Fascinating doesn’t do justice to the spectacle, nor to the stakes. Taken in pairs, these giants exchange fluids – products and billion$ – while fiercely fighting with their other half. Each company is the World’s Number One in their domain: Intel in microprocessors, Samsung in electronics, Apple in failure to fail as ordained by the sages.
The ARM-based chips in iDevices come from a foundry owned by Samsung, Apple’s mortal smartphone enemy. Intel supplies x86 chips to Apple and its PC competitors, Samsung included, and would like nothing more than to raid Samsung’s ARM business and make a triumphant Intel Inside claim for Post-PC devices. And Apple would love to get rid of Samsung, its enemy supplier, but not at the cost of losing the four advantages it derives from using the ARM architecture: cost, power consumption, customization and ownership of the design.
At its annual investor day last week, Intel CEO Paul Otellini sounded a bit like a spurned suitor as he made yet another bid for Apple’s iDevices business [emphasis mine]:
“Our job is to insure our silicon is so compelling, in terms off running the Mac better or being a better iPad device, that […] they can’t ignore us.”
This is a bit odd. Intel is Apple’s only supplier of x86 microprocessors; AMD, Intel’s main competitor, isn’t in the picture. How could Apple ‘‘ignore’’ Intel? Au contraire, many, yours truly included, have wondered: Why has Intel ignored Apple’s huge iDevices business?
Perhaps Intel simply didn’t see the wave coming. Steeped in its domination of the PC business — and perhaps listening too much to the dismissive comments of Messrs. Ballmer and Shaw — Intel got stuck knitting one x86 generation after another. The formula wasn’t broken.
Another, and perhaps more believable, explanation is the business model problem. These new ARM chips are great, but where’s the money? They’re too inexpensive, they bring less than a third, sometimes even just a fifth of the price, of a tried and true x86 PC microprocessor. This might explain why Intel sold their ARM business, XScale chips, to Marvell in 2006.
Then there’s the power consumption factor: x86 chips use more watts than an ARM chip. Regardless of price, this is why ARM chips have proliferated in battery-limited mobile devices. Year after year, Intel has promised, and failed, to nullify ARM’s power consumption advantage through their technical and manufacturing might.
2012 might be different. Intel claims ‘‘the x86 power myth is finally busted.” Android phones powered by the latest x86 iteration have been demonstrated. One such device will be made and sold in India, in partnership with a company called Lava International. Orange, the France-based international carrier, also intends to sell an Intel-based smartphone.
With all this, what stops Apple from doing what worked so well for their Macintosh line: Drop ARM (and thus Samsung), join the Intel camp yet again, and be happy forever after in a relationship with fewer participants?
There appear to be a number of reasons to do so.
First, there would be no border war. Unlike Samsung, Intel doesn’t make smartphones and tablets. Intel sells to manufacturers and Apple sells to humans.
Second, the patent front is equally quiet. The two companies have suitable Intellectual Property arrangements and, of late, Intel is helping Apple in its patent fights with Samsung.
Third, if the newer generation of x86 chips are as sober as claimed, the power consumption obstacle will be gone. (But let’s be cautious, here. Not only have we heard these claims before, nothing says that ARM foundries won’t also make progress.)
Finally, Otellini’s ‘‘they can’t ignore us’’ could be decoded as ‘‘they won’t be able to ignore our prices’’. Once concerned about what ARM-like prices would do to its business model, Intel appears to have seen the Post-PC light: Traditional PCs will continue to make technical progress, but the go-go days of ever-increasing volumes are gone. It now sounds like Intel has decided to cannibalize parts of its PC business in order to gain a seat at the smartphone and tablet table.
Just like Apple must have gotten a very friendly agreement when switching the Mac to Intel, one can easily see a (still very hypothetical) sweet deal for low-power x86 chips for iDevices. Winning the iDevices account would put Intel “on the Post-PC map.” That should be worth a suitable price concession.
Is this enough for Apple to ditch Samsung?
Not so fast, there’s one big obstacle left.
Let’s not forget who Samsung is and how they operate. This is a family-controlled chaebol, a gang of extremely determined people whose daring tactics make Microsoft, Oracle, Google, and Apple itself blush. Chairman Lee Kun-hee has been embroiled in various “misunderstandings.” He was convicted (and then pardoned) in a slush fund scandal. The company was caught in cartel arrangements and paid a fine of more than $200M in one case. As part of the multi-lawsuit fight with Apple, the company has been accused of willfully withholding and destroying evidence — and this isn’t their first offense. Samsung look like a determined repeat obstructor of justice. My own observations of Samsung in previous industry posts are not inconsistent with the above. Samsung plays hardball and then some.
This doesn’t diminish Samsung’s achievements. The Korean conglomerate’s success on so many fronts is a testament to the vision, skill, and energy of its leaders and workers. But there has been so much bad blood between Samsung and Apple that one has a hard time seeing even an armed peace between the two companies.
And this doesn’t mean Apple will abandon ARM processors. The company keeps investing in silicon design teams, it has plenty of money, some of which could go into financing parts or the entirety of a foundry for one of Samsung’s competitors in Taiwan (TSMC) or elsewhere in the US, Europe, or Israel. If it’s a strategic move and not just an empty boast on PowerPoint slides, $10B for a foundry is within Apple’s budget.
To its adopters, ARM’s big advantage is customization. Once you have an ARM license, you’ve entered an ecosystem of CAD software and module libraries. You alter the processor design as you wish, remove the parts you don’t need, and add components licensed from third parties. The finished product is a SOC (System On a Chip) that is uniquely yours and more suited to your needs than an off-the-shelf processor from a vendor such as Intel. Customization, licensing chip designs to customers — such moves are not in the Intel playbook, they’re not part of the culture.
I don’t see Apple losing its appetite for customization and ownership, for making its products more competitive by incorporating new functions, such as voice processing and advanced graphics on their SOCs. For this reason alone, I don’t see Apple joining the x86 camp for iDevices. (Nor do I see competitive smartphone makers dropping their SOCs in favor of an Intel chip or chipset.)
Intel isn’t completely out of the game, but to truly play they would need to join the ARM camp, either as a full licensee designing SOCs or as a founder for SOCs engineered by Apple and its competitors.
These are risky times: A false move by any one vertex of the love triangle and tens of billions of dollars will flow in the wrong direction.