We don’t have a chance to report on good news very often these days, so when the Fabless Semiconductor Association (FSA) says that investments in fabless semiconductor startups for the first half of this year have surpassed the number of deals for the same period last year, it’s time to celebrate. What’s more, deal pace in the first three weeks of the third quarter has gone through the roof.
For the first six months of the year, a total of 62 fabless chip startups raised $758 million in venture funding, according to Austin, Texas-based FSA and data analysis by Venture Capital Journal. That’s 9% of the $8.4 billion raised by all VC-backed companies in the first half, as reported by PwC/ThomsonVE/NVCA Money Tree survey. Investments in fabless startups were well ahead of those in biotechnology ($639 million) and second only to investments in software ($864 million).
It’s true that the total funds raised for fabless investments is down from the same period last year, when 58 companies raised $940 million, but you’d be hard pressed to identify a sector in which investors are paying less while getting more equity for their money; valuations are down in all sectors. The important point is that more fabless startups are receiving funds this year, and the pace of deals is accelerating. In the first three weeks of the third quarter, VCs pumped $200 million into 20 fabless deals. “The industry is on track to pass the  fabless startups funded in 2002,” says Jodi Shelton, executive director of the FSA.
A closer look reveals that two areas of fabless startups are dominating VC interest. The No. 1 category is wireless companies, with 28 investments totaling more than $285 million and. The No. 2 category is imaging, with 12 investments totaling more than $200 million. The two segments pulled in two-thirds (more than 40% and 25%, respectively) of VC investments in fabless companies in the first half of the year. Smaller segments of investing include startups in storage, optoelectronics and power management.
There is little surprise in the enthusiasm for wireless, which also dominated fabless investment in the last six months of 2002. The opportunities in cell phones, cellular infrastructure and multifunction devices (formerly known as PDAs) have attracted a Who’s Who of venture firms. There are five important sub-categories of investment within wireless: network processors, wireless LANS, infrastructure, modems and dual-mode chips, as well as two smaller areas-audio processing chips (two deals) and integrated circuits for mixed-signal applications (two deals). The most active VCs that made multiple investments in the wireless sector were careful to spread their dollars around the sub-categories.
Venture interest in imaging startups is somewhat more surprising, given assertions by some analysts that the sector is already over-invested. Twelve such companies received first-half funding, the majority of which went into follow-on rounds.
Opinion is divided as to whether too much money is flowing into imaging deals. One camp of analysts believes the imaging market is headed for the kind of me-too investing that led to the meltdown of fabless semiconductor startups in 3D graphics during the mid-1990s. But another camp says that the companies that have raised venture money are fairly evenly distributed across applications markets, technology segments and opportunities.
In the past two years, funds for imaging startups have flowed into either sensor technology – traditionally dominated by charged couple device (CCD) startups – or image processing technology (so-called back-end processors). And while that separation continues, there are two important new areas of investment. One is sensors built in CMOS. Companies in that sector – most prominently Foveon – will compete with CCDs on cost of manufacturing and soon compete for resolution performance.
The other area pulling in new venture dollars is system-on-a-chip (SoC) startups, which integrate multiple functions on single semiconductors to serve mobile telephony or multifunction devices markets. Venture-backed startups in this space include Xelerated, a wireless network processor company, and Cradle Technologies, whose products support data, voice, images and video.
Not So Fab
All of the activity in fabless startups contrasts with the doldrums of the mainstream semiconductor industry. There were hopes for a recovery this year, but it stalled, says Doug Andre, principal analyst for the Semiconductor Industry Association (SIA) in San Jose, Calif. “First there was the Iraq War, then SARS, and both of those had an impact on the sale of handsets,” Andre says.
That’s bad news. The cellular telephone market is now the largest consumer of the products of the semiconductor industry, ahead of personal computers, for which, Andre says, “IDC and others are reporting the highest growth rates since 2000.” But that’s not saying much. Neither the consumer nor corporate market for personal computers provides the kind of growth it once did for semiconductor manufacturers. In any case, the personal computer market is dominated by two players, Intel and AMD, which are locked in a death struggle over performance, functionality and price. That doesn’t leave much room in the sector for venture-backed startups.
