Take the long term view. If mankind goes extinct, then absolutely nothing we every worried about in the short term matters one iota on that day after the last human dies. Right now, and as long as we're only on Earth, any number of catastrophies could kill us all in sort order, some we create, some that just happen. Either way, mankind and every thing it ever did ceases to matter at all.
Or, we keep working to fix this ultimate problem. Taking the million year view, moving sustainably beyond earth is the most important thing humanity will every do.
If you're paying a signer, that means they're signing your image with their private key so it works with their KEK (Key Exchange Key), presumably already in Flash. But that's not the correct process.
Rather, the original intention was for the user to be in charge of keys. You can go with the PK (platform key, which is the root key) supplied by your board manufacturer, but you're supposed to have the ability to replace it (and of course, clear out any KEKs) if you like. And you're supposed to be able to install your own KEK.
So for example, you compile up your own Linux kernel to use with Slackware or whatever. Long ago, you generated your own private key for signing your personal Linux binaries, and generated a KEK for your motherboard for that private key, which lives in the signature database right next to the Microsoft KEK and any others you or your motherboard manufacturer might have installed. So you sign the new binary, and it just loads (actually, the signature database can deal with both signed and unsigned binaries -- if you image isn't signed, the OS can be securely installed by storing a SHA-256 hash of the boot image in the signature database). No need to involve Microsoft, no need to risk boot from a virus.
But of course, making it this simple gets around Microsoft's ultimate plan to control your personal hardware. Can't have that, can we. Microsoft's got the only KEK pre-installed on every UEFI motherboard, and they do because the industry let that happen.
Eventually all hardware will be like this. What will be your solution then? Don't buy a computer?
No, all hardware will not eventually all be like this, simply because it's not a function of the hardware at all. It's the software -- specifically, the BIOS. Not everyone on this list can necessarily design their own PC motherboard (I can -- not a huge deal, but certainly not cost effective against a company making 100 million PCBs a year), but I'll wager there are more than a few, just reading this, who could modify an open source BIOS (OpenBIOS, CoreBoot, etc), bust out a JTAG programmer, and re-flash a locked-down PC motherboard or a well documented standard ARM motherboard, to allow easy loading of other operating systems.
Of course, that re-flashed system won't run any version of Windows that requires a locked BIOS. But if you needed that version of Windows, you wouldn't have done the reflashing. And with Windows on the fast track to not be a useful desktop OS in a generation or two, it may not even be a huge concern.
And given how easy this is, given that it doesn't change the hardware one iota, the market isn't vanishing. It may shrink, sure, if most users are happy with locked-down tablet devices. There are healthy companies making various motherboards for industrial and mobile computers only selling 25K-50K per year, but doing so profitably. The market for real PCs is much, much larger than that, and will still be in 20 years.
Microsoft actually did float this same basic effect, if perhaps via different tools, back in 2001. Their SecurePC was a super locked-down, Windows-only PC. That's been their goal for awhile now. The fear generator back then was piracy -- this was, after all, right after the whole Napster thing blew up. If you had a SecurePC, only Microsoft-approved things would work, you'd have DRM to the bones (or at least the BIOS), etc.
Funny thing was that the OEMs rejected it. And mostly, I think, because they saw Microsoft as being too powerful already, and didn't want to make them even more powerful by essentially giving them a permanent OS monopoly... there wasn't going to be a way to run any other OS on these systems, as I recall.
So it's here again, same basic idea, only rolled out component-wise and within the auspices of industry support -- after all, the UEFI BIOS is an industry standard -- they just screwed the pooch on the handling of keys. Or left that to the Microsoft people on the UEFI Forum. In the UEFI specification, they allow only a single platform key (PK) installed on a system at any given time, and you, the hardware owner, was expected to be in charge of installing your platform key of choice. The PK is used to generate KEKs (key exchange keys), which then authenticate various bits installed on the PC. When you buy a Windows 8 PC, Microsoft's KEK is pre-installed. The problem is that in practice, the user has been removed from this process. The OEM installs the PK and the Microsoft KEK. They could also pre-install KEKs for Ubuntu or Red Hat or Haiku or AROS... but they don't. And, contrary to the original intent of the UEFI Form, you can't, either. In short, the whole thing is broken. UEFI was never intended, at least in theory, to protect you from your own legit use of your purchased hardware. It was intended to ensure nothing evil got control of your PC at boot time, at least not without your explicit permission (presumably, installing a KEK would be a fairly simple but intentional process... I suspect you'd go to the Ubuntu site, download their key onto a USB drive, reboot into the BIOS, go to the key management tab, enter the key and generate your KEK for that OS. Then you could install it under Ubuntu's key, no need for Microsoft... and no way for a malware application to force you into doing this, either. But that wouldn't work to block non-MS-OSs, so that's not the way Microsoft does it.
In fact, Microsoft is in effect reverse-subsidizing the Surface systems. They're charging very high prices... the Surface RT is inferior to the latest iPad and many recent Android tablets. The purpose here is simple: Microsoft's overcharging the customer will potentially give them Apple-like margins, and it will definitely let them charge OEMs the $95 or so they have to pay for Windows RT + Office, yet still be price-competitive with Microsoft... if not Apple or the Androids.
To see this in action, look at Asus. They have a successful line of Android tablets, the "Transformer" series. I have a TF700, which has a 1920x1200 screen, a nVidia Tegra T33 processor (1.7GHz), DDR3/1600 RAM.. this ran me around $499 with 64GB Flash. They also make the TF300, which has a 1280x800 screen, Tegra T30 processor (1.2GHz), DDR3/1333, and runs around $350 with 32GB Flash. The TF600, Asus' Windows RT version of the "Transformer", comes with the standard Windows tablet resolution of 1366x768 (16:9 for Windows, 16:10 for most Android tablets), the same Tegra T30 processor, but at 1.3GHz, the same DDR3/1333 memory, but 2GB worth (Windows am hungry for RAM), and 32GB for $499.... though only about half of that Flash is available. That's the "Windows Tax" at work... a slightly upgraded TF300 for $150 more, basically at the price of the TF700 with twice the Flash storage.
So that's what Microsoft is "subsidizing" here. They have no per-unit costs other than patent licensing, so they could go head to head with Apple and Android, but they're leaving room for the OEMs, at least right now. Which, of course, makes any Windows RT system a horrible value compared to Android devices, even if you neglect the fact you probably can't do much of anything with it. There's little software... it does run all those juicy Windows 7 Phone apps. Right. And sure, if you need MS-Office, but I suspect anyone really working in tech around here long ago gave up Office -- the last company I worked at demanding Office was in the mid-1990s. Business folk don't want Windows RT either, since they can't join Windows domains. These are for unsuspecting consumers.
Actually, no... it's a general purpose machine with a bootloader (BIOS) designed to not load anything (even another bootloader) not signed with a Microsoft key. The problem isn't the hardware, it's the software in flash in the machine -- Microsoft's proprietary UEFI BIOS. Replace that with something else, and you'll be able to use the machine for other purposes.
