Bloomberg article said: "to try to sharpen the company’s focus on consumer products that generate the bulk of its revenue". So basically iPhone, Inc. I suspect Mac Pro and Mac mini will be next on the chopping block.
Damn shame. One of Apples best kept secrets was how rock solid those routers were, at least on the latest firmware. They fell behind on features in the last few years, but still seemed to have a lot more uptime than even the most modern equipment by some others, thanks maybe to the FreeBSD/NetBSD core. Off to Eero I guess.One wonders if the Mac Pro which also hasn’t been updated in 3 years will get put into a can that looks like itself.That’ll be a shame. Having a POSIX certified FreeBSD based operating system using workstation that could take oodles of ECC RAM out there, even if I didn’t need such a thing anymore, was somehow…It sounds really odd, but somehow a comfort. This essay does it justice.https://marco.org/2016/11/05/world-without-mac-pro
they reported a $406 million net loss for the latest quarter and $446m loss for the last 9 months. That's on top of net losses before this. Under assets, they have $2.8b current assets and $1.4b current liabilities. They had negative equity last year, which is now positive but the money has come from additional paid in capital of $1.25b, i.e from investors/shareholders and not from earnings. Their stock price has gone up over 4x in the last year for some reason, despite making over $400m in losses.
This period was a major launch for AMD with Polaris and they still aren't making profit from it. They made an announcement about some workstation GPUs earlier in the year:
but these are mostly old GPU architectures. The W9100 is 5.24TFLOPs (4K display support max), the 2013 D700 is 3.5TFLOPs. A single NVidia 1080 is 8.2TFLOPs. They really need to be using 100%+ faster GPUs after 3 years. The CPUs would go up 50% (18-core max). The PCIe bandwidth they have is limited, same PCIe lanes as before so how do they double the throughput per TB port for TB3 (they might have to do 4x TB3 + 6x USB-C, all supporting display output)?
The component manufacturers have to deliver something worthwhile to make a packaged upgrade worthwhile and they aren't. There are cases where having the high Xeon cores helps but when you look at how the Xeons perform vs the more modern i7 architectures, they don't hold up very well:
https://browser.primatelabs.com/mac-benchmarks (iMac on multi isn't that far behind the top Mac Pro and that's 4-core vs 12-core)
iMac quad-core 2015 Cinebench = 871, Mac Pro 2013 = 1524. (12-core is just 75% faster than quad-core)
The iMac (including 5K Retina display) costs $2500, the Mac Pro 12-core alone costs $7000. With the latest CPUs, you'd get 50% extra performance but the Skylake CPUs will increase a bit too. If Intel had a 6/8/10-core Skylake i7 (e.g Skylake-X 2017) and Apple used dual NVidia 1080 (18TFLOPs), priced at $3-5k, that would probably serve this level of buyer better than Xeon+AMD. The price/performance ratio would be better and raw performance would be better as it uses the latest architecture from every vendor. This is enthusiast-class vs workstation-class but the workstation market is much smaller and it's down to results in the end. The i7X chips support similar PCIe lanes as the Xeons so the main downside is no ECC and max 64GB RAM (Intel can change this if they wanted).
If AMD doesn't last, Apple can't just switch to Quadro because the price is far higher. I think AMD has bought themselves another couple of years but long-term, the lower-end hardware is going to keep getting closer in performance to workstation class. There's a threshold of demand below which Apple can't be expected to continue supporting any product because the demand (actual sales) is their reference for how necessary that product is.
When it comes to networking hardware, the router vendors all get chips from common component manufacturers, Apple is in the list here:
As with any product, there are differences in the user experience with the whole product like software support/features but routers are pretty basic in what they do. Apple doesn't make ethernet cables because they have a basic function and routers are like this now. They have cumbersome web interfaces but they are usually setup by whoever installs the network and never need to be touched again.
Apple's router sales have to be under 1 million units per year because they are noted as having under 1% WLAN marketshare. This isn't even a significant portion of Mac users. This is the same reason they dropped the XServe, some iPod models etc. There's no point in supplying something when there's not enough demand for it. Given that most people will get their base router from the ISP, the bigger market is for easy to setup wifi repeaters.
They can even incorporate powerline networking here where they send the network to each repeater over the electrical sockets. The first repeater can plug into the base router over ethernet and every other repeater just gets plugged into a socket. All communication would be encrypted between repeaters and they'd be paired together to avoid conflicts with other properties like apartment blocks with shared electrical systems. This way you'd never have to deal with the base router as the ethernet connection doesn't really need any configuration and you wouldn't need to get the repeaters to communicate with the base router over wifi. This would work well even with buildings with solid stone walls or very large buildings.