WSJ reported recently that Intel will merge its PC and mobile divisions, with both being run by Kirk Skaugen, who is currently the chief of the PC arm of the business. This announcement comes after Intel has already lost over $6 billion with its mobile division over the past two years by heavily subsidizing OEMs who would choose its Atom chips in their mobile devices.
Intel has been hoping that once it reaches a certain level of sales in the mobile market, it can end the subsidies and can either increase the price of its chips or keep the price the same, but offer lower-end chips for that price. That's a strategy Intel has already used with Chromebooks this year.
After making a big splash in the media about all the new Haswell-based Celeron Chromebooks (emphasis on "Haswell-based") that cost $300 or less, Intel and its PC partners have quietly removed the Haswell-based Celerons and replaced them with slower Atom-based Celerons in new Chromebooks that didn't need to be subsidized by Intel anymore.
This strategy has mostly gotten Intel what it wanted to achieve, by having customers believing that "Intel-based Chromebooks" are faster than ARM-based Chromebooks (which was true in the case of the Haswell Celerons), even though the company later added chips that were only as fast as, if not slower than, high-end ARM chips.
Because Intel is already in a position where it must heavily subsidize Atom, and it's losing roughly $1 billion every quarter to do it, it may not be possible for the company to repeat this strategy in mobile because it would need to spend even more money to subsidize Celeron, or higher-end models of its Haswell or Broadwell microarchitectures. Furthermore, unlike with Chromebooks, those chips may still be too power-hungry for a mobile phone.
So then, what is Intel's strategy for the mobile market, and why is it merging the mobile and PC divisions? Here are three theories.
Hiding mobile losses
This is a tactic Microsoft has used as well. By combining the revenue it received from royalties from different Android OEMs, Microsoft managed to hide how well or poorly Windows Phone was doing in terms of sales, because it was getting billions of dollars virtually cost-free from the patent royalties, making it possible to hide even billions of dollars in losses with Windows Phone, if necessary.
Intel's PC business is still doing quite well, with over $4 billion in operating profit last quarter, as PCs have been starting to see slow growth lately. By integrating the mobile business into its PC business, Intel could hide how much it's really losing in the mobile market, which could buy the company a few more years of trying to succeed in this business without being questioned every quarter about its losses by the shareholders.
Another theory that the company is also pushing is that Intel will achieve great synergies by combining the two chip businesses. For instance, the company could move Atoms to a new process node as soon as its Core chips do it, too. It's already too late for 14nm, but Intel could launch an Atom chip on 10nm at the same time with its next-generation (presumably Skylake-based) Core chips.
Although by then Intel shouldn't have as large of a difference in the manufacturing process as it used to have, its Atom chips should have a competitive edge over ARM chips -- provided those chips don't cost much more than the high-end ARM competition, and Intel doesn't have to subsidize them again.
Intel has had a hard time competing on price by manufacturing its own Atom chips, so it has recently partnered with low-end chip designers such as Rockchip and Spreadtrum in order to decrease the cost of Atom chips. The problem with that strategy is that these new Atom chips will also have lower performance because of being built on TSMC's 28nm planar process. If Intel can't become competitive on price by building its own chips, then being ahead of ARM chips with the process node may not matter. For the past year or so, Intel's Bay Trail Atom has already been on a 22nm FinFET process when most high-end ARM chips were built on 28nm planar, and that didn't seem to matter either.
A third theory for why Intel is merging its mobile and PC chip divisions is that the company is looking to discard its "mobile chips" and just use one micro-architecture across all product categories, including smartphones and tablets.
Killing Atom may not be that radical of an idea, at least not to Intel. It wouldn't be the first time Intel killed its mobile division (at the time, based on ARM) because it couldn't make any money with it. The division was also likely incompatible with Intel's cost structures and goals.
Intel has been heavily promoting its Core chips lately, especially Core-M/Broadwell-Y, as "suited for mobile," or at least for tablets. It's doubtful if Core-M is as suited for mobile devices as ARM or Atom chips are right now, but we can at least entertain the idea that somewhere down the line, perhaps on 10nm or 7nm, a low-enough-power/low-enough-performance (always a compromise between the two) Core chip could be used in smartphones and tablets of all kinds.
Unfortunately for Intel, it's unlikely this strategy would work. Ideally, Intel would prefer to do this, as Core chips are much more profitable for the company than Atom chips are or could ever be. Plus, it's only a matter of time before Atom/ARM-level performance becomes "good enough," even for higher-end machines, which could pose a threat to the relevance of its Core chips.
That's not a future Intel would like, because the company wouldn't want customers to buy Atom-powered notebooks on which they make little money over Core i3, Core i5 and Core i7 PCs.
Both Microsoft and Intel would hope to sell the "heavier" version of their products, such as Core chips in Intel's case, or Windows RT, which was based on the desktop version of Windows, because they can be priced higher than the lighter versions, such as Atom and Windows Phone.
Trying to fit the "heavier" product in the mobile world didn't work for Microsoft because of competitive pricing pressure, and it likely will not work for Intel in the mobile market, either. Core chips cost too much to ever be relevant in smartphones or most ranges of tablets, except for the highest-end products such as the Surface Pro 3.
In reality, Intel is likely to use all three of these tactics in the next few years. It will take time to get Core chips suited for smartphones and lower-priced tablets, at least from a performance/power consumption point of view.
During this time, Intel can try to make its Atom chips more competitive by building them with its own cutting-edge process nodes, and if "hiding" the mobile losses gets shareholders off the company's back for the next few years, Intel may even use that strategy, as well.
Whether any of these strategies, or even all of them combined, will work to make Intel succeed in the mobile market is something only time will tell, but Intel certainly has many hills to climb before that outcome can be declared.