This article was taken from the October 2012 issue of Wired magazine. Be the first to read Wired's articles in print before they're posted online, and get your hands on loads of additional content by <span class="s1">subscribing online.
Some of the biggest bets in the computing industry were made on the fifth floor of Intel's Robert Noyce Building, on the north-east corner of the chip giant's main campus in Santa Clara, California.
It was here that CEO Andy Grove cemented the "Wintel" partnership that would dominate the PC industry for more than a decade. It was here in 2005 that the newly installed CEO Paul Otellini announced an agreement with Steve Jobs, bringing Apple computers into the fold.
Despite these gambles, a staid, conformist ethos extends throughout much of Intel's headquarters.
The office decor on the fifth floor consists of a boring grey matrix of chest-high cubicles set against walls emblazoned with an ugly dusty-rose-coloured stripe. Khaki-clad engineers filter silently through the aisles with identical black laptops, filing into the windowless conference rooms where they spend much of their time. In all, Intel's offices seem to have as much spontaneity as one of its sterile chip-making factories.
Until you get to the pirate flags.
The flags, which hang on a row of cubicles near Otellini's desk, mark the territory of Mike Bell, whom Otellini hired in 2010 to make Intel the phone industry's premier chip provider. An alumnus of Apple -- where the Mac team famously flew a Jolly Roger of its own -- Bell is a rebellious presence in the buttoned-up confines of Intel. In an office filled with hushed-toned, Supercuts-coiffed engineers, Bell speaks quickly and laughs often, brushing back his mop of reddish-brown hair. And then there's his wardrobe, which seems to consist entirely of Hawaiian shirts. (Today's selection is a navy-blue number with an outrigger canoe emblazoned on the back.) "As long as it's not illegal, Paul's given me permission to break any policy," Bell says. "I don't like lots of meetings. We make decisions and we go." At a place like Intel, that's tantamount to anarchy.
But then, that's the point. Otellini chose Bell because Intel's usual modus operandi, which had helped it utterly dominate the PC-microprocessor market, had left it shut out of the phone industry. Intel had built its business by predictably releasing increasingly powerful chips that could accomplish ever-more-impressive tasks. Indeed, Gordon Moore-- who coined the eponymous law -- was one of the company's cofounders.
The chipmaker's control of its industry was so absolute that the US Federal Trade Commission, California chipmaker Advanced Micro Devices, the State of New York, the European Union and South Korea accused it of anticompetitive tactics. The EU eventually levied a $1.3 billion (£833 million) fine against the company for anticompetitive behaviour.
That, however, was before the iPhone, the iPad, Android phones and tablets hijacked the future of computing.
Suddenly consumers didn't need high-powered machines that could deliver the latest 3D gaming experience at over-clocked speeds.
They wanted portable, energy-efficient devices that could run lightweight applications without draining the battery.
Unfortunately for Intel, its rivals knew how to build chips for just those kinds of devices. In the mid-80s, British computer company Acorn created what came to be known as ARM (originally Acorn RISC Machine) architecture. Apple partnered with Cambridge-based Acorn to create a chip for its first mobile device, the Newton; the two spun out ARM (tweaked as Advanced RISC Machines), an independent company. It launched in 1990 with Robin Saxby installed as CEO, and was in profit by 1993. Whereas Intel's chips used brute force to make computers run faster -- more transistors performing increasingly more-complicated functions -- ARM focused on writing leaner, more efficient instructions, which required fewer transistors, used less energy, generated less heat, and needed no cooling fans. However, ARM architecture struggled until the arrival of the smartphone in the early 2000s. (Apple sold its stake in the late 90s.) Today, ARM chips are found in more than 95 per cent of mobile phones on the market -- including the iPhone and almost every Android phone. The most advanced smartphones in the world contain multiple ARM processing engines --
Samsung's Galaxy S3 i9300 holds as many as 14.
This represents an existential crisis for Intel; the future of computing is small and battery-powered. As sales of smartphones and tablets quickly outpaced those of traditional PCs, the chipmaker faced obsolescence unless it could become a player in these markets. Meanwhile, developers were writing hundreds of thousands of apps for ARM-powered iPhones, iPads and Android devices -- apps that wouldn't work on an Intel processor -- creating network effects that threatened to shut Intel out of the mobile market for good.
Intel's early smartphone efforts were less than inspiring. Chips based on the company's low-powered Atom architecture -- previously used in netbooks -- were still too power-hungry, and phone manufacturers avoided them. In 2010 Otellini stood onstage at the Consumer Electronics Show (CES) and promised an LG Atom phone by the end of the year. It never came. In 2011 Intel would sell almost 330 million microprocessors, nearly all of them going into traditional PCs and servers. ARM, on the other hand, would sell licences to make 7.9 billion chips, which were deployed in everything from refrigerators to cars, set-top boxes to coffeemakers. ARM chips even began infiltrating the server industry, long seen as an Intel stronghold.
