I have been asked by one of you forum members for my opinion about David Kanter's prediction regarding Intel's plan for 10 nm or even 7 nm.

Okay, so why not answer in public! Did I sell my POET shares? No! Did I buy more? Well, not today, but recently.

Without going too much into technical details, there are some obvious similarities between Intel's image from 2009 and POET's work, namely the use of GaAs – though not as a substrate, but as a "nucleation and buffer layer", as Intel puts it. Other common terms we recognize are "strained InGaAs" or "quantum well". There is, however, a remarkable difference in that Kanter assumes InGaAs for the n-channel and SiGe for the p-channel, while POET uses InGaAs for both.

Aside: If you want to dive deeper into the technical details (or at least into some of them), you should read "GaAs: The logical successor to CMOS" and "Moore's Law Has No End in Sight". Make sure you also read the comment section of the latter as it has a comment from someone at POET Technologies, cited by the article's author R. Colin Johnson. End aside

Generally speaking, competition never sleeps, will it? If Intel is really going to deploy III/V materials in their upcoming technologies, this would not only be competition in some areas, but it would also be a nice confirmation of POET's approach.

However, please keep in mind that Kanter's prediction is not necessarily Intel's roadmap, but rather just, well, a prediction.

We'll have to wait and see …

• whether this is indeed the way Intel's will go forward,
• how fast they will be able to proceed, especially if they are targeting 10 or 7 nm,
• whether and how they will have to circumvent Dr. Taylor's patents,
• whether they'll have to license POET's IP,
• whether they will go for 10 or 7 nm with their extreme costs.

The most prominent difference, however, is that Kanter's prediction is all about transistors. No thyristors, no optics, no integration of optics and electronics, no analog, no mixed-signal. Much less than what POET is offering!

Now let's assume for a moment that everything plays out fine for Intel and that they'll have a fast GaAs chip technology in 10 or 7 nm. Would be great if they'd have to use POET's IP! But what if not? Bye, bye, POET's monopoly in the integrated electronics area!

On the plus side Intel's competitors would be under pressure to come up with an alternative. And yes, there'd be something we could recommend …

A lot would still be left over, see above. I believe we cannot overestimate the relevance of the optoelectronical integration. For example, have a look at the transcript of the CIC video and have Stephane Gagnon explain that to you, starting at 12:21 and especially from 14:04!

Oh, and then there's the question of different target node sizes, but this post is already long enough, so I'll leave that for someone else who might want to jump in here.