Chips
The Unlikely Tale of How ARM Came to Rule the World
By Ashlee Vance
This is a story about ARM Holdings (ARMH), the mobile technology company. But before it gets going, here are a few things you need to know:
1. ARM is a company made up mostly of chip engineers. They design parts of chips—such as graphics and communication bits—and they design entire chips.
2. ARM sells these designs and licenses its chip architecture to dozens of companies, including Apple (AAPL), Samsung Electronics (005930:KS), Qualcomm(QCOM), and Nvidia (NVDA).
3. As a result, just about every smartphone, mobile phone, and tablet runs on an ARM chip.
4. In fact, you can argue that ARM-based products are now the most-used consumer products in the world, outflanking even Coca-Cola (KO) and McDonald’s(MCD) by some measures. (I recently made just such an argument.)
5. A great many people have not heard of ARM. This is because the company has largely kept to itself from a headquarters in Cambridge, England.
6. Its anonymity is sort of incredible when you think about it, considering that ARM has arguably had a more profound effect on modern living than just about any other company.
Picture an idyllic scene in the English countryside—like Downton Abbey, but bigger. That’s Cambridge. It has these ancient college buildings that are cathedrals to learning. On a warm day, students stream out and head to the river to go punting or get a pint and lounge around in grassy fields. It’s a place that begs you to get lost in thought.
Near the center of town, right between a Gap (GPS) and a Marks & Spencer(MKS:LN), is a dark alleyway that’s encrusted with bird poop. It’s in this alley that Acorn Computers got its start in 1978.
The young and/or non-British may not remember Acorn. The important thing is that it made computers when PCs were just starting to appear. In those days, a company with an idea for a computer couldn’t just order up a batch from Foxconn(2317:TT). No, it had to have real engineers to design the parts. Acorn had those: It had a team of young, clever electrical engineers plucked from the local schools and electronics companies. As Acorn endured financial ups and downs over the years, it came to the conclusion that its chip team was a luxury it could no longer afford. The company’s management twice tried sell its chip design department only to have the deals fall through. “We felt quite unloved, really,” says John Biggs, one of Acorn’s chip engineers.
Then, in 1990, a new and determined customer arrived: Apple. The Mac maker had developed plans for a handheld computer and wanted a breakthrough microprocessor for the device. Apple, Acorn, and VLSI Technology, a chip design tools maker, formed a joint venture to build the silicon. Twelve Acorn engineers were picked to staff the company, which they named ARM. It’s an acronym within an acronym: Advanced RISC Machines—RISC stands for “reduced instruction set computing.”
ARM established its headquarters in Swaffham Bulbeck, a village 8 miles from the center of Cambridge. The office was a converted 18th century barn. The first CEO, Robin Saxby—now Sir Robin Keith Saxby—managed to outfit the place with almost no budget. He won ARM’s first boardroom table in a coin toss with a local furniture salesman, then tried to win a set of desk drawers through a game of billiards at the pub. (He lost, and the engineers went without drawers for some time.)
The design specifications for Apple’s handheld computer called for an affordable product that could fit in someone’s hand, yet still have the computing muscle to cope with handwriting recognition and other advanced functions. “There was a need to be cheap, and being cheap meant making a small chip, and that meant a low amount of power would be sent through it,” Biggs says. At a time when the chip industry was focused on making bigger, faster products, ARM went the opposite direction and became the low-cost, low-power specialist. “It was really a happy accident,” he says.
Apple named the handheld Newton, and it flopped. But ARM’s chip caught the attention of companies looking to make small and sophisticated devices. In 1993,Nokia (NOK) hatched plans for a groundbreaking mobile phone. Unlike the rudimentary brick phones of the time, Nokia’s would have an actual menu with icons and basic games and would be aimed at businesspeople. Nokia picked ARM for this device, which would emerge a couple of years later as the Nokia 6110, one of the first blockbuster handsets. “Once the royalties started kicking in for that product, things took off,” Biggs says.
Today, ARM’s campus, built on top of Roman archaeological remains, is in farm country about a 15-minute drive from the center of Cambridge. On weekdays, employees—the vast majority are men—arrive by the dozens and gather for breakfast in a sunlit atrium in the company’s main building. The lads then dine on bacon and eggs pumped out by a street-vendor-size cafeteria before heading off to some drab cubicles.
