SAN JOSE, Calif.--It came right down to the wire, but in the end Adapteva’s effort to raise $750,000 on Kickstarter to fund a new mask set for its Epiphany processor was a rousing success. Not only did the startup snag nearly $900,000 after some last–minute pledges, it also seeded a developer community of nearly 5,000 potential customers.
Seven people pledged $10,000 to get early access in December to Adapteva’s $99 reference board. It includes a 16-core Epiphany and a Xilinx Zinq FPGA with an embedded dual-core ARM Cortex A9 processor.
Adapteva expects to ship out nearly 6,000 of the boards to backers, most of them in May. That’s when the startup expects to have silicon based on a new mask that the Kickstarter program was geared to fund.
“We went from having ten-ish customers to thousands over night, so it’s pretty exciting,” said Andreas Olofsson, chief executive of Adapteva. “They could become real design wins or maybe people will write open source parallel code for our platform and that will be a big win as well,” he said.
“I wish I would have done this a long time ago,” Olofsson added. “The potential of getting that many developers to sign up to a new platform is impressive--even Analog Devices and TI don’t get that,” said Olofsson, who left a job at Analog Devices to start Adapteva in 2008.
“We had been fighting the market a year-and-a-half, and it was a frustrating trying to sell without a developer community because people kept asking us to show them apps and how this platform will work,” he added.
Kickstarter is a good tool to fund production, but there’s still a need for venture or angel capital to supply seed money for a chip startup, Olofsson said.
“There’s a place for Kickstarter, having your customers fund your tape out,” he said. “We were always an open platform and appealed to a developer community so it works well for us,” he said.
However, “an initial shuttle holds too much risk, it’s not fair to backers because there’s a lot of risk in chip design,” he said. “For Kickstarter you have to have a prototype, and we had them so it was pretty low risk,” he added.