While there are a lot of nice niches in widebandgap electronics, there won't be much left if everyone (and there are more) splits up the market evenly. Let's be frank: our most optimistic scenario estimates that the market may grow to as much as $300 million by 2012, while competing in a sandbox (an available market) of about $1 billion. That $300 million is a great niche if you and a few others share it, but it isn't much if it's divided among 15 or 20 companies. And keep in mind, that's not even our "most likely" scenario.

But first, let's review what widebandgap electronics can do. They can run faster, hotter, and at higher voltage and power than other semiconductors. Other technologies can do any one of these more cost effectively, but GaN and SiC are very good at doing them all at once. That means they are good for high-speed circuits and power circuits, or better yet, high-speed power circuits. It also means improved efficiency, especially when you consider the overall system design. For example, widebandgap devices can save in cooling, battery power, and allow more compact and portable systems. So, we can expect to see them in mobile platforms, hybrid cars, that sort of thing.

For GaN, the hot products are HEMTs; for microwave power amplifiers. Silicon LDMOS will continue at the lower frequencies, GaAs at lower powers, and vacuum tubes at the very highest powers and frequencies (and for microwave ovens!). But GaN is great for new applications like WiMAX base stations, especially since these new technologies not only use higher carrier frequencies (where silicon is at a disadvantage), but also wider signal bandwidths. To send that much information over a wide area, you need an efficient and linear amplifier. GaN (and to some extent SiC) does that well.

For SiC, the hot products are Schottky diodes for power supplies and other power circuits. It's been more difficult to commercialize FETs and bipolar transistors, but there is now some progress there too. With increasing regulatory pressures to make electronics and motors more efficient, and the soaring price of energy, the timing is perfect for SiC (and to some extent GaN) power devices.

Opportunities are not without challenges, though, and there are plenty of those too. As always, a lot depends on unit prices, but prices depend greatly on having good manufacturing yields on 4-inch and even perhaps 6-inch wafers, whether the wafer material is SiC, Si, sapphire, or native GaN. In the case of GaN substrates and epi layers, the defect density is still a challenge. Great progress has been made, but there is more to go.

There is much more to electronics than just the chip. Packages must be improved to match the materials. Circuits must be rejiggered to match the new devices. And while microwave circuit design is complex and time-consuming, the price pressure in power supply design is punishing. There is also uncertainty in the end-user demand. Wireless technologies are notoriously prone to risk. WiMAX is a case in point. When will it finally take off?

There are several WiMAX technologies, and there are other wireless technologies vying for the same application. Even in the slower-moving power components business, silicon is still moving forward after all these years. For example, a start-up called Q-Speed aims to do what SiC can do, but in silicon. That is, its "soft recovery" diodes that are faster than conventional ultrafast diodes, and almost as fast as SiC Schottky diodes. And so far its diodes can go to 600 V reverse voltage, which extends beyond the range of silicon Schottky diodes (but not as high as SiC).

Of course, there are also the competitors. We group them into four categories. For the GaAs foundries like Eudyna, Oki, RFMD, Toshiba, and Triquint, working on GaN is a no-brainer. They already understand microwave markets, and they are paying for their III-V foundries with their GaAs products. For military contractors like HRL, Northrop Grumman, and Raytheon, working on GaN is also an obvious choice, since they need the new technology for their high-margin customers. Cree, Nitronex, and Velox Semiconductor are using their advantage in materials expertise. Moreover, Cree can pay for much of its electronics manufacturing with its LED foundry. Finally, power electronics powerhouses like Infineon, STMicrolectronics, and others can leverage their ties to demanding customers (such as the automotive industry) to make and sell SiC diodes and transistors.

All said, the growth rates are very attractive and the timing is finally right. And very importantly, the challenges facing widebandgap electronics suppliers are not only the usual kind, but they are the kind you want: the "good" kind.