pro and money, I assume they kept the design so moving forward they can annually or semi-annualy update the Mac Pro alongside the Studio with a current gen Ultra chip. I also assume they currently have a team working on a redesigned Mac Pro chassis aimed for the late 2020s, but they might use these metrics to abandon that redesign. But unless that redesign brings back upgradability, it's a moot point anyways.
Some more conventional "dual socket" architecture would have two SoCs connected via... PCIe? Some other interconnect? That leaves very few PCie lanes for anything else today, and it's no longer a unified architecture, but some kind of cluster. Very different.
The DRAM isn't on-die, but it's on the MCM. It's conventional DDR5, but tighter timing on the MCM lets it run faster. They could support another two or four DDR5 channels for off-module DRAM, but tjat is a ton of extra signals to deal with.
Also, you're not using off-module DRAM as paged memory, it's not all much slower, but the overhead would be killer. Rather, the OS would have to manage tiers of memory, like most NUMA architrctures impose. All critical OS memory gets allocated in on-module RAM.
I loved my 2008 Mac Pro. I could not fail to admire the beautiful engineering every time I had reason to switch a graphics card, add a hard drive, then latterly an SSD, or add more bullet-proof ECC RAM. It used so much power it kept my office warm and, when it was working hard, I couldn't hear the rain outside. Those days are gone. My modest M1 MacBook Air that I replaced it with is faster and silent. It drives a huge external display and any other functionality is provided care of Thunderbolt rather than huge, cumbersome PCI cards.
The Mac Pro is a dead parrot. Deceased. It has outlived its evolutionary purpose.
In the Mac Pro's favour I will say that I now require heating in my office and I find the sound of rain, and other things I couldn't previously hear, quite distracting.
IF THE MAC STUDIO IS AS GOOD AS IT GRTS, that’s not that hot. It’s capable and powerful, but not the world destroyer the g5 power Mac / cheese grater Mac Pro was or that the 2019 Mac Pro was.
All the people panning these machines expected their idea of a workstation. This was never on the cards what was on the cards is what we got a very niche machine to host AVID HDX cards for ProTools and SDI video I/O cards for Resolve/Premier/Final Cut.
Plus there are interesting things happening that Apple would seem to want to be involved with as they develop and open up interesting options to be the developers/researchers consumer hardware of choice.
It seems they could go either way. The studio is the successor to the trash can which was the successor to the cubeMac. and they acknowledged that is great for a lot of people but not that top niche they want to be in.
Couldn't have said it better myself. That Apple behaved like a lean, mean startup for almost two-and-a-half decades after its brush with bankruptcy and Steve Jobs's return, I could perfectly understand and accept. Except they've been an absolute juggernaut ever since the iPhone became a thing, nearly fifteen years ago. Heck, they even design their own chips…
Apple is at this point where not only they can, but also MUST produce some inspirational, flagship, loss-leader products. Even if they had to sell a $5000 Mac Pro at a loss (yes, you've read it correctly; they might be taking a hit just due to how expensive designing a specific chip for it might be), if that's the only way they can keep the high-end markets which need obscene amounts of memory and graphics capability and make them them look cool (and actually be useful to society; scientific applications which deal with massive datasets come to mind), so be it.
Consider it a form of a still somewhat self-serving form of hardware philanthropy. They did it before, at a time when it made less sense from a financial or even strategic standpoint (remember the XServe, anyone?), so there's no excuse now. It only makes them look even more like bean counters.
That's already reflected in its price. As long as there's a market and Apple can continue to make a profit from it, they'll keep it around. They updated the Mac Pro knowing the size of the targeted user base, especially after the Ultra Studio was released.
I do have a feeling the next iteration will be a smaller chassis with half the number of PCI expansion slots though. I can't imagine there's any significant share of users that need all of those slots, especially since none can use a GPU card.
The Mac Pro case is a machined case, motherboard, with SoC. DONE...
(oh and yes the wheels, for the smart a$$es out there, but...)
I think they can say, ONE MAC PRO, coming up... zzrt zzrt rrr, there's the case, put a MOBO, in done...
I mean it's pretty much mastered! (2000, 3000, 5,000, how ever many they make)
2¢