Apple's next-gen A7 processor reportedly being readied for production by TSMC
A report from the Far East cites sources who expect Taiwan Semiconductor Manufacturing Company to "tape out" Apple's A7 SoC this month ahead of risk production in early summer, suggesting that future iOS devices may not rely solely on silicon manufactured by Samsung.
According to DigiTimes, industry sources say that TSMC is ready to "tape out" the new A7 chip on its 20nm process, with the processors scheduled to be implanted into Apple products in the first quarter of 2014. While the publication has a questionable track record, Thursday's report echoes recent accounts that said TSMC has started production of certain Apple chips as part of a long-rumored partnership.
As noted by CNET, TSMC is expected to reach the "tape out" stage ? which references the photomasking process that is one of the final steps in circuit design ? for the A7 in March and move into early production in either May or June. The firm will reportedly expand operations at its "14-fab" located in Tainan Science Park, with investments in production facilities said to reach nearly $17 billion.
TSMC set a capital expenditure budget of $9 billion for 2013, an increase of $700 million from 2012. The report claims that about 90 percent of this year's capex will be put toward bolstering output of chips based on company's 28nm process. The Taiwanese firm said in January that it forecasts a near 100 percent ownership of the 28nm chip market during 2013, a statement that sparked speculation of massive orders from Apple.
According to DigiTimes, industry sources say that TSMC is ready to "tape out" the new A7 chip on its 20nm process, with the processors scheduled to be implanted into Apple products in the first quarter of 2014. While the publication has a questionable track record, Thursday's report echoes recent accounts that said TSMC has started production of certain Apple chips as part of a long-rumored partnership.
As noted by CNET, TSMC is expected to reach the "tape out" stage ? which references the photomasking process that is one of the final steps in circuit design ? for the A7 in March and move into early production in either May or June. The firm will reportedly expand operations at its "14-fab" located in Tainan Science Park, with investments in production facilities said to reach nearly $17 billion.
TSMC set a capital expenditure budget of $9 billion for 2013, an increase of $700 million from 2012. The report claims that about 90 percent of this year's capex will be put toward bolstering output of chips based on company's 28nm process. The Taiwanese firm said in January that it forecasts a near 100 percent ownership of the 28nm chip market during 2013, a statement that sparked speculation of massive orders from Apple.
Comments
2. TSMC getting '100% ownership of the 28nm chip market during 2013' is that actually a good thing, competition-wise?
3. I wonder if their subsidiary WaferTech (outside of Portland) is able to produce chips for Apple, putting more US build parts into iPhones et cetera.
That doesn't make sense. Tape-out is done by the chip designer.
Even with risk wafers they would never get SoCs into production before October.
(To many uneducated news reporters that talk about Intel, TSMC, Samsung and so on... Its not just to move production. A new tapeout is required for each manufacture redesign of the SoC to meet the specific manufactures rules. Apple need 50-70K wafer starts per month. People have to understand how much that is. TSMC total output is 90K. )
1. Schiller may never do another interview again in his life. Seriously, though, I wonder what that WSJ thing was. It was not an interview, it was some kind of ambush. We may hear something about this.
2. If they seriously have such a process jump on everbody else, how is it not a good thing for Apple to be using their whole output exclusively?
3. "Project Azalea." So mysterious. Is this a huge strategy shift for Apple toward Taiwan and the U.S., and away from you-know-who?
4. They gotta do something about that logo.
TSMC is supposedly far along with their next node so a pilot production effort is a real possibility. The real question though would Apple be able to adapt that quickly to TSMCs production technology. That iIs debatable, there are many factors involved here. This is all true to an extent but like everything else automation is part of the production chain. Beyond that if Apple has the processors detailed in a portable HDL it ought to make the move easier. Beyond that Apple would have started such a move more than a year ago.
Apple also has enough cash to build as many plants as TSMC could possible need. If the chip is shrunk enough that could cut wafer starts in half. Frankly I don't expect a big shrink, rather I'd expect Apple to leverage that chip area for more functionality.
The A7 is now to be 20nm in length.
Photo masking, which involves painting masks on photos will be a $17 billion industry by 2014
Bloody damn good I think.
Can I be an analyst now?
Thank you! At last i understand...
Quote:
Originally Posted by shompa
From tapeout to production its at least 6 month! For example X86 at least 12 month.
Even with risk wafers they would never get SoCs into production before October.
(To many uneducated news reporters that talk about Intel, TSMC, Samsung and so on... Its not just to move production. A new tapeout is required for each manufacture redesign of the SoC to meet the specific manufactures rules. Apple need 50-70K wafer starts per month. People have to understand how much that is. TSMC total output is 90K. )
Hence:
Quote:
Taiwan Semiconductor Manufacturing Company (TSMC) is expected to tape out Apple's A7 processor on a 20nm process in March and then move the chip into risk production in May-June, which will pave the way for commercial shipments in the first quarter of 2014, according to industry sources.
This matches up with GF and TSMC being wafer ready for 20nm and FinFET at 14XM
A well written FAQ: http://www.globalfoundries.com/technology/14XM-FAQ.aspx
Hah. Actually you can learn more real stuff from the average streetwalker.
I have to say, though, that the AI reporters do a better job with Apple-centric subjects than the average NYT, WSJ or NPR reporter. This comes mostly from knowing the subject better.
The first course in journalism schools ought to be Humility Before the Truth 101. Instead the schools and the job itself tacitly teach arrogance born of ignorance.
About that TSMC logo: does it suggest precision or even competence? I think it suggests an engineer knocked it out in an afternoon.
Is it a silicon wafer with 11 out of 216 complete chips being bad? Nothing about it is visually appealing.
Precisely, that is what makes it so good! It highlights that the company is a no non sense engineering company.
The reality is this, I've seen companies spend huge bucks to contract with firms to produce logos and other marks for a company to replace the old and then to have such work thrown away a couple of years later. Often the materials generated just don't ring with customers or have no linkage with the products the company markets.
Then we have TSMC logo which is simple to implement and easy to recognize. Even if you are only loosely familiar with semiconductor manufacturing it immediately gives hint to what TSMC is all about. Further it is a logo that can be printed anywhere even on wafers or individual chips. Its simplicity highlights a company that is practical and focused.
Really what do you guys expect from a Logo?
When you think about this a bit and then think about things like Apples rumored iWatch, you can't help to think that such a product would really mature in 2015. These power reductions we are seeing are huge compared to 28/32nm tech. Displays are already huge power draws for portable devices, this shrunken tech just highlights the display power problem even more. It really makes you wonder what is being done to address that issue at Apple.
Whatever happens with Apples custom chips I still have this dream that my next iPad will be a sixty four bit device with four cores and at the very least 2GB of RAM. That would be a killer iPad platform.
Why does it need to be 64 bit? There's little or no advantage for a device like this - even if it does have 2 GB of RAM. When you add the extra overhead, it might even be slower than a 32 bit processor. Better to use all those extra transistors for something useful.