But don't do that. These need to fail in the marketplace. And people who understand the difference need to support companies that support open boot, out of the box.
It's not. It's an $800-$900 laptop missing its keyboard (available as an option), running a processor from a $500 laptop, and shipping with too little RAM or SSD space to be a very effective Windows machine.
Surface is a product for the Microsoft fanboi, no one else. Linux people just on principle shouldn't buy Surface computers, even if they get to the point where they run Linux just dandy. I mean, you really want to be feeding the beast on hardware, too? Better to buy a device from a Linux-friendly company.
Intel is going to SOLDERED ON CHIPS so frankly? Too late.
They did, a long time ago, in laptops and other smaller devices. They have already announced that no, they're not going to ONLY soldered-on chips. But that's not even the issue -- the CPU doesn't contain the BIOS. It's the BIOS -- a chunk of software that is, in this case, provided by Microsoft, that's the problem. Folks are asking Microsoft to allow this chunk of software to load other OSs, because in the past, the BIOS did that. Of course, in the past, the BIOS was an independent thing, not a Microsoft thing, so that was rather natural.
But the on-board Flash was soldered on long ago.. that's not a new thing. You might need a hack to re-program this in software, particularly if protected boot sectors are used, but in hardware, because these things are soldered on, there's usually some additional interface, such as a JTAG port, used at production time for programming. Or available for that -- some large volume companies may order the flash chips pre-programmed, but they can still be overwritten.
Also if you don't like UEFI? Then DO NOT BUY INTEL, its called voting with your wallet, is that REALLY so hard? Everyone forget that AMD embraced FOSS and was switching to Coreboot (probably have on the FM series but I'm still building Am3+ ) so again just DON'T BUY IT!
Absolutely agree... if you don't like a thing, don't buy it. When you buy something, that's a vote to support that thing, far as the manufacturer is concerned. When they have warehouses full of unsold goods, that's when they start reconsidering bad decisions. As techies, we have an obligation to not simply not buy unacceptable products, but tell all those people who ask for technical advice not to buy those things as well.
Aren't other companies building the Surface for Microsoft? I mean, they don't have an ARM plant tucked away somewhere I don't know about... So, what you're buying is MS branded hardware...
They're doing the same basic thing they did on the X-Box, same thing Apple does on the iOS devices: they design the Surface devices, which are then built by a contract manufacturer. This is different than, say, Google and the Nexus devices: Google contributes design input, but the company they work with is doing the whole design and manufacture. Google is an OEM in this scenario, Microsoft is the actual developer, using a CM to actually build the device.
Not unusual.. in fact, at latest count, Hon Hai Precision Industry Co. (aka Foxconn) makes about 40% of all the electronics made on the planet. Similar things happen in silicon at places like TSMC and Global Semiconductor: AMD, Broadcomm, Qualcomm, nVidia, Xilinx, Marvell, etc... these guys don't have their own chip fabs, they do the chip design, but have some other company make the chips (Global was actually spun out of AMD's former in-house manufacturing capability).
The key in all this, which is different than it was 20 years ago, is that ALL these contract companies do are contracts. If I put a personal computer together back in the 1990s, I could hire a CM to build the PCB, but it was probably someone like Motorola -- a company that built a great assembly plant for their own stuff, then decided to sell off excess capacity to other companies. When I worked at Commodore, we had HP make some of our chips... other companies used IBM. Today, most large semiconductor companies do sell out excess capacities, probably some product-level companies, too (we also had one product, the original Amiga 1000, produced outside the company, at Sanyo)... but the bulk of this is done today by companies that just do contract work -- they have no products of their own. And that's why they're the guys you choose -- no conflict of interest.
And that's essential to their business model. Reportedly, Apple was ready to spend an extra billion to buy all of TSMC's 28nm output. They were told "no"... which is why we're reading rumors of Apple making a deal to have their ARM SOCs made at Intel (they're currently made by Samsung's in-house chip fabs, but relations between Apple and Samsung haven't been stellar, thanks to Samsung taking over much of the mobile market that Apple thinks belongs to them).
The main reason is that, at least for the US (not sure about Europe), Microsoft's legally pronounced Monopoly position is limited to x86 personal computers. So at present, that's not extended to ARM. That's the way the judge wrote it.
Sure, it's easy to claim this is applicable to any personal computer. An Android phone or iOS tablet is just as much a personal computer as an x86 PC. There are x86 tablets that are essentially indistinguishable from ARM tablets. Motorola's got an x86 version of the RAZR Smartphone, and there are a couple of others. ARM chips are in laptops (the latest Samsung Chromebook, various others more popular in the Far East) and will certainly be in desktops again at some point. So sure, it's an artificial distinction... then again, Law IS artifice.
Once the courts agree that x86 and ARM are the same market, though, then it's much harder to argue for Microsoft's continued monopoly. Once you add up all the Androids and i-things, Microsoft doesn't even have 70% of the market -- around the point at which you first get accused of being a monopoly. And their share is dropping, fast, at least in numbers (in reality, sure, x86 devices are kept much longer than ARM devices, so ARM isn't building the installed base at the rate sales would suggest. Yet). At the rate they're going, ARM personal computers will outsell x86 personal computers in a couple of years, if not sooner. Maybe Intel and AMD will introduce lower power devices to combat this, but realistically... nVidia's selling Tegras starting at $15. Does anyone here really think Intel's going to be happy selling $15 SOCs, even if they are price/performance competitive, even in 22nm or 14nm processes on 450mm wafers (which TSMC won't have until 2018 or so)?
It's your hardware. You bought it with Microsoft's software, which includes their UEFI bootloader. You knew when you bought it that this loader was not acceptable for your uses, so your only real choice is to replace the bootloader -- not expect it to do other things. There's bound to be a JTAG port on that Tegra3 that would let you at the least wipe the Flash and install your own bootloader on your hardware... but most folks are expecting Microsoft to modify their software to do what you want your hardware to do. They aren't going to do that unless something forces them to (the law, consumer pressure, easing their karmic burden, etc).
You can make all the mods to your house you want, but if you want to put a 20x10 picture window in place of a critical load-bearing wall, you're going to have problems. Better to have bought a house -- or computer -- closer to what you wanted from the start. And don't support products that don't meet you needs. Period. That only leads to the industry producing more products that don't meet your needs.
Microsoft knows well from experience that, by the time anti-trust regulations catch up with them, the damage is pretty much done. And the punishment is little more than a stiff wack across the knuckles. They're not openly defying any existing court orders or regulations, but they're doing everything but.
The problem that Microsoft's been facing for decades now is the fact that Linux is free. You can't under-price free, and you can't, in the current Intel architecture, make a suitable "Windows only" system anymore. (There are exceptions, and some driver support sucks, but for the most part, it's not like it was in the heyday of Microsoft's OS hot war against everyone else.)
It is stupidly trivial for Microsoft to create a standard for "Windows only" systems. It is trivial today, it was trivial ten years ago, it was trivial ten years before that. They didn't.
