All these foundries seem to be chasing the same leading-edge business - which will inevitably shrink as the ROI of scaling diminishes. This is not accounting for Intel making a big play in the Foundry space with its huge capacity and world-class fabrication capability. Someone will be left owning a lot of expensive and under-utilized equipment.
Sure in the past years, i well know this, as Apple devices were a news for customers but now it's totally different. Apple is the very first to ship a 20nm SOC, not even Qualcomm will be able to do this with 810; Apple nearly is working in the risk production (low yields) because their devices are slowly losing appeal, they need of A8 "now" not sometime n H1 next year. The same patterning will follow on 16nm and 10nm because Apple has realized that without a fast refresh of phones and tablets the USA market will shift to good enough devices mediatek/rockchip powered.
And yes, many of the news on latests SOCs are only crappy gadgets, no innovation. Do you call innovation an LTE modem capable of 300Mbps on a carrier infrastructure able to give only 15/20Mbps if all goes well......it's a gadget for idiots, like fingerprint.....in a phone?? LOL. We do not speak of the displays and Gpus with a resolution that the human eye is unable to appreciate on little diagonal.
Looks like you have no idea of what you are talking about. Apple, typically, uses one (or two) nodes removed from the latest, bleeding edge process nodes that the foundaries supply. That's not because they don't want to; like you rightly mention, it takes time to get to reasonable yields on the latest technology nodes and no one would like not to make more money!
And what useless gadgets? There are good uses with every technology innovation and leaps. And that kindles innovation too.
We are in strange times. Apple, Qualcomm and others want a new shrink every 12/18 months, mainly to have "new" produts in shops to attract the customer to buy useless gadgets. Foudries are under pressure because this fast shrink is "NOT POSSIBLE" expecially at finer nodes like 14nm or 10nm.
This is the reason Apple is silent about supplier, she goes to the faster one on a given one, changing every year if possible.
I think than this is a crazy manner to do business. Instead of innovate (a tick tock ala Intel for example), they want to shrink fast adding new features to shitty ARM ISA.
My best wishes to Foundries, still i dont think this approach will be usefull to archieve strong transistor performance and dramatic power reductions, These things need time and the two years cycle is the minimum, three years i think could be even better at these terrific nodes without the help of good tools from ASML.
A Book For All Reasons Bernard Cole1 Comment Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...