Start me up

Starting up a technology firm is not a quick route to riches, it's hard graft folks. Richard Wilson spoke to an expert - Robin Saxby head of ARM

You have to make your product obsolete before the competitor does it for you,"declares the head of Advanced Risc Machines (ARM), but Robin Saxby's philosophy could be taken as sound advice for any of the start-ups appearing on these pages.

Saxby knows all about developing an electronics start-up company. In his own words he knew that he had to "run the company very mean and lean" when he was drafted in from Motorola to create a microprocessor business for Acorn Computer, licensing the Cambridge-based company's ARMRisc microprocessor technology. "For about the first year I was VP of everything, with the engineering team focused on getting the chip designs out. We got to break-even in year two and we’ve been generating profit ever since," adds Saxby, who after seven years travelling the world - this man spends more time in a Virgin Exec Club lounge than many of us do in our front rooms - has amassed a list of 20 licensees for the ARM microprocessor architecture.

Only some things have changed since 1990; Saxby now markets not one but a bagful of microprocessor designs - "Right now there are about a hundred bits of silicon coming out of ARM,"points out Saxby. Another thing that has changed is that the ARM licensees have become "partners".

Saxby's business model is deceptively simple. First create a Risc microprocessor core that competes in processing performance, power consumption and die size with leading industry processor architectures like MIPS and Hitachi's SuperH . Then make it an open standard architecture by licensing the basic design and customised variants of it to some of the world's leading semiconductor makers. "ARMwas the basic building block which we made available to the semiconductor guys. The trick is how do they and ourselves make money out of it," says Saxby.

For start-ups hoping to make a quick killing in the market with an innovative product, Saxby's experience with ARM is informative and salutary. ARM's partners are designing a handful of versions of the basic processor cores: ARM 6, 7, 8 and StrongArm into a wide and increasing range of end applications. These range from the original Acorn PCs and Apple's Newton to today's 56k fax modems, GSM mobile phones and cars. But Saxby has yet to find that "killer application" which will result in ARM chips pouring out of semiconductor plants in numbers to match Risc market leader MIPS R4000 and Hitachi's SuperH families which have cornered the games machine market between them.

According to Saxby, the lack of a volume market for ARM is not a problem. The company is growing at a healthy rate of 78 per cent per year on income from its licensing deals, and a growing list of ancillary businesses which includes design consultancy, application software development, design tools and training.

Saxby is also quietly confident that big volumes for one of the ARM processors will come eventually, and in 1997 they may not be very far away. "Fax modems chips have been shipping for the last two quarters and the first mobile phone design-in volumes will start by the end of the year. Then it will be personal organisers and PCperipherals," says Saxby.

The applications for ARM do not stop there. By the year 2000, Saxby predicts that there will be more ARM chips in cars than any other product. ARM's current list of partners covers most markets but Saxby cannot stand still. He is continually "obsoleting" his processor cores with new designs with a high performance StrongARM and application specific devices for network computers and GSMphones. To stay ahead of MIPS in processing performance, work combining a DSPcore with the ARMThumb processor for next generation mobile phone designs is underway. "Things are coming up to beat Thumb so we need to stay ahead," says Saxby.

Half joking, Saxby says that the CPUcore is the easy bit; what is less so is developing the right hardware and software peripherals to go around it. To do that, Saxby knows he must continue to second guess the developments of more than one end market for the ARMprocessor.

Seven years down the line, with a possible public share offering this year, there is no suggestion of plain-sailing for Saxby. Anyone thinking a technology start-up was the quick route to Bill Gates-style riches should think again.

This article first appeared in Electronics Weekly,
Wednesday 14th May 1997