ARM is basically a company of chip engineers. The business model it invented is simple: Tech companies shouldn’t have to reinvent the wheel every time they need a new chip for a product. Instead, they can look over ARM’s roster of chip parts, buy some basic things, and then do a bit of extra custom design work of their own to make their product unique. Other companies have similar models, but none offer the product breadth of ARM, which has chips that can run pretty much anything that has a computer, from coffee pots to data-center servers and everything in between.
About half of ARM’s revenue comes from mobile products, while the rest comes from chips that go into TVs, media players, sensors, cars, printers, and other gear. Companies pay ARM in two basic ways. The first is a kind if all-you-can-eat subscription to ARM designs. For about $10 million per year, a major player like Samsung can pick and choose from the entire chip catalog. The second way is royalties. Mobile phones bring in 20¢ to 40¢ apiece—a bit more for smartphones—and there are pennies or fractions of a penny to be had from other devices. “They innovate like crazy and take just enough back to keep funding that innovation,” says Tony Fadell, the man who brought the iPod to life at Apple and now runs the smart-home appliance maker Nest Labs.
ARM’s top-tier customers such as Apple, Samsung, Qualcomm, and Nvidia employ hundreds of their own chip designers for customization work. Much of Apple’s success has come from the snappy performance of its products and its long-lasting batteries. The features are a direct result of tweaks Apple made to ARM designs, and the ways in which Apple married its software to the silicon. Despite having its own large chip team, Apple still saves hundreds of millions, if not billions, of dollars a year by not having to produce everything from scratch.
The ARM model is based on this idea of spreading risk, R&D investments, and profits. ARM does the underlying dirty work and makes more money when its customers sell more things. Companies like Apple and Samsung get to focus their attention on higher-level innovations instead of grunt work. And the contract chipmakers like Taiwan Semiconductor Manufacturing (TSM) and Global Foundries cater to dozens of ARM customers and can divvy up their orders among factories to keep them running at full capacity.
Representing the exact opposite of the ARM model is Intel (INTC).
Intel designs its own chips, which are widely regarded as the most advanced in the world. At least as important as that: It builds its own chips. This year, Intel will spend about $12 billion on its factories. The facilities, scattered throughout the U.S., Israel, and China, are probably the most cutting-edge manufacturing facilities ever built. They allow Intel to squeeze billions of transistors, or switches, on something the size of a fingernail, and then double that transistor density every 18 months. The process requires regular breakthroughs in physics, materials science, and software.
While many clever and wealthy companies, including IBM (IBM) and Samsung, have tried to keep up with Intel, no one these days comes close. Intel, according to numerous analysts, has a two- to four-year lead over rivals in terms of raw manufacturing smarts and chipmaking techniques. “The other guys are simply not as good at manufacturing as Intel is,” says David Kanter, a chip analyst and industry consultant. Intel is now down to two or three competitors that are willing to make the investment in similarly advanced factories, and they’re all contract chip manufactures that spread their business among many customers.
When Intel’s business is humming, its plants more or less print money. Its chips range in price from $37 to $4,616 each. (ARM chips tend to max out at about $30 and usually cost less than $5.) The problem for Intel is that many investors think its model has tapped out. The company’s stock has been in a coma for more than a decade, with shareholders finding it hard to get excited about the steady stream of profits derived from PC and server chip business. Intel has inexplicably missed out on the smartphone and tablet explosion. Meanwhile, ARM team member Apple has become the most valuable company in the world, and Qualcomm, another member of the ARM clan, has overtaken Intel as the most valuable chip company.
As far as Intel is concerned, the world is tilting in its favor. Chinese smartphone makers have started building devices around Intel’s low-power [JA1] chips, and Samsung, Microsoft (MSFT), and others have Intel-based tablets on the market. As these devices take on more and more PC-like functions, Intel’s pricier, high-performance chips will come into vogue, says Intel spokesman Bill Calder. “Performance does matter very much now, and that is an area of Intel’s strength,” he says.
