Intel Tries To Conquer Mobile Market By Investing In Chinese Chip Makers

Today, Intel announced that it signed a $1.5 billion deal with Spreadtrum and RDA Microelectronics, both fabless chip makers from China and part of the Tsinghua Unigroup Ltd., in an effort to get more access to mobile devices in China.

"China is now the largest consumption market for smartphones and has the largest number of Internet users in the world," said Intel CEO Brian Krzanich in a statement. "These agreements with Tsinghua Unigroup underscore Intel's 29-year-long history of investing in and working in China. This partnership will also enhance our ability to support a wider range of mobile customers in China and the rest of the world by more quickly delivering a broader portfolio of Intel architecture and communications technology solutions."

China has a huge and booming mobile market, which is a great opportunity for many players, including chip makers. Intel has been struggling to compete with ARM in the mobile market and has seen relatively little success so far. Mobile OEMs have gotten used to ARM chips, and they have employees that have been hired for their ARM chip expertise, making it difficult to start developing on new platforms.

ARM chips also tend to be significantly more affordable than Intel's solutions. Intel has been trying to fight this with subsidies, but that's not a very sustainable strategy for its mobile chip business, so the company is now looking for alternatives such as selling its Atom CPU IP to other chip makers that can build inexpensive SoCs.

Both Spreadtrum and RDA Microelectronics design chips for low-end smartphones and feature phones along with 2G, 3G and 4G modems, making this deal very complementary to another deal Intel made with Rockchip (a tablet chip maker from China) earlier this year. Through these deals, Intel can cover both the budget smartphone and tablet markets, in China or elsewhere.

"We have been tackling mobile on many fronts, from comms, to integrated IA, to performance and to value and entry segment, making steady and significant progress," Intel's Ellen Healy told us. "The entry and value market is exploding with integrated platforms, which is a key focus for us with SoFIA. We have also made great progress in comms with our latest launch of the Intel XMM 7260 LTE platform for all mobile devices, including performance phones."

She added that Intel is focusing on the growth of value and entry market segments as well as the emerging mainstream segment, "comms" and tablets while the company strengthens its overall roadmap.

Intel seems to be giving these Chinese chip makers the IP for its Atom processors and its own 3G/4G modem technology. The company is then asking Spreadtrum and Rockchip to make SoCs around that IP and build them on 28nm in a foundry like TSMC. Intel is essentially replicating ARM's business model here, since it's selling its Atom IP to other companies that can then make chips using that IP.

Intel will only get a (small) royalty from that IP, but it will be Spreadtrum or Rockchip that will make the bulk of the money from the sale of the chip to OEMs. Intel knows that if it can conquer the low-end market with x86 processors, where all the volume is, it can later try to conquer the mid-range and high-end of the market. However, Intel can't do that on its own, because the company's costs for making low-end chips seem to be too high, especially when building them on cutting edge process technology such as 22nm Trigate or 14nm Trigate.

Spreadtrum, RDA Microelectronics and Rockchip can solve the price aspect of the problem by using cheaper components and an older manufacturing process. The question is what happens to the performance and power consumption of Atom chips on a 28nm non-Trigate/FinFET process when Atom has managed to only be competitive with 28nm ARM chips by using a smaller and more advanced 22nm Trigate process.

This can raise new problems for Intel as ARM chips start to move to 20nm next year, while these Atom-based SoCs from Intel's new partners won't be available until the second half of 2015.

Follow us @tomshardware, on Facebook and on Google+.

Lucian Armasu
Lucian Armasu is a Contributing Writer for Tom's Hardware US. He covers software news and the issues surrounding privacy and security.
  • ldo
    Why does Intel persist in squandering its resources trying to push x86 onto a market that doesn’t want it? Why doesn’t it it abandon those useless Atom chips, and use its legendary fab prowess to make ARM chips instead? That way, its mobile division can actually show a profit for a change.
    Reply
  • IInuyasha74
    Why does Intel persist in squandering its resources trying to push x86 onto a market that doesn’t want it? Why doesn’t it it abandon those useless Atom chips, and use its legendary fab prowess to make ARM chips instead? That way, its mobile division can actually show a profit for a change.

    There is a huge amount of money in mobile devices like smartphones and tablets. The PC market is getting smaller because of these devices, and its key for the company to be able to extend into new areas as a result.

