Intel’s NUC illustrates why the company struggles in a post-PC world

What starts as a rant ends as a serious margin problem

Intel - logoWhen Intel first put out the NUC FFRD, SemiAccurate thought that it was not only a bad idea, it illustrated all that was wrong with modern Intel. If you take a closer look at the underlying tech, it paints a grim picture for Intel’s prospects in a post-PC world.

Intel is desperately trying to save the PC from the quick death looming above it, and the NUC is their latest ill-conceived master plan. More importantly, they need to do this in a market that is quickly moving from proprietary to commodity devices. Trying to get a bigger share of a shrinking BoM is decent business, but it has limits. To make matters even more grim, the consumer market has spoken and it is saying that the current Wintel path is one is not wanted, regardless of price. Technology, marketing, and consumers all add up to the NUC crashing and burning in a spectacular fashion.

Unfortunately it also points out that Intel is in far deeper trouble as a company than anyone wants to admit, they simply do not understand how to compete in this new post-PC world. The NUC is prime evidence that they are speeding down the wrong path while building ever more efficient blinders. This cute little PC form factor is the latest canary coughing in the coal mine, and the underlying problems are quite fatal to Intel’s bottom line. Lets take a look at what the NUC says about Intel’s tenuous margin structure and start it out with a merry rant that will make you smile before you realize the trouble your portfolio is in.

If you are not familiar with the NUC, it stands for Next Unit of Computing, basically a little 4×4 inch box that has a full sized Intel CPU in it, not an underpowered Atom. It screams Apple envy more than Ultrabooks, but if anything is an even worse idea. Why? Lots of reasons, price being the most obvious one. By price we don’t mean simply the retail toll of a NUC system, price also encapsulates silicon cost, BoM cost, Windows overhead, and the margins necessary to keep the big blue silicon machine going. Intel wants to reinvent the PC, but is unwilling to take even the tiniest of steps in the right direction to do so, and NUC is the result.

Why was our initial reaction so negative? Price was the big red flag but next up was the fact that it was just a PC in a new box, but it utterly fails to bring anything actually new, much less innovative, to the table. Intel made it smaller and at the same time made it more expensive, but they also removed several necessary PC features in order to cram in to a form factor that consumers were simply not asking for. People don’t want PCs any more, they want something new. PCs are not selling, new things like tablets, uber-phones, and widgets are. Unfortunately the NUC is an underpowered PC in a “been there, done that (by Apple naturally)” box. From there it adds nothing users have been asking for, much less are lining up to buy, and is painfully expensive to boot.

The currently available NUCs cost $299 as a barebones system, but you can get them with a red top for only $15 more. Since it also forces Thunderbolt on you at the cost of a second monitor port, this one is best avoided. These powerhouses come with a massively powerful CPU, the Intel Core i3-3217U CPU, a dual core 1.8GHz screamer with turbo, virtualization, AES-NI, Untrustworthy Compute, and a few other bits fused off. In case you didn’t catch it, that was bitter sarcasm, this thing is dog slow. On the up side, it is a 17W part but given the size of the box, what did you expect? Unfortunately since this CPU is an Intel ULV part, it is priced like all their other low wattage parts, this one costs a rather stunning $225. Intel has a way of matching useful parts with extreme margins, and the NUC CPU should be a very good primer on how they intend to keep margins up in the near future.

The next problem is that the NUC has a Q77 chipset, a business part aimed at the corporate IT set, but the CPU paired with it fuses off the two things that category needs most, virtualization and encryption. You have to admire the marketers here, the stopped the possibility of the NUC cannibalizing both business and consumer PC sales by making it so awful no one in their right mind, business or consumer, would buy it. Would you buy a business PC without virtulization, encryption, trusted compute hardware, or vPro/remote management? Would you buy a media center with underpowered graphics, almost no storage options, and no expandability whatsoever? These features that one potential markets or the other require were either were removed for no good technical reason, precluded by the form factor, or not possible with Intel hardware in the first place. Genius level product strategy there guys, pure genius.

On the up side, the GPU is the most powerful one Intel currently offers, the 16-shader HD4000. Unfortunately, to save power and fit in the 17W envelope, it runs at 350MHz. Worse yet it is an Intel GPU, so that means you will not get working drivers, Intel policy precludes that possibility. So the NUC is saddled with slow graphics, broken drivers, and a form factor so small that it prevents a buyer from of putting in a graphics card. If this was a business PC you could possibly forgive this stupidity, but as we said above, that was precluded with the fusing off of all useful business features like virtualization and AES-NI.

