“Both Taiwan Semiconductor Manufacturing Company’s and Samsung’s 10nm processes have reached lower-than-expected yield rates, according to industry sources,” Josephine Lien and Jessie Shen report for DigiTimes. “TSMC has reportedly obtained 10nm chip orders from Apple, HiSilicon and MediaTek, with volume production set to kick off in the first quarter of 2017. Nevertheless, yield rates for TSMC’s 10nm process technology are not what the foundry expected, the sources said.”
“TSMC is scheduled to start making Apple’s A10X chips for the next-generation iPad series slated for launch in March 2017. Unsatisfactory yields for the foundry’s 10nm process could disrupt the schedule, the sources indicated,” Lien and Shen report. “TSMC reportedly will also manufacture Apple’s A11 chips for use in the iPhone 8 with volume production scheduled to kick off in the second quarter of 2017, the sources said.”
Read more in the full article here.
MacDailyNews Take: Hopefully, yields will improve dramatically and rapidly as the last thing Apple needs is yet another poorly-stocked product-free launch.
The trend is…..announce, launch with limited supply, and then meet demand 6 months later, meet the new Apple……
Supply chain expert Tim
Is TSMC yield another term for insufficient RAM? I can assure you that they launch “Pro” products with insufficient RAM.
It’s worse on the Mac, where Apple goes 4 years without a single update of any kind.
IOS consumers won’t notice a difference, but creators can’t accept this.
Now now. I think you’re referring to the incredible lack of update for the latest (last?) Mac Pro. But it wasn’t ‘4 years’. The latest (last?) Mac Pro release as in the fall of 2013. That’s a little over 3 years ago. But yeah, that’s bad.
https://en.wikipedia.org/wiki/Mac_Pro
Unless you are expecting a new Mac Pro in the next months, the math works for me. 2017-2013=4
Not really, the 4 year mark from announcement would be June, but from shipping this coming December. Most likely there will be a new Mac Pro in Q1 2017 (or Q2 as is apple’s calendar) so a little over 3 years from the shipping date of the product. Which is bad. But on the other hand if something is overbuilt to begin with and is only beginning to show it’s age now, that’s not bad. Honestly all they really need is more powerful GPU’s in there and that new 22 core E-5 Xeon.
How’s this for a config:
22 Core E-5
64 GB ddr4 2133 MHz
PCI-e ssd (3 GB/s)
Dual MI 25 Instinct (Vega Based) GPU’s
6-8 Thunderbolt 3 ports
Audio combo jack
Ethernet + 802.11 AC 3×3
I don’t think anyone would complain about that level of horsepower even if it was in the same form factor. The thing everyone seems to be forgetting is that apple doesn’t update things unless they have to for competitive reasons, or unless they can make the product better. They don’t just change things for change sake, that would be dumb. Look at the MacBook Pro (previously PowerBook), same basic design for approx 4 years, then refined and redesigned when it was possible (wall street 1997/1998, ti book 2001, aluminum 2004) And In 2008 it moved to the unibody, in 2012 the retina machines did away with the optical drives, FireWire 800, and Ethernet ports, but the basic design was essentially the same. Then 4 years later, we got the new one. With the removal of all legacy ports except the headphone jack. So in the last 19 years there have been 5 designs. The Mac Pro/Power Mac about the same. The G4’s had 3 chassis, the G5/Mac Pro shared the cheese grater from 2003-2012, and now the trash can has been 3 years… so historically everything is right on track. And since the processors haven’t been advancing as much the last 4 years, it makes no sense to just throw new things in with potential yield issues which would cause even more problems.
Paul: We’re still in 2016 at the moment. Post this again next week.
If they ever do finally get around to announcing a Mac Pro upgrade, does anybody think they will have the product available on announcement day or at least on launch day within a couple of weeks later? I doubt it…
Color me NOT surprised. Intel has had fab runs for several months and test runs for over a year. The very low yield is why Kaby Lake happened. Expect Cannon Lake next fall at 10 nm.
It’s easy to announce that you’re going to the next step. It’s harder to actually pull it off.
And, it’s an issue of how much of the chips actually conform to 10 nm. Intel has the highest percentage followed by TMSC followed by Samsung bringing up the rear with the least percentage of components at the smallest feature size. (All chips are a mixture of feature sizes with only a fraction of the chip features actually at the stated feature size. In reality, these companies should ALL state, “with features as small as xx nm”.)
We’ve all heard the stories that TMSC and Samsung have leap frogged Intel to 10 nm. It’s just not so. Intel is having problems with yields, but so are the rest.
I’ve read that they play nomenclature games with the size of the process (probably what you’re talking about). Intel’s 14 nm is close to a true 14 nm while TSMC’s 16 nm is more like 20 nm. There are many variables and it’s kind of challenging to pin it down to one size.
good grief! Is there somebody in this world that could lead apple. is there one person that sensing the change in attitude of apple buyers solves the cost problem of apples hardware. the tech was been around for years, old in the tooth, it should be much cheaper by now.
Yields are always lower than desired heading towards volume production. Every percent is critical to a chip foundry’s profits, and it is no surprise that TSMC and Samsung are encountering difficulties. The key question is whether or not these companies can sufficiently improve their processes over the next two or three months.
It is easy to toss red flags into the air. People do that at work all of the time. They want to take credit when a few of their dire predictions come true. But they never seem to own up to the many times that their uninformed and overly pessimistic viewpoints are wrong.