They tried. Microsoft has been trying to close Windows systems for nearly two decades. The last time they tried (2001's "Secure PC" -- the excuse that time was media piracy, the goal exactly the same as now: a Windows-only PC), the OEMs rejected the closed PC. Not out of any particularly good behavior themselves, mostly because they felt Microsoft was already too powerful, and allow that would make them even more so.
This time around, Microsoft's doing it in stages, and making their own hardware, so that OEM participation isn't necessary. Same goal: they own the whole PC. Not the OEM. Not the consumer.
Right now, the monopoly decree around Microsoft is strictly limited to x86 systems -- that's how the judge wrote it. That was a stronger argument than probably necessary, given that the only significant competition was Apple, on PowerPC, at 2% of the desktop market at the time. But that's the argument.
So you have to re-argue for their being a monopoly again. If ARM tablets/phone personal computers are not the same market as x86 personal computers, then there's really no argument here: Microsoft has less than 1% of the tablet and less than 3% of the phone market. If we argue that all these devices are personal computers of effectively the same nature, then Microsoft doesn't even have a 70% share of that whole market, once you start counting Androids and iOS devices. And their rate of growth is stagnant -- it's easy to believe, if nothing fundamentally changes, that ARM personal computing devices overtake x86, at least in volume shipments, in a 1-3 years.
So the one left is the one they used against Microsoft and Netscape -- it's also illegal for a proven monopoly to abuse their monopoly powers in a non-monopoly market in order to take over that market. That's what Microsoft did when they tried to take over the web and kill off Netscape. Oh wait, when they did kill Netscape and nearly took over the web.. then got their pee-pee wacked over this naughty behavior, and had to promise not to do it again (and make it easier to run other web browsers as first-class). But unless someone makes that effective argument over the tablet stuff, in court, Microsoft is free to use their whole dirty-trick playbook in the ARM world. And they do seem intent on doing just that: can't load anything but Windows RT, probably can't even buy an RT upgrade for the system you buy today, can't buy an RT system from anyone without also buying Office, etc. Most of these dirty tricks aren't even Surface-specific.
The Surface RT tablet is a horrible value for the money. You pay less for a better made Asus Android tablet with twice the Flash memory, better screen, better keyboard option, etc... and a functional bootloader.
But for hacking these, I wonder why anyone bothers with the bootloader anyway. There's bound to be a JTAG or similar programmer for production-level programming. Ok, sure, I suppose they could all buy pre-programmed Flash chips, but given Microsoft's "creative" software schedules, I bet they don't. Just install your own BIOS... except, even if that's possible, don't. That's even more reverse engineering of the system (well, ok, not that much, given that the Surface is bound to be pretty identical to any other Tegra3 computer at that level), but why send a dime to a company trying to lock the PC down at this level?
Not to mention FTL "subspace" communications... our real devices are limited by the speed of light, obstacles (the relative position of the Enterprise versus on-planet communicators never seemed to be an issue... your body can drop a cellphone signal by 10-20dB), as well as our lack of Star Trek's advanced power sources -- anything capable of powering a tiny Phaser-1 wouldn't be hard pressed to deliver on much more range and battery life in a smartphone/communicator. Sure, the pocket device evolved in dozens of ways not imagined by Roddenberry and crew... that's how things happen in the real world.
Much simpler math: price an iPod Touch at retail. Add the extra $25-$35 worth of component cost to make that into an iPhone, and add whatever markup you'd like on top of that. That'll get you the rational MSRP Apple would charging in a world that didn't have phone "subsidies". The difference between that and the $650-$800 or whatever the MSRP actually is for your model-of-choice, Apple or Samsung or anyone else, is the fake pricing imposed by the Telco industry.
ALL on-contract phones -- even the dumb ones -- have a completely fabricated MSRP. Those cheap-ass Nokia-style candybar dumb phones can be made for about $25-$30 today, complete... but the MSRP is usually around $150. A generous MSRP in a competitive market would be $50-$60.
The iPhone is a pretty clear example, given how similar it's been over the course of its evolution to the iPod Touch. At any given technology node, the iPhone has maybe an upgraded processor or screen (we don't know what Apple pays for a processor, but the nVidia Tegra 3 reportedly runs around $15-$20 in volume, so this is probably a $5 actual difference), often actually less Flash, and of course, a cellular modem. That's less than $100 difference at retail, and yet, the iPhone usually runs 2x-3x the price of the iPod Touch.
This isn't Apple setting the price per se -- Apple doesn't really control the pricing of their phones anymore than Samsung or Motorola or... well, anyone but Google right now. That's because Apple's market DEPENDS on selling through the major Telcos, and that means setting a price that will enable their distribution through Telcos. All Telcos are negotiating a volume price for the phones they redistribute based on a discount from the MSRP. They do this, absolutely, to keep the "subsidized price" myth going strong, to keep consumers thinking they're getting a great deal before signing that two-year contract. If the manufacturer cuts the price, the Telcos simply won't offer that model. Apple has no real choice in this, anymore than Samsung or any of the others. Sure, Apple could sell an iPhone at $350 or so, but the price Verizon or AT&T would offer per unit, for millions of devices, would be below cost. No way to make Apple-style profits. And even Apple isn't going to buck that system just yet; particularly in North America, most consumers buy their phones through the Telcos.
The one company actually bucking the system is Google... and it took them a few generations to get there. The original Nexus One was supposed to change the whole cellular model by going direct, but it didn't -- Google's $575 price was in the same ballpark as Motorola's direct $600 MSRP on the Droid 1, or the price paid in other countries for the iPhone of the day (not sold unbundled in the USA back then). The reason became clear when they started offering the Nexus One through T-Mobile and AT&T. The other issue: those were different models of the same phone, they didn't have a single device that would work even across all GSM carriers, even just in the USA -- the other critical thing you need to sell an unbundled phone.
But fast forward to modern times, and they did actually make some rational changes. When the Galaxy Nexus went on sale at Verizon, it was $299 + 2-year contract, and had an MSRP around $650. Some months later, Google introduced a direct-sale model, GSM-only, that worked fully on both AT&T and T-Mobile, for $349. That's a fairly reasonable MSRP, at least compared to other smartphone-like device such as the iPod Touch or the various smaller Android tablets. And this continued forward with the Nexus 4, which starts at $299. But it's not available from any carrier -- carriers aren't going to bless the loss of subsidized phones unless it's to their competitive advantage.
Reportedly, T-Mobile is going this way, but by their choice, as a way to differentiate themselves from Sprint, AT&T, and Verizon... particularly given their current lack of real 4G services. This ought to be good news for Google's phone sale model, and may push others in a similar direction. Unfortunately, given that Verizon and Sprint have locked networks rather than locked phones, no one's going to introduce a combination GSM/CDMA2000 + LTE model, despite the fact that, given modern cellular modem chips, it's quite possible to do.
