VCs puzzled over future investments

Xilinx Inc. President and CEO Moshe Gavrielov said it was all over: Venture capital would not return to the semiconductor industry—not even after the current recession—because of the heady sums required to launch a chip start-up and the more than five-year lag before VC investors see a return.

If he's right, it signals dark days ahead for what has always been one of the most dynamic technology-based industries.

Xilinx shareholders, of course, might hope Gavrielov's prediction holds true, because then semiconductor start-ups would increasingly become intellectual property plays, with the IP burned into FPGAs.

At first glance, Gavrielov's logic seems solid. Return-on-investment times have stretched over the past decade, while semiconductor market growth has diminished. And plenty of IP companies are already working with Xilinx. For example, start-up Omiino Ltd. is working on telecom network circuits that will go to market on Xilinx FPGAs. But then there's also Tpack A/S, a vendor of Altera FPGAs programmed to implement communications functions.

Indeed, Gavrielov's argument implies opportunities for rival FPGA start-ups. The industry has seen the emergence of Achronix Semiconductor Corp., targeting the high-performance market, and SiliconBlue Technologies Corp., focused on extremely low-power FPGAs.

So venture capital investment is more complex than Gavrielov may have suggested. Nonetheless, with the global economy mired in recession, few would dispute that these are tough times for many VCs and their portfolio companies.

"It is a case of 'back to basics' in venture capital. A lot of VCs are increasing their interest in cleantech, but the overall trend is towards capital efficiency," said Sven Lingjaerde, cofounder and managing director of VC firm Endeavour Vision SA (Geneva). Lingjaerde is also president and founder of the European Tech Tour Association, a not-for-profit group that helps put VCs in touch with investment opportunities.

Looking for the next big thing
"It is absolutely clear we are looking at how much [money] a company needs," Lingjaerde added. "And VCs are always looking for the next big thing, a really disruptive approach to the next market. But that revolution should come at a cost that is bearable."

That does not necessarily mean a wholesale shift away from semiconductors and towards some other activity, such as the Internet. Appealing as that prospect might be, because of the lower barriers to entry, the flip side of a low barrier to entry is a crowded market.

Capital efficiency comes in various forms, but one of the most obvious is simply to focus on an area other than the leading edge of digital design. It can cost perhaps as much as Rs.497.19 crore ($100 million) to get a leading-edge digital chip out the door, in what becomes a bet-the-company bid to hit a market just as it's taking off. And quick growth just got put on hold by the recession.

But there are still plenty of hardware-oriented opportunities that do not require huge outlays, in areas such as analogue, RF and microelectromechanical system (MEMS) devices. EDA might also be assumed to be an attractive opportunity, since it is software-based and provides a low barrier to entry, but that assumption is not borne out in practice.