OVER THE LAST FOUR YEARS THERE HAVE BEEN some significant changes in the semiconductor ecosystem that will have a major impact on semiconductor companies, particularly startups. These changes will influence every aspect of company strategy and execution and span diverse. In this article, we examine these macro issues. We also believe that there are still some really exciting pockets of activity in a range of areas that make for some attractive investing over the near future that we will highlight towards the end of the article.

Changes in semiconductor consumption pattern: Consumption of chips in Asia/Pacific is approximately 40 percent of global consumption today (vs. 20 percent five years ago). Combining Japan, which is approximately 20 percent of global consumption, that is three times the total consumption of semiconductor devices in the Americas! In addition, the type of products consumed is changing from the jellybean, low end products to increasingly cutting edge communications and computing products. For example, many Asian economies are the leading adopters of next generation communications technologies, as witnessed by the growth of ADSL/VDSL ports and next generation handsets in Asia. Millions of VDSL ports will be under deployment in Korea and Japan much before it will enter the deployment roadmaps of North American RBOCs. Multi-media messaging, gaming, chat and SMS all show significantly greater adoption in the Far East compared to Americas and Europe. This early adoption and significant market potential is beginning to drive two things—customized products and increasingly local technological content and innovation. China’s recent embrace of a new DVD standard—Enhanced Video Disk (EVD), primarily for local consumption, in order to avoid paying DVD royalties illustrates the power of this growing consumption. China has taken a similar position in the wireless services markets with TD-SCDMA and encryption standards for wireless LAN products. This runs counter to historic practices where products were designed for U.S. OEMs and U.S. consumers, and over time were adopted by Asian markets. These developments are both good news and bad news for semiconductor entrepreneurs. The good news is that worldwide, markets are growing faster with decreasing dependence on the economic climate of any one region. The downside is that capital constrained startups need to figure out ways to get closer to an increasingly diverse customer set and be able to rapidly adapt to needs of different markets. The savvy entrepreneur would clearly think ahead through morphing geographical consumption patterns and standards before placing their bets on any particular product/market opportunity.

Democratization of semiconductor talent: Software and semiconductor manufacturing industries have benefited on the expense side from global outsourcing of resources. For the semiconductor chip industry, this is still a trend in relative infancy. By some estimates there are over 250 fabless companies in Taiwan alone. This trend is not far behind in India where TI, Intel, Motorola, Analog Devices and National Semiconductor have all set up large design centers. In addition to the cost efficiencies, with the shift in end markets to Asia these companies will also benefit immensely in being close to a rapidly growing customer base with unique requirements. Further, practically every capability in the product development ecosystem from front-end design to test and validation to applications engineering is developing at a rapid pace in these markets. That clearly highlights the increasing degree of savviness in understanding and building products for end customer demand. This freeing up of talent supply across Asia adds to the skill base that every startup can potentially tap into. The downside to this trend relative to control, IP protection are well understood. However, the biggest downside to semiconductor startups here is increased competition. Earlier, engineers in Austin mostly worried about what their brethren in Silicon Valley or Boston were up to on the competitive front. But now, with the rapid democratization of the talent and startup eco-system they also need to worry about “a couple of smart guys in a garage” in Bangalore.

Capital markets – trials and tribulations: The venture capital industry is undergoing huge shrinkage over the past few years as evidenced by the dramatic drop-off from the over $100B raised in 2000 by VC firms to less than $10B in 2003. Reducing fund sizes implies both smaller number of venture investments and lesser dollars available for an investment. In addition, after the surreal exuberance of 1998-2000, venture capitalists have reset expectations for median exits for semiconductor startups. Likely median exits in the $150 to $200m range imply that the total dollars invested before cash flow breakeven should be less than $30 to $35m to get to required venture returns. That in turn suggests that startup semiconductor companies need to get to tangible technology and customer validation for less than $15m. This does impose some severe constraints on the types of chip development projects that startups can embark on. Increasing design tool and mask costs with decreasing geometries are adding additional strain to the initial capital intensity of semiconductor startups. However it is not all bad news for semiconductor entrepreneurs. As a percentage of total venture dollars invested, semiconductors have gone from 2 percent in 1998 to almost 10 percent of total venture investments for 2002. Semi investing, which was once considered a niche is clearly more of a mainstream category for most venture firms.

