Designing and making integrated circuits, especially single-die monolithic devices, takes a special skill set, one which differs from developing and making higher-level modules which incorporate one or more die plus discrete components in a non-IC type of package. You'd think that vendors of conventional ICs would stay away from non-IC products.
You'd be wrong. In the past year, several leading vendors of power-supply ICs, such as National Semiconductor and Linear Technology Corp., have stepped up the vertical-integration ladder and introduced tiny power modules which are more like complete power subsystems than power-supply building blocks. On one hand, they are taking a big chance, on the other hand, it's an opportunity to increase their market share, sell products with higher added value, and move to proprietary, unique products with a higher average selling price (ASP). Sure, there is a risk to their well-tuned design process, manufacturing margins, support structure, and even corporate culture, but they obviously have decided the risk is worth the possible return, both financially and strategically.
This move to higher integration is not new. IC vendors have been doing it for decades (yes, it's true: ICs have been around long enough that we can now speak of them in terms of decades.) in those older days, some of this integration was via "hybrids" which placed multiple die and passives onto a common (often ceramic) substrate; other times, it was via brick-like epoxy-potted modules. These products spanned power supplies to RF subsystems; they were not limited to a single application niche.
IN many cases, these non-monolithic IC products brought in significant revenue and even profits (the two are not the same, we know) and made a lot of users happy. They replaced a portion of the circuit and system design effort, along with the uncertainties of debug, manufacturing, and test. At the same time, they put the IC vendors in the awkward position of now competing with some of their existing customers. They also significantly changed many of the application and support issues that a vendor must factor into their business model. After all, as a vendor, you are now taking full responsibility for ore of the total design, and you can't blame the customer if your module (aka his subsystem) doesn't work to spec.
But, hey, that's life in the competitive world. There will be winners and losers, and it will be interesting to see how this latest turn of the wheel works out for the vendors and users. So check back in a year or two. ♦