Venture capitalists and entrepreneurs alike have been bemoaning the fact that nobody starts semiconductor companies anymore. It seems that it’s just too costly and too risky for entrepreneurs to challenge old world chip companies who have locked up the market.
But the folks who started Samplify Systems didn’t listen. The Santa Clara, Calif. company is coming out of stealth today with a plan to create a smarter generation of chips that convert analog data to digital data. In other words, they convert things like sound into the ones and zeroes of digital computerese. These chips are necessary for everything from cat-scans to baby ultrasound pictures.
Normally, analog chips just convert data and pass it along. But Samplify’s chips can convert data and then compress it too. This may result in cheaper and more accurate products for consumers.
It may not sound like much. But the move requires expertise in “mixed signal” technology, or the combination of analog and digital functions on the same chip. It’s a trick that is novel enough to get the startup off the ground in the face of much larger competition. And the company has been able to design its first chip with just 16 employees and $11 million in funding. (Some chip companies have raised $100 million).
The analog chip market is the domain of big chip companies such as Texas Instuments, National Semiconductor, Maxim Integrated Products, Linear Technology, and Analog Devices. But Samplify figured out how to make the chips better in a way that takes a lot of cost out of a system.
Typically, analog chips are built in old, depreciated chip factories that use old chip-making processes (such as 0.35 micron or 0.25 micron features, a measurement of the width between circuits). Those older plants are fine for analog circuits and they’re really cheap. But it doesn’t pay to put digital circuitry on the same chip, because the process for the digital circuitry is really inefficient.
Consequently, many data converter chips have a lot of analog and just a little bit of digital circuitry on them. Or they simply do the analog processing and then pass the data on to a second chip (dubbed a field programmable gate array) that handles the digital compression. The problem is that moving data in uncompressed form is far more inefficient than moving it around in compressed digital form. The transmission side and the receiving side need sophisticated circuitry that can handle a lot of analog data movement. That’s costly.
Tom Sparkman, chief executive of Samplify, said the company has figured out a way to put a lot of digital circuitry together in the same chip as analog circuitry.
A system that once needed the ability to send 200 gigabits of data throughout the system each second might now get by with just 60 gigabits a second. An example is a wireless antenna in a train station or an airport. It often requires a fiber optic cable to move data in analog form. That expensive infrastructure could be eliminated if the data were moved in digital form.
By doing this, Sparkman said the Santa Clata, Calif.-based company can dramatically reduce the costs of electronic gadgetry. The company is focused on markets where this kind of solution matters a lot, such as chips for ultrasound machines or cat-scan machines. The chips can also be used in cell phone infrastructure, military equipment, test equipment, and data acquisition equipment.
It is, in essence, a better mousetrap, dreamed up by Al Wegener, chief technology officer of Samplify. The road has been a long one. Wegener started working on the technology in 1999. His bosses at Texas Instruments didn’t want the technology, so he got permission to take it to a startup. He received initial funding from Charles River Ventures in the fall of 2006. The company raised a second round in March 2007. Besides Charles River Ventures, Formative Ventures and angels also invested.
At first, the company sought to license its technology to others. It has been showing the technology to potential customers or chip makers in the past year. But there is more money in making chips. So the company is now designing its own chips, which will be fabricated by a contract chip manufacturer, Taiwan’s United Microelectronics. Sparkman said the company was able to get its first sample chip done on a budget of $5.6 million.
He expects customers will start deploying systems with the chips in the fall of 2009. Data converter chips, which Gartner predicts will be a $3.9 billion market in 2011, typically sell for $30 to $50. Samplify will sell its chips for $39.50 to $79. Over time, Sparkman said the company might be able to integrate its chips into other higher value chips.
Brian Connors, a partner at Formative Ventures who invested in Samplify, says he can count on one hand the number of semiconductor companies that came to him seeking venture money since that time. He said that Samplify was unique in the tricks it used to get a chip done quickly and cheaply. It also had strong patents and pending patents.
“I’m afraid this trend is going in the wrong direction for the industry,” said Connors. “It’s one of the reasons we love the Samplify model. Someone has to prove that the semiconductor industry can be capital efficient or its bad news for electronics in general.”
For many years, only big companies ruled the analog space. One of the reasons is that analog engineers are rare commodities. They have specialized talents that aren’t taught in universities today. Hence, the big companies coddled them, giving them plenty of stock options and bonuses. But the analog stocks peaked a few years ago, and so the rewards from employers aren’t want they used to be. Analog chip engineers are thus more motivated to join startups. The problem was that there were very few who found a hole in the strategies of big players.
Sparkman himself spent decades at Maxim. The stock is trading around $15 a share now. But a few years ago when he sold some of his shares, Maxim’s stock was above $87.50 a share. Now Maxim’s stock price just doesn’t have the same power to retain employees. Sparkman doesn’t think he’ll have any trouble hiring when the company starts recruiting again sometime next year.
Sparkman said Samplify plans on raising a new round of funding in the first half of next year. But whatever he raises, Sparkman says the company will still serve as an example that it doesn’t take a gigantic war chest to start a chip company.