Prabhat Dubey is busy steering his company— TelASIC Communications —transitioning from product development stage to delivering products to customers worldwide.

A developer of analog and mixed-signal integrated circuits and sub-systems for the 3G wireless infrastructure markets, TelAsic recently closed a $26 million series D financing round, the largest deal in the semiconductor space, so far, in 2005.

Dubey might be sitting pretty happy. However, everything isn’t that rosy. Venture investment in semiconductor industry has decreased by 40 percent in first quarter (Q1) 2005 when compared to Q1 2004, according to the Venture Capital Counter by Silicon Strategies.

$397.55 million was invested into semiconductor startups in Q1 2005, up 13.4 percent from Q4 2004 but down 40.2 percent from the $664.5 million recorded in Q1 2004.
Philip Gianos, General Partner at Interwest Partners doesn’t like to read too much into the falling index.

“So what?” he questions. “The falling index for a particular quarter is not a trend to draw the conclusion that the venture activity in the space has taken back seat.”

According to PriceWaterhouseCoopers, in 2004 $1.4 billion was invested in the semiconductor space.

“The cost of building a chip company is more than ever before. The venture capitalists’ will continue to pump in money in their existing portfolios and do whatever best they can to get a good multiple on exit. So I do not see any fall off yet. At the same time there is a nagging doubt in the minds of the investors whether they can support as many deals as is being funded. Investors are struggling to make money in a capital-intensive space.”

Sumant Mandal, managing director at Clearstone Venture Partners agrees. “Increasing complexity of geometries calls for sophisticated tools and other resources. So it takes not only more money to build a semiconductor company but also takes longer time to stabilize.”

Interesting Themes
One of the fundamental aspects of silicon business is that the economics of silicon are centered primarily on cost reduction applied to existing high volume markets. The innovations do not tend to be huge disruptions. Reasonable but not extremely disruptive advantages can swing vast revenues toward a newcomer. Some domains are not for startups, given the resource requirements like computer processors. But there are plenty of other spaces in digital communications, RF and memory technologies.

Any company building innovative chipset for the wireless segment—802.11, Wi Max, Ultra Wide Band—is a hot bed for venture capitalists. A lot of money has been poured into this space. However, it isn’t clear whether everyone is making money! The UltraWideBand space excites Bill Tai, partner, Charles River Ventures.

“While we are a believer, and made our decision to invest in the area in 2002, there's a huge crowd of latecomers just coming to the party now—I suspect for the latecomers it will be like funding an Internet bookstore in 2000; coming in years after sector leadership has already been established,” he jokes.

With mobility being the buzzword and proliferating cellular devices across continents, video processing, decompression, image processing seem to be the promising areas to build innovative chipsets. Though the space has just started to get crowded, the good news is that there is still room for innovation. Then there are a handful of companies developing the next generation Ethernet chipsets—making 10 Gbps transmission a reality—a promising area to bet on.

Memory is an exciting place and offers a huge marketplace as well. “Memory is a hard place to make a dent. It scares a lot of people—both VCs and entrepreneurs. Though there have been a couple of memory deals, all those have been long difficult deals,” says Mandal.

Electronic Design Automation (EDA) is again a tough place to build a big company. Magma’s IPO was the last EDA exit on the public markets. With increasing complexity of geometries, there is enormous room to build innovative EDA tools. VCs often make interesting investments in this realm since big players dominating the EDA space acquire the startups.

Being Successful
“In Silicon, its all about execution. Designing it right and building it right. You have to deliver the best price performance, as this is a cost game in the end, assuming you offer the best performance,” says Tai.
In the semiconductor business it is difficult to build things speculatively hoping that there will be a big market. One has to understand that it will cost millions to get to the marketplace. The management has to have tight control over the scope of the project and target a specific customer group. Close relationships with customers also play an important role. “The key is to shorten the time to revenue and increase the revenue ramp,” says Gianos.

To shorten the time to revenue, Gianos says, one has to focus on innovation at the system design level. This allows the company to differentiate itself in the marketplace. Process technology is not a differentiating factor. By getting closer to system level design there is a better chance of being successful, he remarks.

Semiconductor certainly is a ‘cyclical business’ in which bottlenecks keep shifting over time. “You need to have the ability to identify those bottlenecks and build companies around it,” says Mandal.

Exit Strategy
“There are lots of ‘single silicon product cycles’ that will be acquired by companies with efficient channels of distribution, and a handful of ‘breakaway opportunities’ that can go public. The ratio will shift more and more toward acquisition versus IPO though due to the added expenses of being a public company from Sarbanes Oxley legislation and the increasing burn rate required in general to develop multiple products simultaneously,” says Tai.

But Brent Holliday of Greenstone Venture Partners questions, “Why be down on M&A?” He adds, “A typical fabless semiconductor company requires $35 million in investment to get to profitability. In the past four years, the largest M&A transaction for a semiconductor company was $122 million.

I happen to know that the investors did 3x on that deal. Big whip! A hit based business like venture capital does not get rich on 3x. And that is the best one. IPOs in the semiconductor space, like Portal Player, SiRF and Atheros did much better, all going public at around $500 million company value. That is great performance and closer to that magical 10x hit. Everyone grumbled that the best one can expect in an M&A transaction is $200-$500 million and there will be very few of those.

Companies need to be tracking on $50-$100 million in revenue and profitable to go public these days. That is a big company.”

It takes time to get to that level. M& A will be the order of the day till then.