We know Intel shunned ARM processors and played virtually no role in the smartphone revolution. But we now learn Steve Jobs asked Intel to build the iPhone microprocessor. Paul Otellini, Intel's departing CEO, admits he should have followed his gut - and made the smartphone world a very different place.

CEO valedictions follow a well-known script: My work is done here, great team, all mistakes are mine, all good deeds are theirs, I leave the company in strong hands, the future has never been brighter… It's an opportunity for a leader to offer a conventional and contrived reminiscence, what the French call la toilette des souvenirs (which Google crudely translates as toilet memories instead of the affectionate and accurate dressing up memories).

For his farewell, Paul Otellini, Intel's departing CEO, chose the interview format with The Atlantic Monthly's senior editor Alexis Madrigal. They give us a long (5,700+ words) but highly readable piece titled Paul Otellini's Intel: Can the Company That Built the Future Survive It?

[caption id="attachment_5538" align="alignnone" width="460"] Photo: Guardian.co.uk[/caption]

The punctuation mark at the title's end refers to the elephantine question in the middle of Otellini's record: Why did Intel miss out on the smartphone? Why did the company that so grandly dominates the PC market sit by while ARM architecture totally, and perhaps irretrievably, took over the new generation of phones -- and most other embedded applications?

According to Otellini, it was the result of Intel's inertia: It took a while to move the machine.

Madrigal backfills this uneasy explanation with equal unease:

"The problem, really, was that Intel's x86 chip architecture could not rival the performance per watt of power that designs licensed from ARM based on RISC architecture could provide. Intel was always the undisputed champion of performance, but its chips sucked up too much power. In fact, it was only this month that Intel revealed chips that seem like they'll be able to beat the ARM licensees on the key metrics."


Note the tiptoeing: Intel's new chips "seem like" they'll be fast enough and cheap enough. Madrigal charitably fails to note how Intel, year after year, kept promising to beat ARM at the mobile game, and failed to do so. (See these 2010, 2011 and 2012 Monday Notes.) Last year, Intel was still at it, dismissively predicting "no future for ARM or any of its competitors". Tell that to ARM Holdings, whose licensees shipped 2.6 billions chips in the first quarter of this year.

Elsewhere in the article, Otellini offers a striking revelation: Fresh from anointing Intel as the microprocessor supplier for the Mac, Steve Jobs came back and asked Intel to design and build the CPU for Apple's upcoming iPhone. (To clarify the chronology, the iPhone was announced early January, 2007; the CPU conversation must have taken place two years prior, likely before the June, 2005 WWDC where Apple announced the switch to x86. See Chapter 36 of Walter Isaacson's Jobs bio for more.)

Intel passed on the opportunity [emphasis mine]:

"We ended up not winning it or passing on it, depending on how you want to view it. And the world would have been a lot different if we'd done it, […]


Indeed, the world would have been different. Apple wouldn't be struggling through a risky transition away from Samsung, its frenemy CPU supplier; the heart of the iPhone would be Made In America; Intel would have supplied processors for more than 500 million iOS devices, sold even more such chips to other handset makers to become as major a player in the smartphone (and tablet) space as it is in the PC world.

Supply your own adjectives…

Indulging briefly in more What If reverie, compare the impact of Intel's wrong turn to a better one: How would the world look like if, at the end of 1996, Gil Amelio hadn't returned Apple back to Steve Jobs? (My recollection of the transaction's official wording could be faulty.)

So, again, what happened?

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."


A little later, Otellini completes the train of thought with a wistful reverie, a model of la toilette des souvenirs:

"The lesson I took away from that was, 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," he said. "My gut told me to say yes."


The frank admission is meant to elicit respect and empathy. Imagine being responsible for missing the opportunity to play a commanding role in the smartphone revolution.

But perhaps things aren't as simple as being a "gut move" short of an epochal $100B opportunity.

Intel is a prisoner of its x86 profit model and Wall Street's expectations. It's dominant position in the x86 space give Intel the pricing power to command high margins. There's no such thing in the competitive ARM space, prices are lower. Even factoring in the lower inherent cost of the somewhat simpler devices (simpler for the time being; they'll inevitably grow more complex), the profit-per-ARM chip is too thin to sustain Intel's business model.

(Of course, this assumes a substitution, an ARM chip that displaces an x86 device. As it turns out, the smartphone business could have been largely additive, just as we now see with tablets that cannibalize classical PCs.)

Another factor is the cultural change that would have been required were Intel to have gotten involved in making ARM devices. As both the designer and manufacturer of generation after generation of x86 microprocessors, Intel can wait until they're good and ready before they allow PC makers to build the chips into their next products. The ARM world doesn't work that way. Customers design their own chips (often called a System on a Chip, or SoC), and then turn to a semiconductor manufacturer (a foundry) to stamp out the hardware. Taking orders from others isn't in Intel's DNA.

And now?

The answer might lie in another French expression: L'histoire ne repasse pas les plats. Google Translate is a bit more felicitous this time: History does not repeat itself. I prefer the more literal image -- History doesn't come around offering seconds -- but the point remains: Will there be seconds at the smartphone repast?

Officially, Intel says its next generation of x86 processors will (finally!) topple the ARM regime, that their chips will offer more computing might with no cost or power dissipation penalty. In their parlance "the better transistor" (the basic unit of logic processing) will win.

I doubt it. The newer x86 devices will certainly help Microsoft and its OEMs make Windows 8 devices more competitive, but that won't prevent the spread of ARM in the legion of devices on which Windows is irrelevant. For these, Intel would have to adopt ARM, a decision Otellini has left to the new tandem leadership of Brian Krzanich (CEO) and Renée James (President). Will they stick to the old creed, to the belief Intel's superior silicon design and manufacturing technology will eventually overcome the disadvantages of the more complex x86 architecture? Or will they take the plunge?

They might be helped by a change in the financial picture.

In 2006, that is after throwing Jobs in Samsung's arms (pun unintended), Intel sold its ARM business, the XScale line, to Marvell. The reason was purely financial: for similar capital expenditures (costly fabs), ARM processors achieved much lower per-unit profit, this because of the much more competitive scene than in the x86 space.

Now, if Intel really wants to get a place at the smartphone table with new and improved x86 devices, the company will have to price those to compete with established ARM players. In other words, Intel will have to accept the lower margins they shunned in 2006. Then, why not do it with the ARM-based custom processors Apple and others require?

-- JLG@mondaynote.com

----------------------------

(I'll confess a weakness for The Atlantic and, in particular, for its national correspondent James Fallows, a literate geek and instrument-rated pilot who took upon himself to live in Beijing for a while and, as a result, can speak more helpfully about China than most members of the Fourth Estate. Going back to last week's reference to the Gauche Caviar, when my Café de Flore acquaintances fall into their usual rut of criticizing my adopted country for its lack of "culture", I hold out that The Atlantic -- which sells briskly at the kiosk next door -- is one of many examples of American journalistic excellence.


And, if you're interested in more strange turns, see this other string Alexis Madrigal piece in the same Atlantic: The Time Exxon Went Into the Semiconductor Business (and Failed). I was there, briefly running an Exxon Information Systems subsidiary in France and learning the importance of corporate culture.)--JLG


--



Print Friendly