| SAN JOSE, Calif. — The climate for venture capital investments in semiconductor technology is healthy but challenging, a panel of experts concluded. Before an audience of hopeful entrepreneurs, analyst Steve Szirom, president of Insidechips.com, said the average forecast for the semiconductor industry this year is 21 percent growth, revenues are solid and shortages are starting to appear for some products. Still Szirom urged caution. Executives remain cautions about capital spending, and the future impact of higher interest rates, wars in Iraq and Afghanistan along with rapid increases in energy prices is not fully understood. In this environment, venture firms are starting to fund again, reported Mike O'Neill, general partner at Kodiak Venture Partners. O'Neill described a venture enviroment very different from the pre-bubble days of the late 1990s, but perhaps healthier in the long run. "Just as you are out with your foil set pitching [venture capitalists], we are out with our foil set pitching to the limited partners who put the money in our funds," O'Neill said. "It's a painful process, just as it is for you." O'Neill said money no longer pours into funds. Instead, it has to be solicited, often amidst heavy competition. Limited partners, mostly institutions with huge investments, are increasingly selective. "Just as we evaluate business plans, limited partners evaluate VC," O'Neill said. "They look for a track record, the skills to be selective and to stay with a growing company. They are not interested in someone who has only done a few C rounds for their friends." The good news it that there is a lot of money — maybe too much money, O'Neill suggested — looking for investment funds. A major difference in the current cycle, O'Neill continued, is the absence of the corporate venture funds, which haven't returned to the market. The new environment has also made fund managers much more selective. John Stockton, partner at StockLabs (Austin, Texas) suggested that venture capitalists' preferences have changed, and the landscape has changed too. He listed wireless — now past its peak and declining — and digital photography as the big drivers in the semiconductor industry this year. But he warned that the environment for digital fabless startups is deadly, with little differentiation and little chance for a profitable exit. "We look for startups grounded in good, hard science," Stockton said. "Companies who build their business on proprietary technology in analog/mixed-signal, MEMS or new nonvolatile technologies make sense." Stockton also sees some opportunities in EDA. "The good news is that there are lots of new problems in chip design," he observed. "That means there are lots of companies sprouting up to attack them. And that represents a decent opportunity." Will Strauss, president of Forward Concepts (Phoenix) said high-growth end markets depend on DSPs, but often in embedded forms rather than DSP chips. But with little in the way of differentiation, DSP tends to be a market of narrow windows and many casualties. Thomas Rosch, general partner at Intervest Partners, added that the digital fabless model is broken. "It takes $20 million to complete a new chip design," Rosch said. "But at the same time, there is no money for obscure markets, and too much competition in the obvious markets. It is bleak." One way to fix the broken model might be less expensive routes to silicon such as those offered by Open Silicon or eSilicon, or by outsourcing. Another would be new, configurable architectures that would shift the process of entering a new market from chip design to software development. Rosch said many emerging markets were already effectively closed to new entrants, in some cases before products are shipped. This is true of the ultrawideband communications market, anything 802.11-related, telecom applications, offload-engines and storage processors. In contrast, Steve Buehler, editor-in-chief of Semiconductor Manufacturing magazine, said there were opportunities for new ventures. "Yields are coming up across the semiconductor manufacturing industry, even at 90 nm," Buehler said. "But manufacturers are focusing on metrology, and on tools and software for yield management. And they are battling problems at every stage of the process. Panelists also said that some investment funds are moving away from the U.S. Strauss said he has been asked to participate in due-diligence reviews of a number of non-U.S. investments — often cases in which non-U.S. investors are partnering with U.S.-based investors. Panelists agreed that the next overcapacity-driven slowdown would hit sometime in 2006. But Beuhler said the length of the business cycle has been shrinking for some time. "Of course, if executives are still so badly burned from the last time that they stay conservative on capital spending, that could pull it in a little bit." | 