For Windows 8 vs Windows RT, maybe the tie goes to Intel. On Android, the tie certainly goes to ARM. But keep in mind, you're comparing a quad core ARM to a dual core x86, and this isn't even the best ARM for comparison anymore.... plus, the Surface doesn't even use the faster T33 verson. The Atom in question doesn't have a CPU speed advantage, but it has huge memory bus advantage over the Tegra 3: nVidia's single 32-bit DDR3 bus versus Intel's dual 64-bit DDR3 bus. A comparison o the Nexus 10 might be more meaningful, or a better Tegra 3 device, like the Asus TF700.
And sure, a Windows 8 system is more useful than a Windows RT system, given the app support. And yet, that 2-core Atom on 32GB of Flash (half available) with only 2GB RAM and 32-bit mode is a bottom-level device for any real Windows apps. The quad core Tegra 3 at 1.7GHz with DDR3-1600 or so is a pretty top of the line Android device -- the experience will be far superior.
That's actually part of the Open Handset Alliance agreement, and has been from the get-go. Carriers can change things to an extent, but can't mess with the APIs... all Android devices are supposed to be compatible at the app level. And they largely are -- the fragmentation thing has been blown way out of proportion, mostly by Apple fans as they ran out of other arguments as to why their iOS wasn't better than Android.
There are two problems Google needs to address. One is the initial OS in a device: you have had developers releasing devices on 2.2 even, well after Android 4.0 was out. Google needs to address that.
The second problem is getting the new OS out in the first place. An ordinary development model for an OS will have early releases available to developers long before the new version ships. This gets them an early start on porting and testing, etc. Google's current M.O. is to select one vendor and one device to work on (usually a Nexus device these days), then work intensely with that partner. The new OS version isn't released to other OEMs until that new device ships. This is a big delay in getting the new OS adopted. And it results in far less testing than would otherwise take place. Maybe this is needed for Google's two-release-per-year schedule to be kept, but that, too, is part of the reason new devices don't always have new OSs.
There are a few things Google could do. Ideally, they could re-engineer the basis of Android, and build a hardware abstraction layer under Linux. Android/Linux would have class-drivers (display, touchscreen, keyboard, etc) that hit the vendor-supplies HAL layer. The HAL layer would contain all hardware dependencies, cell phone baseband, etc. This would basically allow any new version of Android to run on any device without the need for the manufacturer or cellular provider (argh!) to be involved. In short, just what PCs do.
That's true, and it's pretty much guaranteed. When you're winning, you don't innovate. Innovation, after all, is nothing but risk judged over time as successful.
Microsoft didn't innovate for years. They didn't have to -- they could copy the better things being done by other companies, deliver incremental improvements and changes, and keep their empire pretty much intact. They stumbled on Vista, badly... that was the start of their fall from being "The Most Valueable Tech Company On the Planet". Vista was stupid risk -- it was trying to deliver a bunch of strategic things for Microsoft, without much concern for the end-user. Not innovation, because [a] it failed and [b] it just wasn't that interesting.
Apple's doing that now... steady but very conservative improvements. Why not? They have a model that, even if they don't dominate the mobile world quite as much as a couple of years ago, they make the most money of any mobile computing company. As Microsoft was, they're now on top in that segment. It's much easier to imagine falling than climbing much higher.. so they don't screw with the formula. Apple's actually a worst-case, because they only release a model or two in each product line every year (ok, actually three iPads this year, but the iPad 4 was an apology for the iPad 3). So if they gamble and win, wow, that's a great innovation. If they gamble and lose... no money, and maybe they lose a bunch of seemingly unloseable customers to Android or, hell, even Windows Phone/RT.
Short answer: Apple's way too successful to innovate, and they don't really have a corporate culture anymore that allows innovation. The guys still jockeying for position in this industry, that's where you find innovation. So Samsung's winning on phones, and their SIII is basically just bigger-better-faster-more over the SII, no surprises. But they were only so-so on tablets. So they introduced, shocking the world apparently, a 5.2" tablet.. and create the "Phablet" class. Not just that, but the Galaxy Note has a Wacom digitizer in it -- they brought back the stylus, but one that doesn't suck. Now that's in their higher-end 10" tablet, too. That's innovation -- they took a risk, it was well received. Microsoft's copying this in their Surface Pro, so it'll actually be possible to run real Windows applications without a mouse. One sign of real innovation: someone's stealing the idea.
Look at Motorola... nearly dead, they built what was basically the anti-iPhone, in the original "Droid" (aka Milestone), and delivered the first significant Android product. Innovation, because it did well. They did the same thing recently with the revamped RAZR series... a smartphone that actually survives typical smartphone use without a case? They're also doing what companies that will be seen as innovators do: lots of experiments. In the time that Apple went from iPhone 4S to iPhone 5, Motorola fielded five different RAZRs, not to mention a bunch of other smartphones. That's not the way to get to Apple-like profitability any time soon, but it's also the way to, by brute force and refinement, figure out what you can do that Apple, HTC, or Samsung can't.
Microsoft, of course, has done some crazy shit lately.... because they're scared. Sure, they still dominate, still have most of the desktop PC market. But that's actually the wrong question to ask, given the rise of mobile. They have 69% of personal computing, once you factor in mobile.. and that's an older number. It's actually possible that Microsoft will fall to under 50% in the next year or two. Now, sure, any other company would be happy to have that share, but Microsoft's whole business has been built on the fact they can bully pretty much anyone they want in the PC industry. They've made some real enemies, and won't be the same Microsoft without that being their main superpower.
So Apple's floated this Metro/WinRT thing, a tablet version of Windows that's more or less the same thing (with some mysterious compatibility matrix) between different devices. They've basically th
Intel acquired DEC's StrongARM in a law suit settlement with DEC over CPU patents (Intel was using some of the Alpha designs), and that eventually became X-Scale. Which they sold to Marvell some years back. Both Marvell and Intel retain ARM architecture licenses, though Intel's clearly not building any new ARM chips.
Intel's been doing some fab work for a few very small companies. I don't think they're offering anything even close to what Apple needs. Not to mention the old "out of the frying pan, into the fire" problem that would cause -- Intel's hell bent on making x86 compete on price, performance, and power against ARM... and it looks like Microsoft is their biggest booster (they're never getting iOS business, they're still a second fiddle processor on Android, though most apps work -- but on Windows, they have the advantage).
Apple apparently offered TSMC a $1 billion bonus for their entire 28nm capacity... they were turned down. So it may well be they're having or had some issues finding capacity -- they may even be bankrolling some of it. Apple's not looking for average capacity, either, but something fairly cutting edge. Not that their designs are unusually challenging, they're pretty standard. But smaller dice equals higher profits and lower power -- it's important in CPUs in general, critical for Apple's mobile existence.
In fact... Samsung has used a large number of competitors' chips in their own devices. That's almost certainly due to clearing foundry space for Apple. Once Apple's out of the way, you may never see a Qualcomm or TI chip in an Samsung product again... unless they have a better offer. Samsung may also get the competitor's chips cheaper than others, not just due to volume, but perhaps foundry exchange deals -- Qualcomm is fabless, TI not entirely but they lack on the digital side of things. Both usually use TSMC, but not exclusively.
