The final session of the SEMI Industry Strategy Symposium (The CxO Panel) was the most interesting for me because executives from three of the four most influential semiconductor companies were on the panel: Dr. Goeff Yeap of Qualcomm, Dr. Jack Sun from TSMC, and Mr. Mark Bohr of Intel. Who is fourth you ask? That would be Apple of course. In my opinion the most interesting question asked of the panel was: Which foundry will be first to FinFET?
The CxO panel closes the conference with a conversation between Dan Hutcheson and several esteemed executives at a CxO level. This year, we will be examining what it takes to practically transition a new technology to manufacturing and through its yield ramp. There’s no magic in it for these execs: they have to make finFETs yield today while rolling 10nm and 7nm processes down the gangwayfrom pathfinding into development. It is a task that has consumed many at prior nodes. They have closed their doors on Moore’s Law. Our panelists are the champions who won at these nodes and will keep moving down the scaling path. They will do this with a combination of their own internal excellence and an unparalleled ability to partner with others to align a global base of researchers and engineers — all towards hitting a specific ramp date that is exactly two years after the last one. They’re going to talk about how it’s done against a background in which every node gets bigger, faster, and riskier. They’ll provide their vision about the future, the challenges, and the opportunities.
Before I give you my ranking of which foundries will ship 16/14nm FinFETs first let me say a few words about Mark Bohr. I have always said the key to success in the semiconductor industry is showing up. Meeting customers and partners, writing papers or blogs, and attending conferences. Some call it collaborating but I call it showing up. Speaking at conferences is important too but if you only speak and do not actually attend that does not count. And don’t think we do not know who arrives right before their time slot and leaves right after because we most certainly do.
Mark Bohr not only spoke at ISS he showed up. I sat in the row next to him and chatted with him during breaks. He was not on his phone, he paid attention and took notes. He also graciously accepted a copy of “Fabless: The Transformation of the Semiconductor Industry” and I sincerely believe he will read it.
Which Foundry will be First to FinFET according to me:
Mark’s answer to the question by the way was “not Intel Foundry”. Another interesting comment from Mark was that Moore’s law will continue another 10 years but he says that every 10 years. Mark also explained to me the difference between the 14nm processes used for the Intel CPUs and SoCs. It really is the same process, same basic transistor structure, materials, design rules, and minimum interconnect pitches at the lower layers. The SoC version however has a wider range of devices (high-voltage and ultra-low leakage transistors) and a few more process steps. The bottom line is the processes are similar enough to run in the same “copy exact” fabs and that is quite an achievement, absolutely.