The NUC is a set top box on steroids according to Intel, things like signage and kiosk duty are just variants of the same “play a video and do light input duty” theme. The one thing you want to avoid if at all possible on a media-center type machine is Intel graphics, they simply do not have the power to do the job much less drivers that work right. With the NUC there is no other choice. To balance this out, the machine has more CPU power than is actually useful for a set top box, kiosk, or digital signage box, but the graphics are nowhere near adequate. How do they expect silicon that is ill-suited for either potential task that the external form factor works in to sell?

The current NewEgg deal on the NUC has a special offer for potential buyers, a deal so sweet that it may change a few skeptical minds. NUC is sold as a barebones chassis with board and CPU, but for a limited time, for only $299, you get the NUC and, wait for it, a *FREE* power cable! Free! Holy sh*t, a free power cable with every NUC purchased! Be still my beating heart, according to NewEgg, that is a $4.05 value! No, put “!!!” at the end of that last sentence, any deal that good deserves it. If this isn’t a reimagining, nay a reinvigorating of the PC, I don’t know what is. How many enthusiasts building their own PCs do you think lack a few dozen spare power cables? I just ordered 71 NUCs to take advantage of this limited time offer, you should too, at least Intel seems to think so. The fact that the NUC does not include a power cable should clearly illustrate the margin problems on this box.

Back to reality, to make a working system around a NUC barebones, you need memory, a HD, and an OS, although a Wi-Fi card is probably necessary too. While the NUC doesn’t come with Wi-Fi built in, it does have Thunderbolt, but you lose half your monitor outputs for that model. If you put Wi-Fi in it, that takes up the mini-PCIe slot, singular, and that is 100% of the NUC’s expansion capabilities. Dear god, who thought this thing up, and how did he or she remember to breathe regularly?

It seems that Intel purposefully set out to put all their most expensive but useless technologies in to one package, and somehow thought a long list of bullet points would crush their competition in to the ground. Did no one actually bother to wonder if, oh say the fundamentals like CPU power, GPU power, and expandability mattered to customers in the slightest? Really, you would have trouble if you set out to design something more ill-suited to all potential markets on purpose, a crank powered electric car with a microwave, HAM radio tower, and 4000 gallon salt water aquarium makes more sense to me.

So to recap, we have a $300 barebones system that has one mSATA port and one mini-PCIe slot, and that is the entirety of the NUC’s internal expandability. If you put in an mSATA SSD, there is no room for an actual SSD much less a HDD, and Wi-Fi, that is all of your expansion used, all of it. It has Thunderbolt ($15 extra), an underpowered GPU, broken graphics, and zero chance of further expansion. It may have more CPU power than needed for the media functions, but not enough for real business use. The same marketing games around business oriented features precludes that the NUC from that market also put the business oriented Q77 chipset on the board, no other options. A business chipset without the CPU power to do business work, but enough to do light media center work that the lack of GPU power precludes. Does this make any sense to you at all?

Back to the system cost, you just need to add an expensive mSATA SSD, currently about $100 for 100GB or so, DRAM, say $50 for a couple of sticks, and $25 for Wi-Fi. Toss in $100 for a copy of Windows Home because as miserable as Intel’s GPU support is for that OS, it is worse on Linux. If you are wise enough to buy a NUC, Windows is the right fit for you. Sarcasm again there. Anyone capable of using Linux would got for a machine that adds up to far less than the $575 here but does infinitely more.

And that is the point really. When kitted out to bare minimum PC standards, the NUC is scraping the $600 mark system mark for a very meager set of specs. Add in a tolerable amount of storage and that total spirals up quickly. You don’t think 100GB is enough for Windows and a few programs, much less a media center machine do you? Movies and music take up a lot of space, more than just about any other file type. That brings me to the thing that convinced me the NUC was as badly thought out, ill-conceived, and out of place for all markets as I originally thought.

It came in an ad from a local Tier 2 PC vendor with a large regional presence. They are not international, but are a force in a large part of the US and own several lucrative niche markets. They don’t have the volume of a Dell or Lenovo, so their prices are probably up a bit from what the big guys offer, but not more than about $50 a system. That said, you get unquestionably better flexibility and support than the three models backed by phone banks from hell than the big guys offer. To many, that is worth the slight premium. Their price for a NUC? The price from this vendor for the Thunderbolt version with 4GB of DDR3, an 80GB mSATA drive, and Windows 7 Pro? $699.

Nope, I am not joking. Even if Dell, Lenovo and the rest of the big boys can shave $100 off that, you are still at $600. Add in more RAM, essentially free, and a >200GB mSATA SSD, and you are right back above $700. $700 for a a freaking Apple TV with none of the class, style, software, or, well, basic functionality that even a low end PC offers? Is Intel on crack? Do they actually expect to sell a single one of these? Seriously, that was not a rhetorical question, if whoever green-lighted this turkey still has a job, Intel is more lost than even their worst critics believe. But that isn’t all the bad news.