Slashdot Mirror
Take the long term view. If mankind goes extinct, then absolutely nothing we every worried about in the short term matters one iota on that day after the last human dies. Right now, and as long as we're only on Earth, any number of catastrophies could kill us all in sort order, some we create, some that just happen. Either way, mankind and every thing it ever did ceases to matter at all.
Or, we keep working to fix this ultimate problem. Taking the million year view, moving sustainably beyond earth is the most important thing humanity will every do.
If you're paying a signer, that means they're signing your image with their private key so it works with their KEK (Key Exchange Key), presumably already in Flash. But that's not the correct process.
Rather, the original intention was for the user to be in charge of keys. You can go with the PK (platform key, which is the root key) supplied by your board manufacturer, but you're supposed to have the ability to replace it (and of course, clear out any KEKs) if you like. And you're supposed to be able to install your own KEK.
So for example, you compile up your own Linux kernel to use with Slackware or whatever. Long ago, you generated your own private key for signing your personal Linux binaries, and generated a KEK for your motherboard for that private key, which lives in the signature database right next to the Microsoft KEK and any others you or your motherboard manufacturer might have installed. So you sign the new binary, and it just loads (actually, the signature database can deal with both signed and unsigned binaries -- if you image isn't signed, the OS can be securely installed by storing a SHA-256 hash of the boot image in the signature database). No need to involve Microsoft, no need to risk boot from a virus.
But of course, making it this simple gets around Microsoft's ultimate plan to control your personal hardware. Can't have that, can we. Microsoft's got the only KEK pre-installed on every UEFI motherboard, and they do because the industry let that happen.
Eventually all hardware will be like this. What will be your solution then? Don't buy a computer?
No, all hardware will not eventually all be like this, simply because it's not a function of the hardware at all. It's the software -- specifically, the BIOS. Not everyone on this list can necessarily design their own PC motherboard (I can -- not a huge deal, but certainly not cost effective against a company making 100 million PCBs a year), but I'll wager there are more than a few, just reading this, who could modify an open source BIOS (OpenBIOS, CoreBoot, etc), bust out a JTAG programmer, and re-flash a locked-down PC motherboard or a well documented standard ARM motherboard, to allow easy loading of other operating systems.
Of course, that re-flashed system won't run any version of Windows that requires a locked BIOS. But if you needed that version of Windows, you wouldn't have done the reflashing. And with Windows on the fast track to not be a useful desktop OS in a generation or two, it may not even be a huge concern.
And given how easy this is, given that it doesn't change the hardware one iota, the market isn't vanishing. It may shrink, sure, if most users are happy with locked-down tablet devices. There are healthy companies making various motherboards for industrial and mobile computers only selling 25K-50K per year, but doing so profitably. The market for real PCs is much, much larger than that, and will still be in 20 years.
Microsoft actually did float this same basic effect, if perhaps via different tools, back in 2001. Their SecurePC was a super locked-down, Windows-only PC. That's been their goal for awhile now. The fear generator back then was piracy -- this was, after all, right after the whole Napster thing blew up. If you had a SecurePC, only Microsoft-approved things would work, you'd have DRM to the bones (or at least the BIOS), etc.
Funny thing was that the OEMs rejected it. And mostly, I think, because they saw Microsoft as being too powerful already, and didn't want to make them even more powerful by essentially giving them a permanent OS monopoly... there wasn't going to be a way to run any other OS on these systems, as I recall.
So it's here again, same basic idea, only rolled out component-wise and within the auspices of industry support -- after all, the UEFI BIOS is an industry standard -- they just screwed the pooch on the handling of keys. Or left that to the Microsoft people on the UEFI Forum. In the UEFI specification, they allow only a single platform key (PK) installed on a system at any given time, and you, the hardware owner, was expected to be in charge of installing your platform key of choice. The PK is used to generate KEKs (key exchange keys), which then authenticate various bits installed on the PC. When you buy a Windows 8 PC, Microsoft's KEK is pre-installed. The problem is that in practice, the user has been removed from this process. The OEM installs the PK and the Microsoft KEK. They could also pre-install KEKs for Ubuntu or Red Hat or Haiku or AROS... but they don't. And, contrary to the original intent of the UEFI Form, you can't, either. In short, the whole thing is broken. UEFI was never intended, at least in theory, to protect you from your own legit use of your purchased hardware. It was intended to ensure nothing evil got control of your PC at boot time, at least not without your explicit permission (presumably, installing a KEK would be a fairly simple but intentional process... I suspect you'd go to the Ubuntu site, download their key onto a USB drive, reboot into the BIOS, go to the key management tab, enter the key and generate your KEK for that OS. Then you could install it under Ubuntu's key, no need for Microsoft... and no way for a malware application to force you into doing this, either. But that wouldn't work to block non-MS-OSs, so that's not the way Microsoft does it.
In fact, Microsoft is in effect reverse-subsidizing the Surface systems. They're charging very high prices... the Surface RT is inferior to the latest iPad and many recent Android tablets. The purpose here is simple: Microsoft's overcharging the customer will potentially give them Apple-like margins, and it will definitely let them charge OEMs the $95 or so they have to pay for Windows RT + Office, yet still be price-competitive with Microsoft... if not Apple or the Androids.
To see this in action, look at Asus. They have a successful line of Android tablets, the "Transformer" series. I have a TF700, which has a 1920x1200 screen, a nVidia Tegra T33 processor (1.7GHz), DDR3/1600 RAM.. this ran me around $499 with 64GB Flash. They also make the TF300, which has a 1280x800 screen, Tegra T30 processor (1.2GHz), DDR3/1333, and runs around $350 with 32GB Flash. The TF600, Asus' Windows RT version of the "Transformer", comes with the standard Windows tablet resolution of 1366x768 (16:9 for Windows, 16:10 for most Android tablets), the same Tegra T30 processor, but at 1.3GHz, the same DDR3/1333 memory, but 2GB worth (Windows am hungry for RAM), and 32GB for $499.... though only about half of that Flash is available. That's the "Windows Tax" at work... a slightly upgraded TF300 for $150 more, basically at the price of the TF700 with twice the Flash storage.
So that's what Microsoft is "subsidizing" here. They have no per-unit costs other than patent licensing, so they could go head to head with Apple and Android, but they're leaving room for the OEMs, at least right now. Which, of course, makes any Windows RT system a horrible value compared to Android devices, even if you neglect the fact you probably can't do much of anything with it. There's little software... it does run all those juicy Windows 7 Phone apps. Right. And sure, if you need MS-Office, but I suspect anyone really working in tech around here long ago gave up Office -- the last company I worked at demanding Office was in the mid-1990s. Business folk don't want Windows RT either, since they can't join Windows domains. These are for unsuspecting consumers.
It's not the hardware. It's a bog standard nVidia Tegra 3 platform, with software (the BIOS) that demands a proprietary key.
