Timing is Everything… the concept of the 3D-IC has been around for decades, so why now?!
We all hear “Timing is Everything”, and it is so very true for a startup – too early could be as bad as too late. Being an entrepreneur, I am probably far too optimistic to present an objective view and especially with respect to the Monolithic 3D-IC. So you could be the judge and hopefully reality will soon provide the final judgment.
Let me tell you why I believe the time for monolithic 3D is now!
Before presenting the argument for “why now”, we may want to consider why the 3D-IC has not happened until now. My best explanation is that it was not really needed. As they say, “If it ain’t broke, don’t fix it”. Scaling worked great over the past few decades – 2X increased density, coupled with higher speed and reduced power, were worth dying for. The system industry needed to work very hard to develop and bring to market products that utilized this rapid improvement of device cost performance and complexity. Semiconductor vendors had their hands full with all the challenges associated with scaling. In short, we were too busy staying on course to explore “futuristic” ideas.
So what has changed?
It is now broken!
Scaling now is associated with escalating costs of process development and capital investment as illustrated by the following charts.
This escalating cost has brought the scaling cost beyond most vendors’ financial abilities. In fact, from the good old days where we had tens of semiconductor manufacturing vendors, the days where semiconductor executives said “real men have fabs”, it is expected that only 4 vendors will scale into the 22/20 node, as indicated by the following chart courtesy of iSuppli.
It is clear that scaling is broken – at least for those vendors who are not part of the “lucky 4”
And to make matters worse, when these escalating costs are calculated for their impact on the end device cost, we now face, for the first time in semiconductor history, an increase of device cost with scaling for many applications and vendors. See the following chart (Courtesy IBS).
The writing on the wall is loud and clear – scaling down is broken. So, let’s fix it if we can – “yes we can”
Scaling up using monolithic 3D would allow us to continue Moore’s Law, something the high tech industry and its users are so addicted to. This would double the number of end device transistors at similar or reduced prices and with additional performance improvements.If monolithic 3D can do it, then the time of
monolithic 3D is NOW!