And so in July 2010, three months after the release of the iPad, Otellini brought in Bell. This spring, after an intense two-year sprint, the first consumer phones to use Intel processors went on sale in India, China and Europe. By the end of this year, Motorola Mobility, now owned by Google, is expected to launch its own Intel-powered phone.
If Intel can establish itself successfully in the phone industry, it will impact far more than just its shareholders. As with all of Intel's big moves, billions of dollars and the future of the company -- indeed, the very future of computing -- hang in the balance. So far, mobile semiconductors have been improving at a healthy pace, by some measures even besting Moore's law, all while extending battery life. But if Intel becomes a true competitor, mobile devices could get a lot smarter -- and a lot more quickly. No other chip company has Intel's track record of innovation or the same resources to pour into R&D.
Rivals will have to match Intel's furious pace or get swept aside.
Which means that in just a couple of years, that iPhone in your pocket could look like the mobile equivalent of a TRS-80.
Bell describes himself as just "a phone guy". That can sound like modesty, surrounded as he is by semiconductor wonks who have dedicated their lives to understanding the finer points of Intel's x86 chips. They have excelled at performing a kind of engineering magic --finding new ways to pack ever-more transistors into ever-smaller packages. Bell, on the other hand, just knows how to build cool phones. He spent 17 years as an engineer at Apple, where he worked on software for the Mac and the original iPhone before decamping to Palm, where he led hardware development of its well-received but ill-fated Pre and Pixi devices.
But a phone guy is just what Intel needs. For years the company focused so intently on making its semiconductors pack more processing punch that it failed to understand that the mobile market required a different approach.
Even when Intel tried to go small, it still went too big. Compared with ARM's, its initial Atom chips were electricity hogs. They didn't work in phones, so manufacturers crammed them into Frankenstein devices such as the Lenovo IdeaPad https://www.pocketables.com/2008/08/lenovo-ideapa-1.html<span
class="s2">U8 and Aigo P8860D. They both flopped.
At Apple, on the other hand, Bell learned that there's an art to integrating hardware and software, especially in mobile devices. It means designing phones to accomplish only the specific tasks that customers expect, using only as much power as is absolutely necessary.
The goal isn't just to sell as much silicon as possible but also to please phone customers -- and by extension the carriers who sell to them.
Otellini hired Bell to make low-powered chips that could compete with ARM, and he gave the engineer free rein to hire whomever he wanted and use whatever resources he needed to get the job done. "We expect a different approach out of you guys, and we expect you to do things from an outsider's perspective," Otellini told Bell.
Bell accepted the challenge and upped the ante: not only would he build a low-powered chip, but he'd also build a phone to put it in. It would mean a huge amount of work -- software and hardware integration, testing, and certification -- especially for a chip company with little phone experience. But Bell felt he had no choice. If he just produced another Intel chip, nobody would trust that it would work any better than the power-hungry Atoms had. But if he could show carriers a finished phone, they'd have to believe him. "When a carrier or a handset manufacturer asks for benchmarks, we can say,
'Here's the phone; run them yourself,'" Bell says. "It was the fastest way for us to demonstrate real progress."
Bell built his team by snagging people from his old employers Apple and Palm, as well as from rival chipmaker Qualcomm -- a powerhouse in the phone-chip industry that licenses ARM tech and builds on it. He hired a huge number of Android software developers to adapt Google's open-source mobile operating system so it could run apps on the Intel phone. One month after Bell was hired, Intel paid $1.4 billion (£870 million) to acquire German chipmaker Infineon Technologies' wireless division (which uses ARM designs for its communications chips). Then, in a conscious move to insulate his team, Bell piled everyone into an unmarked Intel building a few miles from headquarters and loaded it with junk food, caffeinated drinks and couches. Soon after, he raised the pirate flag. Just so nobody would miss the point, he installed pirate flags at his desk near Otellini's.
The phone team operated like a typical startup, its members toiling through nights and weekends, collapsing on to couches after working 36 hours straight. In just over a year, they had built a reference design phone that, with only a few tweaks, carriers could manufacture and sell directly to their customers. In January 2012, Otellini showed it off at CES, announcing that Lenovo had agreed to produce the phone in China.
Called the K800, it included a camera that could fire off 15 photos in less than a second and a software fix that enabled it to run the majority of Android apps -- meaning that most developers wouldn't need to rewrite their software for the Intel handset. Perhaps most importantly, Otellini announced that other customers had already signed on to build their own phones using Intel's new chips.
Another Intel phone, the Xolo X900, was released in April 2012 by Indian handset company Lava, beating the K800 to market by a month.
European carrier Orange recently began selling its San Diego model in the UK and France.
Reviews started coming in. Bell's phones had the processing power that consumers had come to expect from Intel. They were above average by almost every measure -- energy efficiency, display, call quality and processor speed. They were no match for the iPhone --these were just the first round of Intel phones, after all, and were meant to be modest -- but they were good value.
Orange's San Diego sells for about £200 in the UK versus around £200 to £400 for a comparable Samsung or HTC phone.