    You shouldn't assume no one wants Intel in smartphones and tablets. A great number of people do. If nothing else they want it there for competetion, but a lot of people would enjoy the idea of Intel having an energy efficient high performance CPU in their device.

    Its new Broadwell 14-nm based atom chips, which have already been looked at a little by reviewers from samples, look to over much greater performance than ARM CPUs while hitting the same power level, making them highly competitive.

    Intel's mobile division has been profitable, just not to the point they want. It would be a terrible thing for them to make ARM chips instead of their own. It would be like saying we should completely get rid of AMD in the PC market and only have a choice of Intel. Its close to the worst possible thing that could be done.
    Reply
  • ldo
    I don’t need to “assume”, I just look at what’s actually happening in the market. Yes, there is a huge amount of money to be made in mobile; why doesn’t Intel make some of it, instead of just throwing it away?
    Reply
  • IInuyasha74
    You are assuming that no one wants it, when many people do. The fact that companies aren't making a lot of products with Intel CPUs in the mobile space doesn't mean that customers don't want it, it means for some reason (there are various reasons) that they do not want to make an Intel device. A lot of this is usually because they are already used to working with ARM hardware and don't want to transition into something new.

    Intel has a lot more to gain by managing to enter into the mobile space with its own products.
    Reply
  • ldo
    I don’t need to “assume”, I just look at what’s actually happening in the market. Companies have tried using Atom chips in mobile devices lots of times, but they have never sold well. That’s why they stop making them.

    Remember how, when announcing the first Atom chips in 2008, Intel proudly proclaimed that it was *two years* ahead of ARM? It didn’t take long for that smirk to be wiped off its corporate face.

    And ever since then, Intel has been playing catch-up. Just wait for the next generation of chips, they kept saying: wait for Stoneybridge, or Tacoma Bridge, or Oilswellthatendswell, or whatever the latest code name is. This time for sure, it will whip ARM’s arse.

    Only it never did. Because the ARM ecosystem is so much more diverse and competitive than anything Intel can manage on its own.
    Reply
  • Jaroslav Jandek
    I don’t need to “assume”, I just look at what’s actually happening in the market. Companies have tried using Atom chips in mobile devices lots of times, but they have never sold well. That’s why they stop making them.
    That must be why ASUS T100 (with Atom) was the top 3 best selling tablet on Amazon in Q1/2014.
    btw. I own one and a lot of my friends and coworkers own an atom-based tablet or 2-in-1. Not sure where you got the "nobody wants them" from?

    Edit: it was also #2 best selling tablet on Amazon in Q4/2013.
    Reply
  • ldo
    Only on Amazon? And only for one quarter?

    I think we’ve seen the same kind of achievement claimed for Nokia Lumias...
    Reply
  • Sakkura
    14257666 said:
    Its new Broadwell 14-nm based atom chips, which have already been looked at a little by reviewers from samples, look to over much greater performance than ARM CPUs while hitting the same power level, making them highly competitive.
    Broadwell is not Atom. The new Core M (Broadwell-Y) chips that have been previewed are not Atom chips, they belong to Intel's high-performance x86 lineage. Intel's newest Atom design is the Silvermont architecture. The phone-targeted Silvermont chips are the dual-core Merrifield and quad-core Moorefield, while for tablets, netbooks etc. it's Bay Trail.

    The die shrink of Silvermont is called Airmont, and it will go into new chips like Braswell (phones) and Cherry Trail (tablets). Note that Braswell is completely different from Broadwell.
    Reply
  • somebodyspecial
    ROFL. A 28nm x86 chip won't take out the ARM 20nm armada that is already better even at 28nm. The only solution to the ARM problem is to OWN a VERY good ARM company (Nvidia is about the only option to buy) and pump out THEIR stuff on your best process. Intel doesn't seem to understand how to win or for some reason can't get this done. Either Jen won't sell or Intel just plain refuses to give him whatever it takes to get this over with. Either offer him a few billion outside the company buy price, or offer him the position he wants in Intel (CEO). OR face continuing 4B losses in mobile and even more as mobile becomes MORE than mobile. IE, I bet money you'll see denver etc desktops at some point on 20nm or 14nm with a PC like heatsink/PSU, discrete vid cards etc giving you a FULL PC experience power wise rather than nibbling at the notebook market with chromebooks in a ~10w envelope. Just like those stole 21% of the ENTIRE notebook market, these PC LIKE machines geared for 500w power supplies etc, will steal yet another 21% (or more) of the desktops also (and no doubt more notebooks stolen too) as they strap on PC like parts for real and 64bit ARM software takes hold giving even app devs a reason to run to the massive numbers of units on ARM's side. The ARM side can easily strap on more power in say a 50-85w SOC model @3.5-4ghz once the software is 64bit and these things start coming with 4GB-32GB of ram so you can run much more complicated apps. Devs won't just ignore a soon to be 2Billion+ per year unit market just to make Wintel happy on x86. They will move, just like the game devs have already.