Actually, no... it's a general purpose machine with a bootloader (BIOS) designed to not load anything (even another bootloader) not signed with a Microsoft key. The problem isn't the hardware, it's the software in flash in the machine -- Microsoft's proprietary UEFI BIOS. Replace that with something else, and you'll be able to use the machine for other purposes.
But don't do that. These need to fail in the marketplace. And people who understand the difference need to support companies that support open boot, out of the box.
It's not. It's an $800-$900 laptop missing its keyboard (available as an option), running a processor from a $500 laptop, and shipping with too little RAM or SSD space to be a very effective Windows machine.
Surface is a product for the Microsoft fanboi, no one else. Linux people just on principle shouldn't buy Surface computers, even if they get to the point where they run Linux just dandy. I mean, you really want to be feeding the beast on hardware, too? Better to buy a device from a Linux-friendly company.
Intel is going to SOLDERED ON CHIPS so frankly? Too late.
They did, a long time ago, in laptops and other smaller devices. They have already announced that no, they're not going to ONLY soldered-on chips. But that's not even the issue -- the CPU doesn't contain the BIOS. It's the BIOS -- a chunk of software that is, in this case, provided by Microsoft, that's the problem. Folks are asking Microsoft to allow this chunk of software to load other OSs, because in the past, the BIOS did that. Of course, in the past, the BIOS was an independent thing, not a Microsoft thing, so that was rather natural.
But the on-board Flash was soldered on long ago.. that's not a new thing. You might need a hack to re-program this in software, particularly if protected boot sectors are used, but in hardware, because these things are soldered on, there's usually some additional interface, such as a JTAG port, used at production time for programming. Or available for that -- some large volume companies may order the flash chips pre-programmed, but they can still be overwritten.
Also if you don't like UEFI? Then DO NOT BUY INTEL, its called voting with your wallet, is that REALLY so hard? Everyone forget that AMD embraced FOSS and was switching to Coreboot (probably have on the FM series but I'm still building Am3+ ) so again just DON'T BUY IT!
Absolutely agree... if you don't like a thing, don't buy it. When you buy something, that's a vote to support that thing, far as the manufacturer is concerned. When they have warehouses full of unsold goods, that's when they start reconsidering bad decisions. As techies, we have an obligation to not simply not buy unacceptable products, but tell all those people who ask for technical advice not to buy those things as well.
Aren't other companies building the Surface for Microsoft? I mean, they don't have an ARM plant tucked away somewhere I don't know about... So, what you're buying is MS branded hardware...
They're doing the same basic thing they did on the X-Box, same thing Apple does on the iOS devices: they design the Surface devices, which are then built by a contract manufacturer. This is different than, say, Google and the Nexus devices: Google contributes design input, but the company they work with is doing the whole design and manufacture. Google is an OEM in this scenario, Microsoft is the actual developer, using a CM to actually build the device.
Not unusual.. in fact, at latest count, Hon Hai Precision Industry Co. (aka Foxconn) makes about 40% of all the electronics made on the planet. Similar things happen in silicon at places like TSMC and Global Semiconductor: AMD, Broadcomm, Qualcomm, nVidia, Xilinx, Marvell, etc... these guys don't have their own chip fabs, they do the chip design, but have some other company make the chips (Global was actually spun out of AMD's former in-house manufacturing capability).
The key in all this, which is different than it was 20 years ago, is that ALL these contract companies do are contracts. If I put a personal computer together back in the 1990s, I could hire a CM to build the PCB, but it was probably someone like Motorola -- a company that built a great assembly plant for their own stuff, then decided to sell off excess capacity to other companies. When I worked at Commodore, we had HP make some of our chips... other companies used IBM. Today, most large semiconductor companies do sell out excess capacities, probably some product-level companies, too (we also had one product, the original Amiga 1000, produced outside the company, at Sanyo)... but the bulk of this is done today by companies that just do contract work -- they have no products of their own. And that's why they're the guys you choose -- no conflict of interest.
And that's essential to their business model. Reportedly, Apple was ready to spend an extra billion to buy all of TSMC's 28nm output. They were told "no"... which is why we're reading rumors of Apple making a deal to have their ARM SOCs made at Intel (they're currently made by Samsung's in-house chip fabs, but relations between Apple and Samsung haven't been stellar, thanks to Samsung taking over much of the mobile market that Apple thinks belongs to them).
The main reason is that, at least for the US (not sure about Europe), Microsoft's legally pronounced Monopoly position is limited to x86 personal computers. So at present, that's not extended to ARM. That's the way the judge wrote it.
Sure, it's easy to claim this is applicable to any personal computer. An Android phone or iOS tablet is just as much a personal computer as an x86 PC. There are x86 tablets that are essentially indistinguishable from ARM tablets. Motorola's got an x86 version of the RAZR Smartphone, and there are a couple of others. ARM chips are in laptops (the latest Samsung Chromebook, various others more popular in the Far East) and will certainly be in desktops again at some point. So sure, it's an artificial distinction... then again, Law IS artifice.
Once the courts agree that x86 and ARM are the same market, though, then it's much harder to argue for Microsoft's continued monopoly. Once you add up all the Androids and i-things, Microsoft doesn't even have 70% of the market -- around the point at which you first get accused of being a monopoly. And their share is dropping, fast, at least in numbers (in reality, sure, x86 devices are kept much longer than ARM devices, so ARM isn't building the installed base at the rate sales would suggest. Yet). At the rate they're going, ARM personal computers will outsell x86 personal computers in a couple of years, if not sooner. Maybe Intel and AMD will introduce lower power devices to combat this, but realistically... nVidia's selling Tegras starting at $15. Does anyone here really think Intel's going to be happy selling $15 SOCs, even if they are price/performance competitive, even in 22nm or 14nm processes on 450mm wafers (which TSMC won't have until 2018 or so)?
It's your hardware. You bought it with Microsoft's software, which includes their UEFI bootloader. You knew when you bought it that this loader was not acceptable for your uses, so your only real choice is to replace the bootloader -- not expect it to do other things. There's bound to be a JTAG port on that Tegra3 that would let you at the least wipe the Flash and install your own bootloader on your hardware... but most folks are expecting Microsoft to modify their software to do what you want your hardware to do. They aren't going to do that unless something forces them to (the law, consumer pressure, easing their karmic burden, etc).
You can make all the mods to your house you want, but if you want to put a 20x10 picture window in place of a critical load-bearing wall, you're going to have problems. Better to have bought a house -- or computer -- closer to what you wanted from the start. And don't support products that don't meet you needs. Period. That only leads to the industry producing more products that don't meet your needs.
Microsoft knows well from experience that, by the time anti-trust regulations catch up with them, the damage is pretty much done. And the punishment is little more than a stiff wack across the knuckles. They're not openly defying any existing court orders or regulations, but they're doing everything but.
It is stupidly trivial for Microsoft to create a standard for "Windows only" systems. It is trivial today, it was trivial ten years ago, it was trivial ten years before that. They didn't.