If you take a barebones machine with a tiny SSD and skimp on most everything else, you can squeak into a NUC system for a bit over $500 if you build it yourself. You would have to keep your movies, music, and everything else off-device, streaming media or loading everything on the fly before every playback. It is at that point a set top box with delusions of grandeur, but lacking the ability to actually achieve anything grander. But there is a bigger problem.

If you look at the NUC as a $500-$700 set top box with the ability to step beyond that role on rare occasions, it can fill that niche fairly adequately, leaving aside the general business use case. It will be hamstrung by storage with a single mSATA port, expandability with a single PCIe port, both used in the basic configuration, and extremely limited graphics capabilities with again no option to expand. The bigger problem is what other things can do the same “set top box plus a little” job, give or take a few features here and there. Take a look at this little thing, it is called the G500A.

G500A Android set top box

This box is the NUC killer

I first saw the G500A in Taipei in the Guang Hua market last June, but it was just one of many. There were dozens of variants on the theme, size, color, capacity, performance, and everything else, but this one was one of the best overall. It is barely wider than my business card, and the depth is about the same, so maybe 4″ x 4″. For size it is a dead ringer for the NUC but, well, just looks a whole lot better in my eyes.

Going down the feature list, it has an anemic low end ARM CPU of some sort, likely an A8 variant, around 512MB of DRAM, and 4GB of Flash. It doesn’t really matter though, it runs Android 4.x so that is more than enough to do all the things an Android set top box needs to. Windows, as shown off by the Surface Pro, needs a stunning tens of GB with only the OS and mandatory patches installed. Add in necessary antivirus, antimalware, utilities, and the odd program or two, and that 80GB mSATA SSD is going to be mighty cramped.This is before a movie player, something not included in either Windows or the $600+ price. Big programs like Office are almost precluded until you add hundreds of GB more storage, something that won’t fit in the aforementioned $600 price.

The i3 in the the NUC has hundreds of times the raw compute capabilities of the feeble ARM CPU in the G500A though, that should help right? Well no, as a set top box, Intel’s broken drivers matter far more. The G500A will play 1080p video smoothly, it runs Android apps passably, and didn’t bog down when I played around with it. It definitely will struggle if you load it down, but this is a set top box. The NUC struggles with video whereas the G500A burns through most standard video formats with ease. It just works and you don’t have the miserable Windows driver and malware problems seemingly every week either.

But the NUC has far greater expandability right? It has ethernet built in, for $25 or so more you can add Wi-Fi too, and it has not one but two USB ports. So does the G500A, plus the Wi-Fi is included along with Bluetooth, something that is not really an option on the NUC unless you want to forgo the Wi-Fi. Not really a sane trade there. One area where the NUC crushes the G500A is wired networking, it has GbE while the ARM box can only do 100Mb. Then again, who wires their set top box with ethernet? How many consumers have GbE networks in their house, not enthusiasts mind you, but average consumers?

Moving on to uncharted waters, the G500A has a card reader in front, the NUC does not, front or back, standard or optionally. Seriously, no card reader on a $700 PC, it won’t fit. The G500A only does one HDMI out while the NUC can do two, one if you get Thunderbolt, but that would just be an excessively stupid trade-off. G500A is totally unlocked while the NUC bows to Microsoft’s mis-Trusted Computing program and locks the user out of their own hardware then adds DRM too. Win/Win there. The coup de grace is that the G500A lists “cable TV” functionality, albeit in Chinese so it may not be what it sounds like, but if it is, clean kill.

When you tally the spec sheets up, the NUC is vastly more powerful than the G500A in just about every way that matters except those that modern consumers actually want. On that front, the comparatively overwhelming CPU power add no value but the GPU power is inadequate for the task at hand. Users are saddled with an expensive OS that takes up more than 10x the resources that the G500A has in total just to show the desktop, and takes lots of added software to do basic media center tasks like play a DVD. The G500A is ahead on expandability by a card reader and Bluetooth, but behind on wired networking. Take your pick unless the cable TV part is actually what it sounds like, then it isn’t really a fight.

Both devices can play movies at 1080p all day and both are effectively bound to streaming device duty due to storage limitations. When push comes to shove, they are both about dead equal in consumer usefulness, fungible goods in economic terms. There are minor feature differences, huge but meaningless spec differences, and but both will play and stream 1080p content until your local Teh Intertube runs out of things to watch. They are both set top boxes.

If you want more storage, features, or connectivity, you can hunt down multi-function bits for the NUC but they are pricey and drivers are problematic. The G500A is effectively non-expandable, but if you walked 3 booths over in Guang Hua market, there was a variant of the ARM + Android tiny box theme on sale that probably met your needs. Some had more feature, some less, and some had lots more storage too. While you could easily find half a dozen versions that met your specific requirements, it might take 20 minutes of shopping around to find. In short, both devices could probably do most of what you want with a bit of time, money, or both put in.

