Hi Andy,

I remember I posted a couple of years ago following a phone conversation I had with Pierhal that IBM was especially surprised/upset by POET because they had already invested 10s of millions of $$ in silicon photonics. I'm not sure if that is what you were thinking of, but I never meant to imply that wouldn't touch POET.

I recall one of our posters on SH saying they spoke to someone close to the R&D for Intel and they said they would acquire a company if it moved them ahead 6 months R&D-wise. I can't remember who posted it, and obviously money has to be a factor at some point. But it's interesting to think about. I think the people with the R&D money and those actually doing the R&D don't always have their motivations aligned. It could come to pass that an Intel or IBM R&D department could ask for more money, saying their 'almost there', and the folks working the budget declining in favour of picking up a new technology instead. These same R&D guys could end up working *for* our guys in a year's time :)

As for the merits of POET itself, fairchij's posts do a great job showing the metrics. GaAs is a direct band gap material while Si has an indirect band gap. That means that energy fed into a GaAs material can be converted into photons while Si can't emit light - it generates heat instead. This means there is no Si laser and Si will then automatically require at least one other material (and likely another packaging step) to be the light emitter on the chip.

Having said that, what I've learned fairly recently is that POET's advantage isn't as much that GaAs is so unlike Si. It's more that POET is so much like existing processes and on-chip devices. Electronically, everything on POET has an Si analog. It's compatible with the current state of the art.

The added bonus is the photonics capability which solves a decades old problem of how to get the data traveling inside the box at light speed. There are other photonics chips out there, but they are packages of smaller chips or they are purely optical and have no electronics capability and therefore can't be used as memory, processor, etc.

Anyway, I'm rambling. But check out this article that I have posted literally dozens of times.

Linking with Light

It still is the best article for explaining the problem that POET solves. At this point it's almost prophetic in the way it describes the timelines for advances in optoelectonics. I thought POET was going to beat the predictions, but we hit a couple of bumps :S