They tried. Microsoft has been trying to close Windows systems for nearly two decades. The last time they tried (2001's "Secure PC" -- the excuse that time was media piracy, the goal exactly the same as now: a Windows-only PC), the OEMs rejected the closed PC. Not out of any particularly good behavior themselves, mostly because they felt Microsoft was already too powerful, and allow that would make them even more so.
This time around, Microsoft's doing it in stages, and making their own hardware, so that OEM participation isn't necessary. Same goal: they own the whole PC. Not the OEM. Not the consumer.
The problem is still making Microsoft a monopoly.
Right now, the monopoly decree around Microsoft is strictly limited to x86 systems -- that's how the judge wrote it. That was a stronger argument than probably necessary, given that the only significant competition was Apple, on PowerPC, at 2% of the desktop market at the time. But that's the argument.
So you have to re-argue for their being a monopoly again. If ARM tablets/phone personal computers are not the same market as x86 personal computers, then there's really no argument here: Microsoft has less than 1% of the tablet and less than 3% of the phone market. If we argue that all these devices are personal computers of effectively the same nature, then Microsoft doesn't even have a 70% share of that whole market, once you start counting Androids and iOS devices. And their rate of growth is stagnant -- it's easy to believe, if nothing fundamentally changes, that ARM personal computing devices overtake x86, at least in volume shipments, in a 1-3 years.
So the one left is the one they used against Microsoft and Netscape -- it's also illegal for a proven monopoly to abuse their monopoly powers in a non-monopoly market in order to take over that market. That's what Microsoft did when they tried to take over the web and kill off Netscape. Oh wait, when they did kill Netscape and nearly took over the web.. then got their pee-pee wacked over this naughty behavior, and had to promise not to do it again (and make it easier to run other web browsers as first-class). But unless someone makes that effective argument over the tablet stuff, in court, Microsoft is free to use their whole dirty-trick playbook in the ARM world. And they do seem intent on doing just that: can't load anything but Windows RT, probably can't even buy an RT upgrade for the system you buy today, can't buy an RT system from anyone without also buying Office, etc. Most of these dirty tricks aren't even Surface-specific.
The Surface RT tablet is a horrible value for the money. You pay less for a better made Asus Android tablet with twice the Flash memory, better screen, better keyboard option, etc... and a functional bootloader.
But for hacking these, I wonder why anyone bothers with the bootloader anyway. There's bound to be a JTAG or similar programmer for production-level programming. Ok, sure, I suppose they could all buy pre-programmed Flash chips, but given Microsoft's "creative" software schedules, I bet they don't. Just install your own BIOS... except, even if that's possible, don't. That's even more reverse engineering of the system (well, ok, not that much, given that the Surface is bound to be pretty identical to any other Tegra3 computer at that level), but why send a dime to a company trying to lock the PC down at this level?
Not to mention FTL "subspace" communications... our real devices are limited by the speed of light, obstacles (the relative position of the Enterprise versus on-planet communicators never seemed to be an issue... your body can drop a cellphone signal by 10-20dB), as well as our lack of Star Trek's advanced power sources -- anything capable of powering a tiny Phaser-1 wouldn't be hard pressed to deliver on much more range and battery life in a smartphone/communicator. Sure, the pocket device evolved in dozens of ways not imagined by Roddenberry and crew... that's how things happen in the real world.
Much simpler math: price an iPod Touch at retail. Add the extra $25-$35 worth of component cost to make that into an iPhone, and add whatever markup you'd like on top of that. That'll get you the rational MSRP Apple would charging in a world that didn't have phone "subsidies". The difference between that and the $650-$800 or whatever the MSRP actually is for your model-of-choice, Apple or Samsung or anyone else, is the fake pricing imposed by the Telco industry.
ALL on-contract phones -- even the dumb ones -- have a completely fabricated MSRP. Those cheap-ass Nokia-style candybar dumb phones can be made for about $25-$30 today, complete... but the MSRP is usually around $150. A generous MSRP in a competitive market would be $50-$60.
The iPhone is a pretty clear example, given how similar it's been over the course of its evolution to the iPod Touch. At any given technology node, the iPhone has maybe an upgraded processor or screen (we don't know what Apple pays for a processor, but the nVidia Tegra 3 reportedly runs around $15-$20 in volume, so this is probably a $5 actual difference), often actually less Flash, and of course, a cellular modem. That's less than $100 difference at retail, and yet, the iPhone usually runs 2x-3x the price of the iPod Touch.
This isn't Apple setting the price per se -- Apple doesn't really control the pricing of their phones anymore than Samsung or Motorola or ... well, anyone but Google right now. That's because Apple's market DEPENDS on selling through the major Telcos, and that means setting a price that will enable their distribution through Telcos. All Telcos are negotiating a volume price for the phones they redistribute based on a discount from the MSRP. They do this, absolutely, to keep the "subsidized price" myth going strong, to keep consumers thinking they're getting a great deal before signing that two-year contract. If the manufacturer cuts the price, the Telcos simply won't offer that model. Apple has no real choice in this, anymore than Samsung or any of the others. Sure, Apple could sell an iPhone at $350 or so, but the price Verizon or AT&T would offer per unit, for millions of devices, would be below cost. No way to make Apple-style profits. And even Apple isn't going to buck that system just yet; particularly in North America, most consumers buy their phones through the Telcos.
The one company actually bucking the system is Google... and it took them a few generations to get there. The original Nexus One was supposed to change the whole cellular model by going direct, but it didn't -- Google's $575 price was in the same ballpark as Motorola's direct $600 MSRP on the Droid 1, or the price paid in other countries for the iPhone of the day (not sold unbundled in the USA back then). The reason became clear when they started offering the Nexus One through T-Mobile and AT&T. The other issue: those were different models of the same phone, they didn't have a single device that would work even across all GSM carriers, even just in the USA -- the other critical thing you need to sell an unbundled phone.
But fast forward to modern times, and they did actually make some rational changes. When the Galaxy Nexus went on sale at Verizon, it was $299 + 2-year contract, and had an MSRP around $650. Some months later, Google introduced a direct-sale model, GSM-only, that worked fully on both AT&T and T-Mobile, for $349. That's a fairly reasonable MSRP, at least compared to other smartphone-like device such as the iPod Touch or the various smaller Android tablets. And this continued forward with the Nexus 4, which starts at $299. But it's not available from any carrier -- carriers aren't going to bless the loss of subsidized phones unless it's to their competitive advantage.
Reportedly, T-Mobile is going this way, but by their choice, as a way to differentiate themselves from Sprint, AT&T, and Verizon... particularly given their current lack of real 4G services. This ought to be good news for Google's phone sale model, and may push others in a similar direction. Unfortunately, given that Verizon and Sprint have locked networks rather than locked phones, no one's going to introduce a combination GSM/CDMA2000 + LTE model, despite the fact that, given modern cellular modem chips, it's quite possible to do.
