The analog semiconductor industry is subdivided into essentially four segments: power (54 percent), amplifiers (17 percent), data converters (17 percent), and interface (13 percent).
Of all these segments, the data converter segment includes the most expensive and most profitable analog products sold. Those facts would seem to portend an area ripe for price competition, but lately I've been wondering if price matters when specifying precision analog devices such as data converters.
If one listens to the leadership inside Texas Instruments, Analog Devices, and other high-performance manufacturers of precision analog, they claim their products are sold on performance, not price. But something an old semiconductor sales colleague told me, years ago, suggests this logic is flawed.
According to my friend, if a company can lower the sales price of a product by 50 percent while maintaining the same quality, the sales volume would increase ten-fold. While I have to admit that this logic is a bit hard to believe, it does seem to apply in power management analog.
Earlier this year I wrote about The Race From Power Management to Data Converters . In the blog I showed that Dialog Semiconductor went from less than $100 million in sales to $750 million in only five years selling power management devices at about 60 percent of the price of comparable analog manufacturers. During that same period of time, the other high performance analog companies had sales that were flat to down.
So what about precision analog, an even higher margin analog business? Is it next in line on the price chopping block?
For now it doesn't appear so. I wonder why.
It certainly can't be that the market is too small. Analog Devices alone, the leader in precision amplifiers and data converters, has annual sales of over $2.7 billion. Linear Technology, another precision analog manufacturer sells annually about $1.3 billion of semiconductors. Both of these companies sell their products for three to four times the manufacturing cost so there would appear to be room for a Dialog-type competitor.
As an example of the gold-valued nature of the precision analog business, consider the recent introduction of a 20-bit successive approximation register (SAR) converter by Linear Technology. Whereas the current generation of 18-bit SAR converters is sold for roughly $25 — a handsome price for a 16-pin piece of plastic and silicon — this new 20-bit industrial grade SAR costs over $40 when purchased on a reel of 2,500 parts. Think for a moment about what that means.
A single reel of that 16-pin SAR converter costs over $100,000. If taken off the reel, that $100,000 of product would fit inside the space of a single, slightly melted, ice cube.
Actually, what may be more surprising to a lot of people is that these types of products are built with perhaps 30 cents worth of materials. Consequently, the price must be based upon the cost to test and trim the precision — and the profit of course — lots of profit.
So if a competitor decided to go in like Dialog did for power and sell an equivalent 20-bit converter for $20, would the volume increase 10-fold? Would the customers buying 18-bit converters migrate to 20-bits to increase precision for less money? And if they did, would their products benefit from the higher precision?
I recall once reading a quote by Jim Williams, formerly of Linear Technology, stating that very few things need 18 or more bits of precision. Now this was written a long time ago, and humans aren't necessarily accurate at predicting the future requirements of technology, so who knows?
I suppose there is one big difference between power management and precision analog devices. The former is sold primarily to consumers (i.e., smartphones, tablet computers, TVs, etc.) whereas high precision analog is sold primarily to industrial, medical and test customers. But is this situation a result of price or limited applications where price is irrelevant?
Certainly one would have been hard pressed to accept that 32-bit processors would become necessary in a toy. But now they can be purchased for less than 80 cents in volume. If one could buy a 20-bit SAR converter for 80 cents, would the volume requirements for 20-bits go way up? Would consumer applications start to appear where they were once rendered price prohibitive? Would anybody really care?
I don't know the answers to these questions, but they are interesting questions to ponder. What are your thoughts? Is there any reason to continue to innovate in precision analog for the sake of a lower price?