And though all this, why am I ranting about the NUC? Why am I complaining about the sub-low end PC specs at a higher end PC price? That happens all the time when new form factors are introduced, what is the problem? The problem for Intel is that the G500A I saw in June was on sale for the rather steep price of 35 USD including the power cables. Yes, $35 vs $699, $575 for a box of parts without the Pro version of Windows or Wi-Fi. Adding either of those costs about the same as the complete G500A. If you wanted to seriously bump up the specs of the G500A, you could find a box 5 stalls over for less than the cost of the SSD in the NUC, add in a 32GB flash drive, and still be under 100 USD too.

This device would run most Android apps, have more user accessible storage than an 80GB NUC, and cost a mere fraction of what the NUC does. The G500A alone costs 1/20th of the locally quoted price for the NUC, and less than 1/15th of the do it yourself version. Please note, that is single digit percentages in both cases for a device that does all the useful things a NUC does, but doesn’t pick up malware and rootkits every few days. In fact, Intel wants 6+ times the cost of the entire G500A for the NUC’s CPU alone! Now you see why I think the sales count to non-Intel employees will be single digits? It may be dumb, but is it priced out of line with reality.

In an nutshell, that is Intel’s problem, they want a large multiple of the competition’s retail price for their silicon alone, and BoM cost goes way up from there. Their reinvention of the PC is not just dumb, it is so far out of line with reality that you have to question that they bothered at all. As a concept device, it probably has some interesting packaging technology, but productizing it shows exactly how bad their competitive position really is.

When your competition can equip every screen in a modern house for a small fraction of what the Intel solution costs, it isn’t going to be a fight much less a fair one. Intel may claim advantages until they get bored of listing them out, but the ARM based widget boxes do 90% of what the NUC does, closer to 100% of the actually useful things that it does, and does it for 1/20th of the price. 1/20th, and that was in June 2012. The G500A is likely old news by now with three new versions offering more features supplanting it since I saw it. If you decide on one that costs about double the price of the G500A, you can take home an ARM box with many times the power and storage, but it will set you back an onerous 1/10th of the NUC’s retail price. 1/10th for a more usable machine than the NUC by far, 1/20th for 90% of the functionality, that is just crazy.

Intel thinks the NUC is a re-imagining of the PC. Intel thinks it will save the PC market. Intel thinks they have a packaging marvel on their hands and the price they ask for the advanced tech is fair. In fact, Intel thinks people actually want a piece of crap like the NUC and will pay $600+ for it. Would you? Any casual observer will tell you it is a PC in a small box with most of the useful features of a PC removed. They will also tell you the price is more than two cheap laptops or three low end PCs, both with vastly superior features. They will probably tell you they also don’t want one, but that may only come after they hear the price. Any sane person will take matching His and Hers G500As, a couple of 32GB memory sticks, and pocket the extra $500. Wouldn’t you?

Intel doesn’t get this. Intel is going down the wrong path. Intel’s very core strategy is broken. Their mindset is a decade out of date, and with a three year product cycle, by the time they can turn the HMS Otellini, it will be far too late. Scratch that, it already is too late. The world has moved on from Wintel, just ask any iPad owner. Neither Microsoft nor Intel will admit this to themselves though, they just keep doubling down on ever increasingly expensive things consumers don’t want, and are deaf to pleas to change.

Intel thinks Windows 8 on a NUC for $700 will reverse the PC decline. If not, next year’s Windows 9 NUC2 with quad-cores for only $100 more will surely set the market on fire. Intel’s Google TV died because of the same visionary product development mindset. Ultrabooks sold 1/8th of their 2012 target for the same reasons. PC sales are in the toilet, Microsoft is busy proclaiming victory, and Intel is ratcheting up prices on Haswell to untenable levels.

If Intel does figure out what to do at this point, their margins won’t sustain going down that path any more. If the right people understand why the NUC is so wrong tomorrow and come out with a product that fixes everything the next day, they couldn’t afford to build it and keep the lights on. Maybe Intel does get it after all.S|A

The following two tabs change content below.

Charlie Demerjian

Roving engine of chaos and snide remarks at SemiAccurate
Charlie Demerjian is the founder of Stone Arch Networking Services and SemiAccurate.com. SemiAccurate.com is a technology news site; addressing hardware design, software selection, customization, securing and maintenance, with over one million views per month. He is a technologist and analyst specializing in semiconductors, system and network architecture. As head writer of SemiAccurate.com, he regularly advises writers, analysts, and industry executives on technical matters and long lead industry trends. Charlie is also available through Guidepoint and Mosaic. FullyAccurate