For Windows 8 vs Windows RT, maybe the tie goes to Intel. On Android, the tie certainly goes to ARM. But keep in mind, you're comparing a quad core ARM to a dual core x86, and this isn't even the best ARM for comparison anymore.... plus, the Surface doesn't even use the faster T33 verson. The Atom in question doesn't have a CPU speed advantage, but it has huge memory bus advantage over the Tegra 3: nVidia's single 32-bit DDR3 bus versus Intel's dual 64-bit DDR3 bus. A comparison o the Nexus 10 might be more meaningful, or a better Tegra 3 device, like the Asus TF700.
And sure, a Windows 8 system is more useful than a Windows RT system, given the app support. And yet, that 2-core Atom on 32GB of Flash (half available) with only 2GB RAM and 32-bit mode is a bottom-level device for any real Windows apps. The quad core Tegra 3 at 1.7GHz with DDR3-1600 or so is a pretty top of the line Android device -- the experience will be far superior.
That's actually part of the Open Handset Alliance agreement, and has been from the get-go. Carriers can change things to an extent, but can't mess with the APIs... all Android devices are supposed to be compatible at the app level. And they largely are -- the fragmentation thing has been blown way out of proportion, mostly by Apple fans as they ran out of other arguments as to why their iOS wasn't better than Android.
There are two problems Google needs to address. One is the initial OS in a device: you have had developers releasing devices on 2.2 even, well after Android 4.0 was out. Google needs to address that.
The second problem is getting the new OS out in the first place. An ordinary development model for an OS will have early releases available to developers long before the new version ships. This gets them an early start on porting and testing, etc. Google's current M.O. is to select one vendor and one device to work on (usually a Nexus device these days), then work intensely with that partner. The new OS version isn't released to other OEMs until that new device ships. This is a big delay in getting the new OS adopted. And it results in far less testing than would otherwise take place. Maybe this is needed for Google's two-release-per-year schedule to be kept, but that, too, is part of the reason new devices don't always have new OSs.
There are a few things Google could do. Ideally, they could re-engineer the basis of Android, and build a hardware abstraction layer under Linux. Android/Linux would have class-drivers (display, touchscreen, keyboard, etc) that hit the vendor-supplies HAL layer. The HAL layer would contain all hardware dependencies, cell phone baseband, etc. This would basically allow any new version of Android to run on any device without the need for the manufacturer or cellular provider (argh!) to be involved. In short, just what PCs do.
That's true, and it's pretty much guaranteed. When you're winning, you don't innovate. Innovation, after all, is nothing but risk judged over time as successful.
Microsoft didn't innovate for years. They didn't have to -- they could copy the better things being done by other companies, deliver incremental improvements and changes, and keep their empire pretty much intact. They stumbled on Vista, badly... that was the start of their fall from being "The Most Valueable Tech Company On the Planet". Vista was stupid risk -- it was trying to deliver a bunch of strategic things for Microsoft, without much concern for the end-user. Not innovation, because [a] it failed and [b] it just wasn't that interesting.
Apple's doing that now... steady but very conservative improvements. Why not? They have a model that, even if they don't dominate the mobile world quite as much as a couple of years ago, they make the most money of any mobile computing company. As Microsoft was, they're now on top in that segment. It's much easier to imagine falling than climbing much higher.. so they don't screw with the formula. Apple's actually a worst-case, because they only release a model or two in each product line every year (ok, actually three iPads this year, but the iPad 4 was an apology for the iPad 3). So if they gamble and win, wow, that's a great innovation. If they gamble and lose... no money, and maybe they lose a bunch of seemingly unloseable customers to Android or, hell, even Windows Phone/RT.
Short answer: Apple's way too successful to innovate, and they don't really have a corporate culture anymore that allows innovation. The guys still jockeying for position in this industry, that's where you find innovation. So Samsung's winning on phones, and their SIII is basically just bigger-better-faster-more over the SII, no surprises. But they were only so-so on tablets. So they introduced, shocking the world apparently, a 5.2" tablet.. and create the "Phablet" class. Not just that, but the Galaxy Note has a Wacom digitizer in it -- they brought back the stylus, but one that doesn't suck. Now that's in their higher-end 10" tablet, too. That's innovation -- they took a risk, it was well received. Microsoft's copying this in their Surface Pro, so it'll actually be possible to run real Windows applications without a mouse. One sign of real innovation: someone's stealing the idea.
Look at Motorola... nearly dead, they built what was basically the anti-iPhone, in the original "Droid" (aka Milestone), and delivered the first significant Android product. Innovation, because it did well. They did the same thing recently with the revamped RAZR series... a smartphone that actually survives typical smartphone use without a case? They're also doing what companies that will be seen as innovators do: lots of experiments. In the time that Apple went from iPhone 4S to iPhone 5, Motorola fielded five different RAZRs, not to mention a bunch of other smartphones. That's not the way to get to Apple-like profitability any time soon, but it's also the way to, by brute force and refinement, figure out what you can do that Apple, HTC, or Samsung can't.
Microsoft, of course, has done some crazy shit lately.... because they're scared. Sure, they still dominate, still have most of the desktop PC market. But that's actually the wrong question to ask, given the rise of mobile. They have 69% of personal computing, once you factor in mobile.. and that's an older number. It's actually possible that Microsoft will fall to under 50% in the next year or two. Now, sure, any other company would be happy to have that share, but Microsoft's whole business has been built on the fact they can bully pretty much anyone they want in the PC industry. They've made some real enemies, and won't be the same Microsoft without that being their main superpower.
So Apple's floated this Metro/WinRT thing, a tablet version of Windows that's more or less the same thing (with some mysterious compatibility matrix) between different devices. They've basically th
Intel acquired DEC's StrongARM in a law suit settlement with DEC over CPU patents (Intel was using some of the Alpha designs), and that eventually became X-Scale. Which they sold to Marvell some years back. Both Marvell and Intel retain ARM architecture licenses, though Intel's clearly not building any new ARM chips.
Intel's been doing some fab work for a few very small companies. I don't think they're offering anything even close to what Apple needs. Not to mention the old "out of the frying pan, into the fire" problem that would cause -- Intel's hell bent on making x86 compete on price, performance, and power against ARM... and it looks like Microsoft is their biggest booster (they're never getting iOS business, they're still a second fiddle processor on Android, though most apps work -- but on Windows, they have the advantage).
Apple apparently offered TSMC a $1 billion bonus for their entire 28nm capacity... they were turned down. So it may well be they're having or had some issues finding capacity -- they may even be bankrolling some of it. Apple's not looking for average capacity, either, but something fairly cutting edge. Not that their designs are unusually challenging, they're pretty standard. But smaller dice equals higher profits and lower power -- it's important in CPUs in general, critical for Apple's mobile existence.
In fact... Samsung has used a large number of competitors' chips in their own devices. That's almost certainly due to clearing foundry space for Apple. Once Apple's out of the way, you may never see a Qualcomm or TI chip in an Samsung product again... unless they have a better offer. Samsung may also get the competitor's chips cheaper than others, not just due to volume, but perhaps foundry exchange deals -- Qualcomm is fabless, TI not entirely but they lack on the digital side of things. Both usually use TSMC, but not exclusively.