Intel rode the PC wave with Microsoft and built an seemingly insurmountable lead in the field of "conventional" (PCs and laptops) microprocessors. But, after his predecessor missed the opportunity to supply the CPU chip for Apple's iPhone, Intel's new CEO must now find a way to gain relevance in the smartphone world.
In last May's The Atlantic magazine, Intel's then-CEO Paul Otellini confessed to a mistake of historic proportions. Apple had given Intel the chance to be part of the smartphone era, to supply the processor for the first iPhone... and Otellini said no [emphasis and light editing mine]:
"The thing you have to remember is that this was before the iPhone was introduced and no one knew what the iPhone would do... At the end of the day, there was a chip that they were interested in that they wanted to pay a certain price for and not a nickel more and that price was below our forecasted cost. I couldn't see it. It wasn't one of these things you can make up on volume. And in hindsight, the forecasted cost was wrong and the volume was 100x what anyone thought."
"...while we like to speak with data around here, so many times in my career I've ended up making decisions with my gut, and I should have followed my gut. [...] My gut told me to say yes."
That Otellini found the inner calm to publicly admit his mistake — in an article that would be published on his last day as CEO, no less — is a testament to his character. More important, Otellini's admission unburdened his successor, Brian Krzanich, freeing him to steer the company in a new direction.
And Krzanich is doing just that.
First: House cleaning. Back in March 2012, the Wall Street Journal heralded Intel as The New Cable Guy. The idea was to combine an Intel-powered box with content in order to serve up a quality experience not found elsewhere (read Apple, Netflix, Roku, Microsoft...). To head the project, which was eventually dubbed OnCue, Intel hired Erik Huggers, a senior industry executive and former head of BBC Online.
At the All Things D conference in February, Huggers announced that the TV service would be available later this year. The Intel TV chief revealed no details about how the service OnCue would differ from existing competitors, or how much the thing would cost...but he assured us that the content would be impressive ("We are working with the entire industry"), and the device's capabilities would be comprehensive ("This is not a cherry-pick... this is literally everything").
Intel seemed to be serious. We found out that more than 1,000 Intel employees in Oregon had been engaged in testing the product/service.
Then Krzanich stepped in, and applied a dose of reality:
Intel continues to look at the business model.... we are not experts in the content industry and we're being careful." [AllThingsD: New Intel CEO Says Intel TV Sounds Great in Theory. But …]
Indeed, to those of us who have followed the uneasy dance between Apple and content providers since the first Apple TV shipped in 2007, the Intel project sounded bold, to say the least.
Late September, the project was put on hold and, last week, the news came that OnCue had been cancelled and allegedly offered to Verizon, whose V Cast media distribution feats come to mind...
Even before OnCue's cancellation was made official, the well-traveled Erik Huggers appeared to show an interest in the Hulu CEO job. (If Mr Huggers happens to be reading this: I'd be more than happy to relieve you of the PowerPoints that you used to pitch the project to Intel's top brass, not to mention the updates on the tortuous negotiations for content, and the reports from the user testing in Oregon. These slides must make fascinating corpospeak logic.)
Krzanich quickly moved from doubt to certainty. He saw that OnCue would neither make money by itself, nor stimulate sales or margins for its main act, x86 processors. OnCue would never be an Apple TV "black puck", a supporting character whose only mission is to make the main personal computers (small, medium and large; smartphones, tablets and conventional PCs) more useful and pleasant.
So he put an end to the impossible-to-justify adventure.
That was easy.
Tackling Intel's failure to gain a significant role in the (no longer) new world of smartphones is a much more complicated matter.
With its x86 processors, Intel worked itself into a more-than-comfortable position as part of the Wintel ecosystem. The dominant position achieved by the Microsoft-Intel duopoly over two decades yielded correspondingly high margins for both.
But smartphones changed the game. ARM processors proved themselves better than x86 at the two tasks that are integral to personal, portable devices: lowering power consumption and customization. The ARM architecture didn't have to wait for the iPhone and Android handsets to dominate the cell phone business. Just as Windows licensing spawned a large number of PC makers, ARM licensing contributed to the creation of a wide range of processor design and manufacturing companies. The ARM site claims 80 licensees for its newer Cortex family and more than 500 for its older Classic Arm processors. No monopoly means lower margins.
Intel saw the unattractive margins offered by ARM processors and didn't want to commit the billions of dollars required by a fab (a chip manufacturing plant) for a product that would yield profits that were well below Wall Street expectations.
The prospect of bargain basement margins undoubtedly figured in Otellini's decision to say no to the iPhone. In 2006, no one could have predicted that it could have been made up in volume, that there would be a billion smartphone sales in 2014. (I'm basing the 1B number for the entire industry on Horace Dediu's estimate of 250 million iOS devices for 2014.)
Even if the Santa Clara company had had the foresight to accept lower margins in order to ensure their future in the smartphone market, there would still have been the problem of customization.
Intel knows how to design and manufacture processors that used "as is" by PC makers. No customization, no problems.
This isn't how the ARM world works. Licensees design processors that are customized for their specific device, and they send the design to a manufacturer. Were Intel to enter this world, they would no longer design processors, just manufacture them, an activity with less potential for profit.
This explains why Intel, having an ARM license and making XScale processors, sold the business to Marvell in 2006 - a fateful date when looking back on the Apple discussions.
But is Intel's new CEO is rethinking the "x86 and only x86" strategy? Last week, a specialty semiconductor company called Altera announced that Intel would fabricate some if its chips containing a 64-bit ARM processor. The company's business consists of offering faster development times through "programmable logic" circuits. Instead of a "hard circuit" to be designed, manufactured, tested, debugged, modified and sent back to the manufacturing plant in lengthy and costly cycles, you buy a "soft circuit" from Altera and similar companies (Xilinx comes to mind). This more expensive device can be reprogrammed on the spot to assume a different function, or correct the logic in the previous iteration. Pay more and get functioning hardware sooner, without slow and costly turns through a manufacturing process.
With this in mind, what Intel will someday manufacture for Altera isn't the 64-bit ARM processor that excited some observers: "Intel Makes 14nm ARM for Altera". The Stratix 10 circuits Altera contracts to Intel manufacturing are complicated and expensive ($500 and up) FPGA (Field Programmable Gate Array) devices where the embedded ARM processor plays a supporting, not central, role. This isn't the $20-or-less price level arena in which Intel has so far declined to compete.
Manufacturing chips for Altera might simply be work-for-hire, a quick buck for Intel, but I doubt it. Altera's yearly revenue is just shy of $2B; Intel is a $50B company. The newly announced device, just one in Altera's product lines, will not "move the needle" for Intel — not in 2014 (the ship date isn't specified), or ever.
Instead, I take this as a signal, a rehearsal. 250M ARM SoCs at $20 each would yield $5B in revenue, 10% of Intel's current total...
This might be what Krzanich had in mind about when he inked the "small" manufacturing agreement with Altera; perhaps he was weighing the smaller margins of ARM processors against the risk of slowing PC sales.
Graciously freed from the past by his predecessor, it's hard to see how Intel's new CEO won't take the plunge and use the company's superb manufacturing technology to finally
make ARM processors.