    http://investing.money.msn.com/investments/financial-statements?symbol=intc

    Intel profits down from 12.94B, to 11B, now to 9.62B over the last 3 years. The trend will continue until they figure out how to buy Nvidia. Losing 4B a year on mobile means you should just offer Jen Hsun $4B in cash NOW (before you can't afford to buy them) to walk away + the price of the company. The guy is worth what like 300-450mil? I don't know many people who'd walk away from 10x their net worth in a day to go find something else to do with their time. Then again maybe his bean counters did the math and figure he'll be worth that in 10yrs anyway and gets revenge in the end on Intel (at least to a large degree as we already see in intel profits even before the desktop assault for years to come) for their chipset shenanigans. I'm not sure I'd walk myself seeing the current trend and the gpu ip money that will be coming NV's way for years now that mobile (and Arm's assault on desktops) will be doing the same things they've patented for over a decade for current x86 pc's.

    NV could easily end up becoming Qcom or Intel in profits once the lawsuits are done and people are forced to pay up, cars go driverless (using even more SOCS etc to do this), GRID starts paying, and NV's gpus themselves start to take over the mobile game (instead of the modem being #1 in the last decade+ of mobile) year after year. You're either going to pay them or just simply use them instead of paying them, or you'll be out of the GAME game so to speak. Look at the app stores for google, amazon, apple, msft and you can see GAMES rule, and are expanding their revenue in these stores yearly.

    As more people start to find out these things at 20nm or 14nm (or heck 10nm at about halfway through console lifespan) can do what a console can do and much more the picture just keeps getting worse for Wintel/x86/consoles. They are cheaper and can do most of the stuff the general public needs already, so amp that up and the Wintel/x86/console market just gets more ugly than Intel's profits, sony profits, nintendo, and microsoft gaming division already are showing (they all are sucking wind). Add some REAL apps at 64bit etc and it just gets worse as they start stealing WINTEL apps revenue also (and pc's sold for these jobs).

    Of course as consumers we'd all probably be better off if Intel is never able to buy NV, and we just end up with a 50/50 ARM/x86 race at some point (having ARM vendors replacing AMD as a cpu competitor that is FAR more competent and profitable). If ARM takes 21% of desktops in the next two years do Intel profits end up at 6-7B from current 9.6B? If the current trend continues I don't see how they keep up with the fab race. At some point when profits can't pay for it, R&D comes down yearly just like at AMD and at some point it shows it's ugly face in your products just like AMD now. Tonga is not an answer to maxwell, it's better but not near good enough hence the pretty silent tonga launch. Whatever AMD launches next (their 20nm chips), NV will just answer with BIG maxwell or a similar 20nm version of today's chip amped up (we know they hit 1.5ghz on 28nm now with limited voltages, surely a 20nm shrink of maxwell could ship at these speeds if needed). We are seeing AMD's decreased R&D (3-4 years of shrinking R&D) now. Since it takes 3-5yrs to make a chip it makes sense you see the results of shrinking R&D 3-5yrs later like we see now and will continue to see unless AMD figures out how to raise profits & thus raise R&D again. Anything in their pipeline has had less and less R&D aimed at it over the last 3-4yrs (cpu and gpu, hence totally out of the cpu race, and barely hanging on in the gpu race).

    We need AMD to get bought ASAP, before their gpu tech is so far behind (which I'd predict we'll see in 2-3yrs as that shrunk R&D hits yearly even more), even a company with billions behind it can't fix the problem for years if ever. I'm not sure their cpu tech is worth anything to anyone but AMD at this point as ARM/Intel will eventually just eat that lunch as they meet in the middle. Console profits will shrink for them after xmas pop from microsoft launching in all the other countries. As SOCS hit 20nm/14nm poor people will go to ARM devices and cheap games vs. an expensive console and $60 games for life. The sales from casual users will not develop for this gen of consoles in the back half of their lifespan like before as mobile begins to match them. You will only get the hardcore in the first 3yrs, then kaput. Microsoft/Sony could slow the process a bit if they lowered game pricing to $30-40 but you're still going to battle for sales even if ARM games end up being $10-20 across the board and I'm pretty sure there will always be $1-10 games in the android market to appease truly poor people who never buy consoles which will push devs even further into ARM's ecosystem.

