TSMC already has a huge capital expenditure. It's something like 8 billion last year, and could be more than 9 billion this year. This is already huge compared to its revenue (17 billion last year), so although anything from Apple might help, it's probably not going to be substantial compared to what TSMC's already spending.
Next-Gen iPhone & iPhone Nano Speculation
- Thread starter Arun
- Start date
Allegedly there was a $1+b exclusivity offer on the table from both Apple and Qualcomm, but it was declined by TSMC..
http://www.phonearena.com/news/TSMC...s-for-exclusive-chip-foundry-capacity_id33877
I wonder how much either could afford to offer if really determined to secure this, and at what point would it be something TSMC would no longer refuse? Something like $2b in pure profit could make a big difference even vs an already spent $9b. I wonder how much it would cost to fund an entirely new fab from them..
IIRC TSMC's latest fab (Fab 15) alone cost something like NT$300 billion (~ US$9 billion) to build. Such fab is expected to generate ~ US$5 billion revenue per year.
ltcommander.data
Regular
https://semiaccurate.com/2012/12/06/nvidias-maxwell-process-choice/
Well if you want to believe Charlie, he's somehow concluded that Maxwell still being released on 28nm in 2014 when TSMC's 20nm process should be up and running to indicate that Apple has bought up TSMC's entire initial 20nm production run. Certainly, Apple willing to pay big, pay long-term, and pay up-front is not new for their supplier relations and they wouldn't even need to explicitly require exclusivity if TSMC wants to remain an impartial, open contractor, but simply bid up the wafer price so it's not worthwhile for anyone else to jump in early. But with no love lost by Charlie for nVidia and no actual proof it's definitely a stretch.
ltcommander.data
Regular
http://www.anandtech.com/show/6834/apple-tv-2013-a1469-short-review-analysis-of-a-new-a5
The new true single core A5 cuts power consumption in half compared to the previous disabled A5. Disabled cores aren't normally a major power user right? So along with the large area reduction, it definitely suggests there being more changes than just removing already disabled parts. The design effort put into it really makes it seem like it should see more use than just the Apple TV.
The new true single core A5 cuts power consumption in half compared to the previous disabled A5. Disabled cores aren't normally a major power user right? So along with the large area reduction, it definitely suggests there being more changes than just removing already disabled parts. The design effort put into it really makes it seem like it should see more use than just the Apple TV.
anexanhume
Veteran
I believe they lase out certain features so I'm guessing the current path is actually destroyed.
I wonder how much on the A5 intended for mobile device I/O (radios, USB/lightning) etc. they removed to lower power. Must have dialed the clocks in too.
Why should it have to be used in more than just the Apple TV? Didn't Apple sell around 5 million Apple TVs in 2012? If so, doesn't that volume alone justify a separate chip design just to save some costs?
Now, if they could drive down the price of the Apple TV or maybe add some local storage for the same price, there might be higher demand.
Of course, they're supposedly still working on a TV so this SOC could prove useful in such a product.
I'm not sure what Roku uses but I think DVRs use off-the-shelf media processors from Broadcom and the like. A5 should be more capable than those chips but maybe not enough to run apps, a capability that some suspect Apple TV will gain at some point.
Of course, they're supposedly still working on a TV so this SOC could prove useful in such a product.
I'm not sure what Roku uses but I think DVRs use off-the-shelf media processors from Broadcom and the like. A5 should be more capable than those chips but maybe not enough to run apps, a capability that some suspect Apple TV will gain at some point.
If there is any Fabs Manufacturer that do things on time, very little delay and with high yield rate that would be Intel. Unfortunately Intel has yet to decide on ( Or they are not ready yet ) being a Fab Partner. Especially with its future CEO undecided.
So the safe bet would be to have two manufacture for its SoC. Which is costly for most but Apple. Since they dont sell any chips and doesn't need to make a profits from it. So having both Samsung and TSMC makes a lot of sense. Especially the decision from coming Tim Cook, who has always prefer to hedge his bet.
The thing i still dont understand is why Apple hasn't make its own Baseband SoC yet. It is currently costing $20 ( or $30 for the LTE version ) per one which is the 2nd most expansive item on its BOM.
Anyone has any thought on this?
P.S - I really really want a 5" iPhone. My eyesight is not getting better by the days and it would properly be a few more years before i am forced to switch to larger screen Smartphones.
So the safe bet would be to have two manufacture for its SoC. Which is costly for most but Apple. Since they dont sell any chips and doesn't need to make a profits from it. So having both Samsung and TSMC makes a lot of sense. Especially the decision from coming Tim Cook, who has always prefer to hedge his bet.
The thing i still dont understand is why Apple hasn't make its own Baseband SoC yet. It is currently costing $20 ( or $30 for the LTE version ) per one which is the 2nd most expansive item on its BOM.
Anyone has any thought on this?
P.S - I really really want a 5" iPhone. My eyesight is not getting better by the days and it would properly be a few more years before i am forced to switch to larger screen Smartphones.
This is completely vendor and phone size agnostic: when you have a problem with a 4" display with a sub <720p resolution things aren't going to get any better with a 5" display at a =/<1080p resolution and no resolution is NOT going to remain idle when display estate grows. For those cases it would be best to consult an eye doctor and whatever he subscribes.
The basebands are all standardized and the tech remains the same for decades.
There is no competitive advantage to be gained by designing something that anyone can buy cheaply.
Also, even if you design your own baseband, you are very likely having to pay almost as much for royalties. So, unless there's a significant cost advantage (such as integration into the SoC), there's no clear benefit for doing it.
Yes, you will have to pay some royalties.
No, they won't be nearly as much. Not even close.
The telecom standards committees never adopt standards that have unreasonable license costs. Google 'FRAND'.
Perhaps a better choice of words is "because of" royalties. Otherwise, integration into the SoC won't save much
I was referring to specially about the 5" iPhone, which provides the *same* resolution as iPhone 5, just everything, Text and Icons would look bigger.
Who and why guarantees that if Apple goes for a larger screen estate that the resolution will remain the same? There's no 5" iPhone available from what I can see unless I'm missing something.
anexanhume
Veteran
That was a ratio shift too though. The iphone resolution doesn't need to increase any further though. I think they're pretty set in their current resolutions aside from a retina iPad mini.