Monolithic: So basically, POET does know what they are aiming to do in order to get the 4th pin in the thyristor to work in the Gallium Arsenide platform (BICFET), but they need funds and this goal of a monolithic integration is still slightly up in the air. This 4th pin seems to be the key to commerciallize the POET platform in the GAAs platform as a monolithic chip …

Well, almost. They tried several different ways to get the fourth terminal (connector) to the thyristor, but didn't succeed. And although they had a working DOES thyristor in the UConn lab, they didn't develop the 4-terminal variant at UConn since this was not needed in a lab setting. POET is confident that they can solve that problem eventually, but each attempt is very time-consuming and very expensive – and only afterwards you know whether you succeeded – or have to make another iteration with another approach.

So, as a workaround, they replaced the DOES thyristor by a bipolar transistor (BICFET). A transistor has three terminals only, which they know how to manufacture. That solution does work. However, power optimization is required, and this does not only pertain to the BICFET itself, but causes some feedback on the other components, so work is required there, too. In summary, the BICFET approach is faster than the DOES approach, but it would still require considerable time and money.

As Suresh explained, the market forecasts for GaAs and InP were more or less on par in 2015. However, since then the forcasts changed in favor of InP, and now it is the InP market that is exploding. The decision to put a focus on that market in my opinion is the right strategic direction, because with InP

• we can make more money (higher ROI),
• we can make money faster,
• we have the manufacturing capabilities,
• we have engineering samples of innovative products in the hands of customers as we speak, and more samples will be available in the near future (to avoid the blacklisted S-word).

The GaAs platform is not dead, as some people suggested here, but will be further developed. The focus, however, is clearly on the InP platform now – for good reasons.