Two Buck CMOS
by Paul McGoldrick
You can't live on the West Coast of the US and not know about Two Buck Chuck.
If you don't it's the slang name given to the Chardonnay that is sold in
the Trader Joe's chain for two dollars a bottle in the California stores
(and three dollars outside the Golden State.) It's cheap because there is
an enormous glut of Chardonnay with fewer and fewer drinkers as the wine-imbibing
among us have supposedly become more "sophisticated" in our tastes.
A few wineries, a very few, have the marketing talents to still attract
the buyer who will spend fifteen dollars on a bottle. And that made me think
of CMOS…
One of our favorite wineries is owned and operated by a German named Dieter. He has made Chardonnay from the start of his operations but last year he gave up trying to sell all the hard work he had to put into his Chardonnay just to get a few lousy bucks back. So he pulled all those grapes and planted instead the now fashionable Pinot Noir. He also expanded his growing area by planting on a 50% gradient hillside that he must have to work with mountain climbing gear.
It's the same in the semiconductor industry with CMOS. More and more vendors are seeing lousy returns on their CMOS product lines and are pulling out by handing all the manufacturing to the volume fabs in Asia. They're keeping their Bipolar lines running because they can see the margins in a lot of those products and, of course, the really fashionable compound processes are what attracts attention -- and investors.
But this is still the equivalent of throwing away what could be -- and is for less than a handful of semiconductor vendors -- a much more profitable arena. CMOS product is cheap to make (how expensive can sand get, for heavens sake) and if you compare the two dollar bottle of Chardonnay with the two dollar IC it is perhaps self-evident where the profits are. With the Chardonnay you wait maybe five years for your grapes to be ready to use, you manually collect them, then you process them for maybe two years before the wine is ready to be bottled, labeled and then transported in expensive cardboard packaging to be warehoused and then distributed to retail outlets. (That's one reason that wineries like to sell direct; even offering the product at half the store price they make much more profit selling at the tasting event.) Then look at the two dollar IC: The design could be nine months for a really innovative product then we create many thousands of them on a single wafer and then slice them apart, test them and package them in a really cheap container and tape-and-reel them ready to be shipped to the volume user. And most of the process is totally machine-controlled.
Yes, the equipment needed to produce that two dollar CMOS IC is incredibly expensive compared to the few millions that need to be invested in a winery operation, but the profits are massive. And in many cases you can make to order rather than guessing what the product is going to taste like in two years.
So why do so many vendors outsource their CMOS? Because so many of the products that they have designed cannot be moved into profitable niches.
It's back to the fifteen dollar bottle of Chardonnay. How do they manage to persuade the buyer to put that kind of money down for it instead of going to Trader Joe's? It's called marketing. They identify the demographic of the buyer who will spend that amount -- even if it is just snobbery -- and then offer the kind of product that demographic will buy. It should be the same in the CMOS market. Identify the buyers who will spend that amount and offer them the kind of product they want to buy.
Marketing in our industry should be two-fold: From the field, accept the information from niches where money will be spent, listen and filter the information, target the products that are wanted, design them, make them and sell back to niches while also advertising to get those even more profitable smaller users. It sounds relatively straightforward doesn't it? But in many cases what happens is that a company designs a product for one customer -- and that's the only customer they get -- or they are unsure of the intelligence that has been gathered and take what they think is the safe route of knocking off someone else's product: The "If we make it and sell it for less we can take a market share" syndrome. It can work, but it's rarely profitable.
Two Buck Chuck doesn't make money. Two Buck CMOS does.