    Think about how many people own a mobile device even today (billions with 1.2B sold YEARLY again and climbing) vs. how many own ALL 3 CONSOLES from last gen COMBINED (about 300mil total over 7yrs of life). Game over man...LOL. In the next 7yrs they will probably sell a total of something like 10B+ units of ARM devices and over half of them will match or beat consoles in perf/graphics. Think 14nm/10nm M1 (maxwell or volta? so maybe at V1?), NVlink with TSV, 8ghz+ memory etc surrounding these socs to massively ratchet up perf for lower cost and more abilities vs a console which basically plays games/movies. How fast will a 14nm Volta based soc be after seeing 28nm kepler K1? Somewhere in 2016/2017 you'll be seeing 10nm versions of either V1 (volta) or whatever is after it and all it's competition. Where does AMD console profits go then? What product will get them back to 500mil profits? How do you do that as you have to slash prices immediately to be considered this and further forward vs. maxwell etc?

    http://www.fool.com/investing/general/2014/09/26/nvidia-corporations-gtx-970-and-980-put-amd-to-sha.aspx
    This sums up AMD's problem pretty well and highlights the R&D problem discussed above. I'll say it again, we need them bought, but I don't know who would want them (NV would be bought first by MSFT/INTC/Samsung/Google). Intel surging in cpu vs. AMD also makes matters worse. I don't see how they recover without some market that AMD can get into that these others are NOT playing in already and I don't know where that will come from. AMD might have survived if they'd spent on GPU/Arm SOC/APU instead of consoles but that didn't happen.
    Reply
  • somebodyspecial
    14257666 said:
    Why does Intel persist in squandering its resources trying to push x86 onto a market that doesn’t want it? Why doesn’t it it abandon those useless Atom chips, and use its legendary fab prowess to make ARM chips instead? That way, its mobile division can actually show a profit for a change.

    <snipped>

    Intel's mobile division has been profitable, just not to the point they want. It would be a terrible thing for them to make ARM chips instead of their own. It would be like saying we should completely get rid of AMD in the PC market and only have a choice of Intel. Its close to the worst possible thing that could be done.

    Intel's mobile division is losing 4B per year now (1.1B lost last quarter up from 930mil the quarter before, it's accelerating losses with subsidies). No point in bothering with the rest of your comment when you clearly don't read balance sheets and quarterly reports. Broadwell will do nothing to change things as ARM negates it with a move to 20nm. IE, Intel moves to 14nm and ARM moves to 20nm so you net nothing (and the ARM gpu side just got to desktop gpu mode, so this story just gets worse, especially as NV goes custom with Denver next month - IE, IN HOUSE ARM cpu). See my other post for why Intel is going down. Where your profit goes (up or down) so shall your R&D go (see AMD 10yr summary below and check the R&D in the balance sheets/earnings reports).
    http://investing.money.msn.com/investments/financial-statements?symbol=US%3aAMD

    You're losing the ARM's race (heh, pun intended), and will shortly start losing the FAB race as ARM assaults the desktops and takes even more notebooks (and thus your profits continue to slide for a 4th year, and thus your R&D slides at some point too). You need to read more. Your last comment is the absolute opposite of what Intel needs to do RIGHT NOW. When you can't beat them join them until you can by beating them at their own game. BUY NV now and produce your own ARM IN HOUSE chips (because you'd own K1, M1, V1 etc, and 85% of pro card market, server gpus, 65% of discrete gpus etc) and your new found top end gpus on the best process out there WHILE YOU STILL HAVE IT!

    http://www.eetimes.com/document.asp?doc_id=1322263
    Not sure what you read, but places like JP Morgan don't ask you to QUIT mobile for nothing. This was before the last quarters 1.1B loss. With it growing, I'm sure JP Morgan still thinks the same as before and there is NO sign of it quitting soon as ARM marches to 20nm negating any Intel 14nm gains. They are both marching in lockstep.
    Reply