The conventional approach in the computer business is to make the machines first, then sell them. But a Lexington start-up, Adapteva , has rung up nearly $900,000 in sales of a computer it hasn’t yet made.
“We have a market. We have a customer base and everything else,” said founder and chief executive Andreas Olofsson. “The stressful part is we’ve already taken their money and now we have to go build them.”
To raise the money for his company’s new computer, called Parallella, Olofsson turned to Kickstarter, the online “crowdfunding” service that lets ordinary people invest money like venture capitalists. Visitors to the Kickstarter website can pledge small amounts of money to help launch businesses that appeal to them. Investors don’t get a stake in the company; instead, they get the satisfaction of backing a bright idea, and, in this case, a chance to own one of the first Parallella computers.
Parallella won’t cost much, just $99. But despite the attractive price, it may not be best suited for the typical home user. Instead, the target audience is sophisticated computer users, such as scientists, researchers, and technology developers, who can handle the ad-hoc aspects of Parallella. The “computer” is really just a bare-bones circuit board about the size of a credit card, with connectors for a keyboard, hard drive, and network connection. An extra $20 buys a case.
The first units will be shipped this month, and production is scheduled to ramp up through the spring.
Among the first recipients: Kickstarter donors who put up $99 or more will receive a Parallella computer. Olofsson fell far short of his goal of raising as much as $3 million on Kickstarter, but the company got investments from more than 4,900 people around the world.
“We’d been struggling for years to get 10 customers,” Olofsson said. “All of a sudden you see people come toward you . . . it was very rewarding.”
Olofsson, a former chip designer at Analog Devices Inc. in Norwood, launched Adapteva more than four years ago to design very efficient multicore processors. Each core on the chip can act as an independent computer, or it can share work with other cores, enabling the computer to complete a task much faster. Today’s home computers generally have processors with four cores; the current Adapteva chip contains 16. A 64-core version has been developed for future production.
The new chip works alongside a standard A9 processor designed by ARM Ltd., a major producer of low-power processors for smartphones and tablet computers.
The Adapteva chip is extremely small — about one-tenth of an inch wide. At that size, it leaves out many circuits that other chips use to run operating system software or perform other tasks; instead Adapteva leaves such work to the A9 and focuses on specialized tasks such as sorting through vast databases in search of a single piece of information. By specializing, the chip can be made much smaller and more energy-efficient.
Jeff Milrod, chief executive of Bittware Inc., a Concord, N.H., maker of specialty computers for military and commercial users, was a fan of Adapteva even before the company turned to Kickstarter. Bittware invested $1.5 million in 2009.
“This was a chip that I thought was just incredibly awesome,” Milrod said, “and we had to have it.”
Massimiliano Versace, director of the neuromorphics lab at Boston University, is trying to build robots that will use processors modeled after the human brain. Versace invested $100 in Adapteva because he believes the company’s new microchip is ideal for his purpose.
“We need very, very tiny chips which suck very, very little power,” Versace said. “I’m a strong believer in what Andreas is doing.”
Olofsson hopes to see Adapteva chip technology incorporated into cellphones, where it could deliver a major boost to processing power without draining batteries.
“People always need more performance,” he said.