To celebrate the release of the phones, Bell threw a party in the Intel HQ courtyard. A band played songs by his favourite groups, The Cure and New Order. No one could remember when Intel last served beer during a workday. Nor could they recall a time so many people had worn Hawaiian shirts.
But Intel may not want to celebrate too hastily. The company seems to have revamped its engineering philosophy, but to compete in the mobile industry it will also need to adjust its business tactics. Intel's empire has run on a fairly traditional model: manufacture pricey chips, then sell them at high margins. ARM, on the other hand, licenses its architecture and designs to other chipmakers who add radios, video processors and other circuitry and components. Some companies, such as Apple, design customised chips to use in their own devices.
Others, such as Nvidia and Qualcomm, make them to sell to other computer and phone manufacturers. ARM gets a licensing fee and a royalty of a few per cent per chip. ARM chips outsell Intel microprocessors 24 times over, but Intel's revenue is 68 times higher.
Because of its dominant market position, Intel can sell a Xeon server chip for more than £500, raking in a gross profit of about 78 per cent. But as Intel moves into mobile, it will have to accept much thinner profit margins and focus on volume. The main chips in smartphones and tablets today sell for about £5 to £15, about 40 per cent of which is gross profit.
And although Intel has come up with a hack that allows most Android apps to run on its phones, that's not an ideal fix. The company will have to persuade app developers to optimise their software for Intel's new smartphone chip, which is still based on x86 architecture. Appmakers already have to write software for Android and Apple's iOS if they want their products to run on both platforms. It may be a struggle to get them to add yet another rewrite for Intel-powered chips. "Does the industry want another supplier?" asks Warren East, CEO at ARM. "We do expect Intel to be in this marketplace. We just don't expect it to have a massive share."
Neither do Intel executives, who keep insisting that they're merely looking for a seat at the table. But Intel's decades-long history suggests that the company doesn't usually remain satisfied until it claims every seat -- and then the table itself. Grove instilled a culture of intense competitiveness, seeing the industry as a zero-sum game. Any chip sold by someone else represented a lost opportunity. A rival's breakthrough design could freeze Intel out of an entire line of computers. And Intel's model of making massive investments -- it builds a new £3 billion fabrication plant every five years or so - has required it to sell billions of microprocessors annually. The chipmaker has proven willing to do whatever it takes to make that happen. Just ask the companies and territories that allege every kind of hardball tactic from Intel over the years: offering steep rebates to customers who agree to buy 95 per cent of their chips from Intel; paying PC makers to delay the introduction of machines that use rival microprocessors; bundling and discounting products to undercut its competition.
Then again, carriers and phone makers might welcome another competitor -- even one as historically rapacious as Intel. Manufacturers such as Samsung and Apple charge carriers outrageous prices for the privilege of carrying their devices. But Intel offers almost complete phones, letting carriers customise the software and put their own brand on the hardware.
Meanwhile, less illustrious manufacturers find themselves at the mercy of the chipmakers -- Qualcomm, for instance, dominates the smartphone market, capturing 50 per cent of all revenue, which gives it the power to squeeze phone makers such as HTC and LG for the right to use its technology.
Intel could offer a welcome counterbalance. "Its arrival ensures that nobody in the mobile industry can rest on their
successes," says Patrick Remy, vice president of devices at Orange.
And then there's Intel's proven expertise at process technology -- figuring out ways to reliably innovate and improve its chips. The company's current phone semiconductor -- a single-core chip called Medfield -- works fine, even if it can't yet best efforts from Nvidia, Qualcomm and Samsung. But next year Intel is expected to come out with a chip that crams an even greater number of transistors on to an even smaller frame. That will mean both more processing power and greater efficiency. (Experts expect the chip to have roughly double the performance of Medfield.) And by 2015, Intel plans to produce a still smaller chip -- about half the size of Medfield yet, experts say, potentially almost four times as powerful and efficient.
At that point, some analysts predict, Intel's offerings may become so superior that manufacturers and carriers will have no choice but to switch to them. "The goal is ultimately to get inside Apple," says Patrick Moorhead, a longtime AMD executive turned analyst. "And Intel will pull out every stop to get there." This rising tide could also enable more and more connected devices -- thermostats, medicine-bottle caps, door locks, water heaters -- all running tiny, powerful, energy-efficient Intel chips.
There are plenty of ARM fans who believe that Intel may still stumble in its phone bid -- not because of a lack of technical chops but because the inertial pull of its old PC business practices and a dependence on fat margins will trip up the company.
Bell has heard all the criticism and brushes it off.
Most phone and tablet users don't care what chip is powering their favourite device, as long as it's fast and doesn't drain the battery quickly. "The best product wins," says Bell. "If we come out with the best products, we deserve to have significant market share." Or, as a pirate might say, hand over those doubloons.
Michael V Copeland (@MVC) oversees <span class="s1">business coverage on wired.com in the US
This article was originally published by WIRED UK
