Open Silicon seeks to be profitable by 2007

Open Silicon, a California-based "foundry-less" chip design startup with a strong team in Bangalore, aims to be cash positive "by late 2006 or early 2007", the firm's key backer says.

This confidence is justified, says Pramod Haque, a highly respected venture capitalist in California, because of the way Open Silicon chooses its chips.

Companies such as Open Silicon will also exploit the growing "eco-system" in India for semiconductor chip design, says Haque, managing partner at Palo Alto-based Norwest Venture Partners that is backing the startup.

Satya Gupta, vice president, engineering and one of the founders of Open Silicon says, "Conventionally, chip design firms try to design chips to specifications that their marketing people bring in from clients. We don't even accept a contract if a prospective clients' desired 'specs' don't fall within the range that we can handle."

Examples of Open Silicon's criteria include things like how close two lines of circuits on the chip are (130 nm or more) the frequency of signals the chips can work with (200 MHz to 500 MHz), and the number of "logic gates" that can be built into them (10-12 million).

When the specs do fit, Open Silicon offers clients ASIC (application specific integrated circuits) that almost always work the first time round, Gupta says. Equally important, says Haque, is the offer that most of the times, the firm will deliver these designs on time, critical in the go-to-market-yesterday semiconductor industry.

All this comes at a premium, but Haque won't say how much. "Our cost structures are lower than rivals' because of our business model," which includes exploiting Indian talent and strictly sticking to what the firm can do.

EE Times, an online technology news portal, however, reported in November last year, the premium could be up to "30 per cent on what it would cost a design team to take the finished design to foundry." The news portal quoted Naveed Sherwani, Open Silicon's chief executive officer.

Sherwani and his co-founders, including Gupta, were a team of ASIC engineers at Intel, when the computer chip maker decided to shut down that division. So, with help from Sequoia Capital, another venture firm, "we started off as a separate company in 2002", Gupta said. The startup has raised some $20 million so far, in two rounds of funding, including Norwest's investments.

"ASICs are ripe for changes," says Haque, "because some things aren't optimal." The cost of doing ASIC design has gone up, particularly those with lines running as close as 90 nanometre apart (a nanometre is a billionth of a metre). Just the thing that Open Silicon has cleverly avoided so far.

While the firm is now considering an entry at this newer level, sticking to the well established 130 nm or less complex designs meant fewer trouble shooting hours spent at the fabrication level, which a contract foundry in Taiwan, "or mainland China" will handle. And of course, greater confidence that the chip will work fine the first time round.

Open Silicon has 25 customer wins so far and it has finished 14 projects, Gupta Says. Of them 10 have "reached the testing and validation stage after which they will be manufactured in volumes."

Customers include makers of display units, wireless equipment, network processors and servers, he said. The company has a staff of 85, of which 35 are in the US and 50 in India. This year, the company will add another 15 engineers.