Prediction: when the A10 comes out and the world sees the kind of performance that this chip will deliver, I think there will be a lot of denial among the Intel-can-do-no-wrong crowd.
We'll see. But for graphics, sure.
Which device will A10 actually be? IPhone 7? So for Apple that would be what, a Tock, with the Tick at A11 10nm in 7s?
Yep, Intel did a good job of spinning this story to the public and many, including myself, were duped into thinking that Intel's manufacturing lead would persist/grow. You are right, the gap is narrowing and there's a real chance that TSMC will take the lead within a couple of years.
Yep. I feel really sorry for investors. But that's what's happens in technology, it's always more difficult than anyone thinks.
The paradox is that Intel was actually quite open about 14nm's performance. The thing that made it ugly is that they never met their own yield progression targets, and so their forecast was way too optimistic. The big hit was when they saw that they wouldn't be able to ramp their 10nm yields, that 14nm wasn't an isolated case. But the surprising thing is that probably no one took notice that things would get really ugly for Intel because their beating heart was slowing down.
If you compare what they showed in 2013 IM and 2014 IM, that was the time to leave the ship*.
Where where the days when Russ Fischer was writing all the nice article about Intel and Moore's Law. Have seen little of that. Where where the days when he was forecasting Intel to grow to 100$ per share, or was it $100B revenue?
E.g.
http://seekingalpha.com/article/1990411-intel-upside-potential
* Here's what they showed in sequence:
1. November 2013 (Investor Meeting: IM) Q4'13 --
forecast to match yields in Q1'14
2. September 2014 IDF: Q3'14 -- forecast to match in Q1'15
3. IM '14: Q4'14 -- telling = lying that yields are "healthy" (okay, I'll give them that because as we'll see next, they WERE healthy, but then plateau'ed).
4. IM '15: Q4'15 --
forecast to to match yields in Q3'16.
Notice that from Q1'14 to Q4'14 yields were behind, but they where (sure, all the disclaimers about graph being very vague) porogressing quite well, so they had a lot of confidence. It was in fact 2015 that was the dead blow, which it indeed was because '14 was a pretty good year for Intel. If you take their rate of progress from Q2'14 to beginning of Q4'14, it was looking very good (as you can see from 2), but then it went just BAD to the extreme. I think they already knew this at the investor meeting '14. If yields had followed the curve from 2, and 10nm had done the same, they would have been far better off.
I'm really curious about how 10nm yields will be with Cannonlake. Even if they don't show graphs, it will be very clear from all the cost graphs of products (and other statements) they show at the IM (I remember Stacy giving a nice one in November for 2016, and then giving their forecast for Q4'16).
---
I'm quite an optimistic person I figured, so I would've done a bad job investing, but the pessimistic people were right. I'm not seeing upside potential for Intel if they don't get their transistors up to snuff and they prove in 2017 they still have an advantage (--> great 10nm yields + power + performance + density AND delays from Samsung and TSMC of their "10"s and "7"s). So sell at around $35 and go in at around $30 if their manufacturing proves to be good and it's certain they will grow in revenues and stuff. But take with grain of salt..