We heard that one high profile AMD veteran transferred to GlobalFoundries and we got a crash course on why this is a good thing. John Docherty is now Senior Vice President, Global Operations, at GlobalFoundries and since he had a similar role at AMD, he should do well in this new position.
John is an ex-ATI chap, he worked as a Senior Vice President Operations at ATI and in AMD he was doing the same thing - he was Senior Vice President, Manufacturing Operations. To put things in perspective, Mr. Docherty has hands on experience in both GPU manufacturing and CPU manufacturing, which is not the case with most executives.
In recent years he was helping the manufacture of APUs (Accelerated Processing Units) which can be seen as a subtype of System on a Chip (SoC). He also worked as VP of European Die Manufacturing for Motorola Semiconductor Freescale, VP FE manufacturing (Singapore) for LSI Agere Systems before going after ATI and AMD. That's quite a CV, like we told you - this chap really knows manufacturing.
GlobalFoundries rely on AMD as their primary customer and John knows AMD's needs, this looks like a match made in heaven, at least for AMD and GlobalFoundries/Samsung alliance.
This might result in AMD coming to market on time with 14nm FinFET products, which we expect in the course of 2016. GlobalFoundries definitely need to tune things up in manufacturing, as 20nm is delayed for both GPUs and APUs to at least 2015, while Apple is the only customer who has a SoCs in this process, at least for now. Apple is making its 20nm A8 SoCs in TSMC and Samsung, but not GlobalFoundries.
FinFET in 14nm looks like a real step forward, enabling amazing bandwidth and products especially in computer graphics, a technology that can take all the bandwidth you can provide. Nvidia will make its 16nm FinFET based GPU codenamed Pascal at TSMC, but we are sure that AMD is cooking up something that can compete with Pascal.
Let's hope that 14/16nm FinFET will bring us products much sooner than originally expected, as the 20nm node is more or less broken and will be passed over by many companies, as chip manufacturers are getting all that is possible from 28nm. You can just imagine how power efficient a 14nm GPU can be in theory, considering that Nvidia Maxwell 2.0 has high-end performance at mid-range TDPs, and Maxwell is still a 28nm chip.
