When microprocessor core developer ARM started in a barn outside Cambridge, England, just over fifteen years ago, odds were against it making a global impact. The team of “12 engineers and me”, as then CEO and now chairman Sir Robin Saxby puts it, “had no patents, a working prototype and £1.75m of cash.”
Without the funding to build a fab or to pay someone else to make chips they could sell, the issue was how ARM could exploit what it did have – a proven design for a small, fast RISC microprocessor core, and bucketloads of ambition.
“How could we become a global standard with so little money?” Saxby says. “We had to turn potential enemies into friends, by making them partners.”
“We expect to sell around 2 billion ARM chips next year and are predicting we will sell 4.5 billion per year by 2010.”
ARM now has 165 semiconductor partners, ranging from Intel to SMIC, of which 62 are shipping parts that include ARM intellectual property (IP). It has grown, gone public, survived the dotcom boom and bust, and now employs around 1300 people at 34 locations worldwide. After a slow start, waiting five years for its first royalty revenues, ARM cores are now used in billions of its partners’ chips each year.
“We’ll sell 1.7 billion ARM chips this year, up from 1.3 billion last year,” says Saxby. “We expect to sell around 2 billion ARM chips next year and are predicting we will sell 4.5 billion per year by 2010.” Having achieved a global standard, the question facing Saxby and his colleagues is what does ARM do next.
Present and correct
ARM’s initial growth was driven by a talented engineering team that tackled everything from IC layouts to operating system and applications software. As the resultant core was licensed, the team grew and diversified, diluting that original, intensively co-operative development environment. The response has been to break the organization down into smaller units that can work closely together.
“We have several ‘mini ARMs’ within ARM now,” says Saxby. The microprocessor cores division houses the original ARM activity, of core design. ARM now also has a physical IP group, which includes the result of its controversial acquisition of Artisan Components. Saxby defends that acquisition, which many analysts thought cost too much and will pay back too slowly, as an early stake in an inevitable progression to the outsourcing of physical library development.
“The majority of ARM semiconductor partners have hundreds of library developers working on physical libraries,” he says. “We think these libraries will become a global standard – it’s an economic argument with time. But to get a customer to use our physical libraries takes two to three years, and we’re only threequarters of a year into our acquisition of Artisan.”
Saxby points out that ARM’s early royalty revenues took years to arrive and suffered sharp peaks and troughs when products, such as fax modems, failed to take off.
“The libraries are in some video games, but there isn’t enough of a community yet to take the spikes out of that revenue,” he says. Six of ARM’s semiconductor partners have bought the Artisan libraries so far.
ARM also now has a ‘fabric’ group, that develops the bus standards and other infrastructure elements that ease the creation of systemson- chip. Other ‘mini ARMs’ offer development tools and supporting software to an ever-widening customer base.
“In design tools it is a case of doing more of that, and better,” says Saxby. ARM bought Keil, a German company, in October 2005 for its suite of software development tools for embedded microprocessors. It also turns to partners to enable and channel the use of its IP.
“The EDA industry is very important to us since our customers need those tools,” Saxby says. “As ARM we want to pick partners where there is a win/win. Each vendor has one tool that is best within its sector, and some other tools. But most of our semiconductor partners use many tools from various vendors.”
Saxby believes the real bugbear for the semiconductor industry, though, is the time it takes for customers to develop and prove operating system and applications software.
“The problem with getting a product out is the cost and challenge of software development,” he says. So he wants to sort it out: “As a vision for ARM over the next 10 to 15 years, I’d like to feel we have influenced the software industry in the way that we have influenced the semiconductor industry, helping it become more economic and effective through standardization.”
Saxby points to two pieces of ARM intellectual property that rely on software as much as hardware. The first, TrustZone, provides security features to service operators and those who want to protect media or other forms of valuable content. The second, Intelligent Energy Manager, helps tackle power consumption. “At deep submicron dimensions, transistor leakage versus speed versus cost is important. So you need the Intelligent Energy Manager offering to connect to software and to the design tools,” he says. This is also where the Artisan libraries come in, providing fundamental building blocks for a chip design that can be tuned for minimum power consumption or maximum performance, as appropriate.
The benefit for ARM, of course, is that such solutions get their customers to market faster and add another piece of royaltybearing ARM IP to each chip.
Each of these technologies was developed to serve what Saxby calls “the explicit needs of the customers”.
“Working with Texas Instruments to satisfy Nokia’s needs helped get us started, as did working with LSI Logic to meet Western Digital’s needs and with Sharp to satisfy Nintendo’s needs. Now we need to understand the explicit needs of, say, Vodafone, France Telecom and Universal Music,” Saxby says.
This means ARM also needs to understand the specific enablers for each customer, such as the iTunes service that enabled demand for the Apple iPod or the GSM standard that enabled the mobile phones industry.
Saxby sees personal healthcare as a major enabler of demand over the next 10 to 15 years.
“In 15 years’ time I think the equivalent of today’s mobile phones will be helping to keep us alive,” he says.
He points to work being done by companies such as GE and Amersham International on health devices that can help monitor individual health, control chronic diseases and manage drug delivery.
“That bringing together of sensors and biotech will be a major wave in the next 15 years,” he says. “This world is definitely happening and will come.”
Unsurprisingly, Saxby believes that the ARM core, with its small size, and good combination of processing power and power consumption is the “ideal” architecture for these healthcare applications, although he acknowledges they will take some time to emerge.
“I’m not saying that within my vision of ARM that will contribute to major revenue – it’s further out than that,” he says. “You could say it’s just a bit of Robin nonsense — just like when I said we’d be the global RISC standard.”
What isn’t nonsense is the increasing globalization of ARM, its customer base and its customers’ customer base. ARM rotates its biannual board meetings around its office locations, visiting Bangalore, Shanghai, California and other cities around the globe to keep close tabs on the local business and to fully understand regional customer needs.
That globalization is also already being reflected in changes to the Institution of Electrical Engineers, a UK charitable and educational body of which Saxby is president elect.
“In reality we are a global organization with a larger part of the membership in the UK,” says Saxby. Many of the members outside the UK are in former Commonwealth countries, or have studied in the UK before returning home.
“As the IEE goes forward, where are the engineers? They’re in China, where we have already put an IEE staff member, and in India,” notes Saxby.
“Part of the vision for the IEE is to make it more relevant for the 21st century and a virtual knowledge network for the planet.” It’s an ambitious vision. But then, Saxby has already shown he can deliver on those.