That leaves cellular telephony as the fastest growing market, dominated by captive semiconductor production from the manufacturers of handsets, base station equipment providers and the most nimble of the public fabless semiconductor companies on one side. On the other side is a fast-growing group of Asian suppliers who copy and replicate the work of the first group of companies with astonishing speed. In between those two groups are fabless semiconductor startups. It’s the place where the venture capital industry funds the innovations that keep the U.S. semiconductor industry on the leading edge of technology.
Times Have Changed
So what’s the fascination with fabless semiconductor companies? The market space between huge entrenched manufacturers and the shameless intellectual property thieves of Asia sounds like a tough place to find oneself as a startup company.
John Michaelson, president of Needham Investment Management, whose firm has made 16 investments in fabless semiconductor startups over the last eight years, explains why his firm remains enthusiastic about fabless: “Many of the barriers to entry have been lowered. Design tools have standardized. There are more design centers and the access to fabs is easier with the increase in the number of foundries that support fabless design companies.”
On top of that, the model for building fabless design companies and ramping successful ones to $100 million in annual sales is well understood. It’s one of the few technology markets in which venture-backed firms can still build a business that allows VCs to earn 5x or 6x on their original investments, says Graham O’Keefe, senior principal analyst at Atlas Venture.
It’s the kind of space where smart VCs can make money quickly. Because “two-thirds of all ICs today go into consumer products, and because manufacturers of consumer products are very cost sensitive, there is intense price pressure,” Michaelson says. That means manufacturers are always looking for ways to save money, add features and get to market fast; a succinct summary of what fabless semiconductor startups offer to their customers.
“Speed, is the key,” Michaelson adds. “These companies have to get their products designed and into the market to exploit any new technology [niche] before cheaper manufacturers get out even cheaper products.”
O’Keefe agrees with Michaelson. With VCs locked out of the mainstream semiconductor industry by the high cost of building a new fab (the average cost ranges from $2 billion to $3 billion), all interest naturally flows into the fabless sector where innovation – and profits – are possible.
There is risk, of course, that too many VCs are betting on too many companies. Analysts say that may very well be the case with startups that make imaging chips. Some cheerleaders say the sky’s the limit, citing research saying that the market for cellular camera phones is expected to hit 140 million units in 2005. But Brian Alger, a senior equity analyst at Pacific Growth Equities urges caution: “Anytime you see projections going from zero to 140 million, there is a lot of hype,” he says.
If history is an indicator, VCs are overestimating the market. A similar cycle of investment took place in the mid-1990s around 3D graphics chip startups. Nearly 50 companies were funded. Most were absorbed into other companies. Now, only a half-dozen companies remain and three of them-NVIDIA, ATI Technologies and Intel-hold 85% of the current graphics chip market.
“Lots of these startups have nifty technology solutions, but there are a ton of companies targeting the market,” says Alger of Pacific Growth. He adds that few of the startups have inked crucial OEM deals to ensure their long-term viability.
Moreover, the market is growing more competitive as publicly traded chipmakers eye the cell phone camera market – companies like Genesis Microchip, STMicroelectronics, NeoMagic, Cirrus Logic and Zoran. And many consumer electronics manufacturers already make their own imaging chips, including NEC, Nikon, Sony, Canon, Samsung and Philips.
Imaging startups are bullish about their own prospects. Saul Altabet, senior director of marketing with NuCORE, which produces back-end image processing chips, says the mainstream graphics chips and cell-phone imaging chip markets are different. Yes, it’s harder to get into the cell phone space because the technology hurdles are higher, but it’s also “harder to get designed out,” Altabet says. His point being that the volatility in the earlier graphics market is not necessarily relevant to imaging chips.
Whether any of these startups can grow up to be long-lasting public companies with sales in excess of $1 billion per year is anyone’s guess. As always the ladies and gentlemen of the VC industry are placing their bets.