By Debra Vogler
WaferNews Technical Editor
Familiar themes – real technology, real products, and real revenue – were echoed as often as possible by numerous presenters at the Semiconductor Venture Fair, held recently in San Francisco, CA. If there are any entrepreneurs in Silicon Valley who don’t yet have a clue as to what is expected of them, Dado Banatao, managing partner at Tallwood Venture Capital, put everyone on notice that investing is a business and should be run as such.
Having a product defined in a high-growth or large-market segment isn’t enough, attendees of the fair, an Infocast-sponsored event, were told.
According to Banatao, the technology itself is the ultimate differentiator, and having a team that can truly execute on that technology is vital. So is being able to adjust when something doesn’t turn out as planned. Apparently, many engineers and scientists don’t easily make such adjustments to so-called “trigger points.” One thing Tallwood has learned is that the founder and/or entrepreneur must truly be interested in the company – a lot of start-ups that had to hire founders or entrepreneurs failed.
Asked to name what’s hot now, discussion panelists (representing a cross-section of venture capital funds, analysts, and companies) mentioned wireless connectivity, storage networking, and enabling technologies that help speed the design and verification processes. Erik Org, associate at TVM Techno Venture Management, specifically noted how the current verification tools aren’t adequate.
The extent of the design challenges facing the industry starting with the 0.13-micron node have also given Lucio Lanza, managing partner at Lanza Tech Ventures, pause. “There will be a lot of chips at 0.13-micron that won’t work,” said Lanza. “Copper and low-k dielectrics create new yield sensitivities and reliability issues. We are still learning the interactions between design layout and process.”
In particular, Lanza is concerned that the traditional learning while ramping up to volume won’t work. “Try and re-try is not a survivable option. That kind of learning curve is dead.”
With respect to other kinds of opportunities for investment, Lanza prefers to think of them in categories, such as removing bottlenecks and finding new opportunities as a result of the disaggregation/redefinition of the value chain.
Barry Eggers, general partner at Lightspeed Venture Partners, doesn’t see specific “hot” areas, but thinks the new market created by InfiniBand technology is a robust segment. InfiniBand is a server I/O technology specification – supported by Intel, among others – that addresses the needs of Internet data centers such as server-to-server connections, links to remote storage and other networking devices, as well as inter-processor communication. An Intel white paper described the architecture as one that “separates the I/O subsystem from the CPU/memory complex by replacing the shared bus with a high-speed serial switched fabric.”
Similarly, Mike O’Neill, partner in Kodiak Venture Partners, believes capturing a system level problem and addressing it with silicon is key. For example, system level architecture – how data is managed and stored – is an opportunity, along with connectivity, wireless, and reducing power consumption. O’Neill emphasized that the new situation facing the industry is not based on a single product – the PC – but rather can be characterized as how people and enterprises want to use computing.
The idea that the PC may no longer be the center of the semiconductor market still appears to be a controversial viewpoint, but David Tuckerman, venture partner at CMEA Ventures, made a case for it. “I have all the computing power I need at home,” said Tuckerman. “But the enterprise has other needs.” Tuckerman thinks the conventional PC architecture won’t be right to meet those needs.
If part of building great companies with great financial returns are the goals of the renewed VC sector as expressed in Banatao’s keynote, it will be interesting to see if longevity is a part of the equation. A refreshing outlook on reliability – often so ill-planned and poorly executed in the early stages of a product’s life-cycle in a start-up – was discussed by AuthenTec’s Scott Moody, co-founder, president, and CEO, in a follow-up interview. This spin-off of Harris Semiconductor is a fabless company that designs biometric systems for fingerprint identification and uses TSMC and Chartered Semiconductor as its foundries.
While most of the panelists believed security will be a major market opportunity, AuthenTec wants biometrics to be in everything – from providing secure access to cell phones, PDAs, PCs, laptops, etc., to building access and other corporate uses. But to do so successfully, a biometric product must be reliable.
“If it doesn’t work and you can’t access your computer or PDA, you’ll be upset – even if the technology itself is low-cost,” explained Moody.
He said the company is committed to providing a product that lasts a long time in the field by focusing on the usual sorts of environmental tests, ESD tests, drop tests, etc., and having a full-time reliability tester specializing in biometrics on staff.
Returning to a non-bubble economy has reset the level of expectations for high-tech entrepreneurs, but if short-term concerns related to raising funds can be balanced with long-term issues like creating a useful, reliable product with a high-growth and/or high-volume market, it appears the new start-up philosophy is headed in the right direction.
WaferNews