And yet not that long ago, no one would have dreamt of using a low-power ARM chip in a PC. Now Google (GOOG) has teamed with Samsung, Hewlett-Packard(HPQ), and others to produce ultracheap Chromebook laptops that run on ARM. Dozens of companies, including Samsung and Nvidia, are expected to begin selling ARM chips for servers this year or the next. These new businesses represent a chance for ARM to make much more money per chip and establish itself as the technology standard for everything from simple gadgets to supercomputers. It would be the first company to pull off such a feat.
Intel’s Calder downplays the threats the company faces in the PC and server markets, saying he has seen such competition come and go before. Qualcomm and Microsoft, for example, have tried and failed to sell low-cost, ARM-based computers. In the data center, Intel has a low-power chip that can compete against ARM. “There is a perception and a reality,” Calder says. “The reality is that no large companies are switching over, and they may never.” As for ARM broadening overall beyond its success in mobile, Calder says: “Show me the money. It’s mostly talk. If you take the sheer scope of Intel’s business, I think you would be hard-pressed to argue that ARM Holdings can go after every piece of that and challenge us in every segment.”
Simon Segars took over as chief executive officer of ARM last July. To fulfill his new duties, Segars flew from his home in Silicon Valley to ARM’s headquarters. It’s a journey the 22-year ARM veteran has taken many times, and one that’s been made all the easier because he can crash at his parents’ house in Cambridge. Except this time—with the extra helping of responsibility on his plate—the familiar rhythms gave way to a harried round of on-the-job training.
Over the course of a few days, Segars met with underlings to get the latest information on ARM’s real estate situation, the company’s technology infrastructure, and its marketing campaigns. He picked up a few possessions and moved one door down the hall to the CEO’s office, which had just gotten a fresh coat of paint. He was apprised of the barrage of cable-TV appearances he would need to make and large shareholders who needed coddling. And he was dragged into a conference room to be judged by a group of financial analysts from Lazard Capital Markets. The good news, they said, was that investors largely adored ARM and were in awe of the technology company’s incredible performance in the past few years. Then came the ominous warning. “It’s all downhill from here, Simon,” said one of the Lazard men. To this, Segars issued a chuckle followed by a grimace, as he seemed to weigh how much truth there might be to this playful jab.
Brits tend not to beat their chests about their accomplishments, and British engineers prove even more reserved. It’s basically embarrassing to succeed. Segars has very much been cut from this mold, and big questions remain as to whether the same reserved attitude that has helped ARM to date will serve it as well in future battles. Intel finally seems cognizant of where the computing industry is going and looks prepared to unleash its famous combination of paranoia, hard-line sales tactics, and near-oppressive marketing. If history serves as a guide, ARM can expect to be steamrolled and left for dead just like most of Intel’s competitors.
“I think ARM’s culture is very much a work in progress and that they’re just now gearing up for this challenge,” says Kevin Krewell, a chip analyst with Tirias Research. “Whether or not ARM is really up for this is a good question.”
If the pressure is getting to him, Segars does not show it. He joined ARM as employee No. 16 after having graduated at the top of his class at the University of Sussex with degrees in electronic engineering and computer science. “On my first day, someone was literally soldering a computer together for me to use,” Segars says. He remains fixated on continuing to have engineers—not breathless sales and marketing types—drive the company’s culture.
Yet there are hints that Segars is willing to break with tradition and bring a fresh perspective to the company, to help ARM, in essence, step up its game. Segars is the only member of ARM’s old guard who was willing to leave Cambridge and go to the U.S. to run the company’s overseas operations. Now, for the first time, ARM has a chief who seems to enjoy moving at a faster pace and who has soaked in the frenetic drive of Silicon Valley. Segars has also started hiring a cadre of people who have done battle with Intel in the past, both in the press and in sales, hoping that ARM can develop a nasty streak.
That said, it’s a nasty streak that seems destined to go only so far. Asked if he’ll be the guy who finally manages to take down Intel, Segars says: “Who knows? They’ve got a lot of money. They’ve got a lot of determination. You know, we’ll see what they do.”
Vance is a technology writer for Bloomberg Businessweek in Palo Alto, Calif. Follow him on Twitter@valleyhack.