Thats what I was thinking as well PoetPower. Also the cost savings from using the GaAs Wafers should give POET quite the competive advantage.

One thing that threw me a bit this morning was the post from ahasja, regarding Moores Law.

"The heat is on

GaAs has a number of downsides, of course. GaAs semiconductors are still expensive to produce. And while it can handle higher power inputs than silicon, GaAs also handles heat poorly in comparison to silicon. The chips also leak a significant amount of the power pushed into circuits as heat. GaAs chips have about a third of the thermal conductivity of silicon, and heat tends to build up in some areas rather than distribute evenly across the surface of the chip. That can seriously reduce the expected life of GaAs RF components, since they typically operate at temperatures above 120 degrees Celsius (up to as much as 150 degrees Celsius in military radar systems)."

and further on

"You can't just wait around for 30 years to see if components last that long before putting them in the field. So to accelerate testing of failure times for semiconductors, engineers use high heats as a sort of time machine—using the "Arrhenius-type behavior" of the semiconductor to extrapolate how long it would last during normal operation. Using a formula developed by Swedish scientist Svante Arrhenius (the man who first tried to calculate the greenhouse effect on global climate), it's possible to get a fairly accurate prediction of how long a semiconductor circuit will last. "If you heat them up to 300 to 400 degrees Celsius, you can force them to degrade more quickly," Whelan explained. This degradation happens in a predictable enough way to get a statistical picture of how long the chips will last.

It turns out that it's very hard to get GaAs chips through the Arrhenius test with the kind of performance the military is looking for. Depending on the design, researchers at the Naval Research Laboratory found that GaAs MMICs under continuous use will experience gaps between failures from as little as a year up to 22 years when operated at 120 degrees Celsius (failure being defined as a 20 percent drop in performance)."

I always thought GaAs handled the heat better than Silicon, or is it just that less heat is generated?