These disruptive trends and the steady increase in global semiconductor content continue to provide a fertile ground for entrepreneurs to harvest. However, these trends do point to the need for a potentially different opportunity filter and changes in execution strategy. Here we would like to talk about a few areas that hold some good promise.

Addressing second order issues for mainstream OEMs: The semiconductor industry has accomplished a lot in staying on the treadmill of faster speeds and feeds at lower prices. Unfortunately, increasing clock rates and reducing geometries are generating a set of second order issues that could retard the progress. We believe two of the biggest threats to compute industry OEM roadmaps are thermal and power management. The combination of power hungry processors (dual processors >100W each), high current transients (>10A/ns) and denser form factors (blades/1u/2u) creates non-trivial tradeoffs for board designers. The thermal challenges revolve both around keeping the junction temperature on a chip below prescribed limits as well as transporting the heat away from the surface to the edges of the system where they can be cooled using fans and other ambient cooling systems. Nanocoolers—one of our portfolio companies—is engaged in addressing this problem. Similarly, with respect to power management, the challenges revolve around delivering power supplies that are small, having a high power density and efficiency and the ability to respond to sudden current transients. Powerstone and Prophesi Technologies are examples of two companies in our portfolio that are addressing related issues. We expect this area to be one of fertile activity in the years to come.

Market specific signal processing solutions: Another trend that we think has significant potential is one, which provides market specific solutions by integrating the requisite analog functions, core processing and peripherals on a single chip. Two companies illustrative of this are D2Audio and Big Bear Networks. D2Audio provides a sub-system solution that allows for the transition from analog to all-digital power amplifiers for high power audio-video receivers. Similarly Big Bear Networks is developing solutions that use advanced high-speed signal processing techniques to address distance and legacy fiber support issues that are serious inhibitors of the transition to 10G Ethernet in the enterprise backbone. Both these companies bring a unique combination of deep vertical market knowledge and advanced signal processing capabilities taken to market in a highly integrated, easy to use solution that breaks the barriers to key industry trends - transition from analog to digital amplifiers for D2Audio and to 10G Ethernet on legacy fiber for Big Bear.

Horizontal mixed signal devices: The design of analog parts is considered a bit of an art form. The core skill is to understand fundamental design behavior and to take advantage of parasitics and higher order effects of different components in a circuit. Design tools that automate this process are practically non-existent. In contrast, digital design operates in the world of 1s and 0s and is a fairly programmatic process that involves the use of standard EDA tools and design flows. These methodologies are fairly orthogonal. However, opportunity arises (in addition to squabbles over which one is tougher - A or D?) when these two skill sets are integrated within a company and ultimately rendered on a single chip. After all, the world lives in the analog domain while most of the compute processing takes place in the digital domain. Their integration on a single chip leads to numerous typical benefits—lower cost, smaller form factors, greater programmability and so on. Austin is blessed with plentiful talent in this area and as a result is a core area of strength and activity for us at Austin Ventures. Crystal Semiconductor (CRUS), Silicon Labs (SLAB), Benchmarq (TI), Cygnal (SLAB), D2Audio are all examples of companies in town that have taken good advantage of this talent base.

Venu Shamapant joined Austin Ventures in 1999 and focuses on the firm's Hardware investment area. Previously, Shamapant served as VP of business development for Telecom Technologies, Inc. Prior, he was with McKinsey & Co. and with Mentor Graphics. Shamapant received his MBA from Harvard, an M.S. in computer engineering from the University of Texas at Austin, and a BS in electronics and communications engineering from Osmania University, India.

Krishna Srinivasan joined Austin Ventures in February 2000 and focuses on the firm's Hardware investment area. Previously, Srinivasan was with McKinsey & Co. Prior to that, he was with Motorola Corporate Offices and also an Engineer at International Sematech. Srinivasan received his MBA from the Wharton, where he graduated as a Palmer Scholar. He also has a Masters in operations research from the University of Texas at Austin and an undergraduate degree from the Indian Institute of Technology.