I attended a lunch yesterday with GlobalFoundries CEO Sanjay Jha (formerly of Qualcomm), SVP Gregg Barltlett (Motorola/GF), and CTO Gary Patton (IBM). Having followed GF from the very beginning I can tell you without a shadow of a doubt that GF has transformed from a collection of companies (AMD, Chartered, IBM) to a fully integrated pure-play foundry.
There were a dozen other media people in attendance who will cover the bullet points presented so let me comment on the leadership aspect. Having met Sanjay, Gregg, and Gary multiple times separately, I found it very interesting to see them interact as a team.
You have to remember these gentlemen come from very different backgrounds and experiences but one thing I can tell you is that they truly worked in harmony during this event which is not always the case with other companies. In my opinion the credit goes to Sanjay as he is an exceptional leader and knows the semiconductor industry from the very top (end product) down to the very bottom (device level), absolutely.
In fact, I would put Sanjay in my current top 10 Semiconductor CEOs list with Hock Tan, Jen-Hsun Huang, MK Tsai, Morris Chang, Amit Gupta, Joe Costello, Lip-Bu Tan, Wally Rhines, and Aart de Geus.
In my opinion, under Sanjay’s leadership GF will become an industry leader versus follower. The FDX family of processes is a clear example. GF will be the only pure-play foundry to deliver leading edge processes using both FinFET and FD-SOI technology which I think is astounding. So, while the other foundries are trying to fit every design into a FinFET, GF gives us a choice between high performance and lower power and lower cost.
Okay, back to the presentation, GF is rolling out their 22FDX (FD-SOI) process and announced a 12FDX process. 22FDX is competitive with 14nm and 12FDX is targeted at 10nm FinFET processes. As we all know, FD-SOI enables lower cost and lower power chips and GF will be the only pure-play foundry with sub 28nm FD-SOI process technologies.
In regards to FinFETS, the good news is that GlobalFoundries and AMD are scheduled to start 7nm production in 2018 which means AMD could have 7nm chips a full year before Intel. The bad news is that TSMC will start 7nm production in 2017 which means other fabless companies (ARM based server chips for example) will have 7nm chips 6 months before AMD. TSMC and Intel also have the advantage of a 10nm process yield learning experience that GF and AMD do not since they skipped 10nm, so there is an additional risk of delay.
It will be interesting to see how TSMC and Samsung respond to this. The TSMC OIP Ecosystem Forum is next week so we will find out then but today I saw a new and improved GlobalFoundries (GF 2.0) and am quite impressed with their progress.