Apple Reportedly Buys Into a Chip Fab, Looking to Produce Its Own Chips

[gpat]

While that's true.... It wasn't until a few years ago that Apple owned a chip design company now did they? No reason for a company who builds consumer devices to own a Fab unless they own a chip design company right? I mean what would they do, buy someone else's designs and then FAB someone elses design? Might as well just let them design and build them entirely. However, now that they own the chip design, it only makes sense to move to the next phase and own a FAB. True?

For the same reason keeping them fabless for the other components, it makes sense to them to remain fabless for chips. Design themselves and let someone else build them, so they can negotiate for lower prices, make contracts for the quantities they need, and make sure they have them in time.
Running a fab is an unnecessary waste of energy. They won't be able to keep up with the demand they need, while substaining unnecessary costs in areas they are strangers to.

 

[tdtran1025]

Designing a SoC is a huge undertaking. Manufacturing at processes below 20nm is even riskier. Yet for the business model Apple embraces, there might be little choice for Apple.

 

[donutbagel]

Bad troll is a bad troll. If you really believed that, you wouldn't own so many apple products now would you?

Hey, I own plenty of Chinese knockoff products that are clearly ripped straight from the original. It doesn't make them worse. In other news, American German food is better than German food.

 

[ihuman:D]

Hey, I own plenty of Chinese knockoff products that are clearly ripped straight from the original. It doesn't make them worse. In other news, American German food is better than German food.

No it isn't, it's only "better" because Americans pump them full of fat and sugar.

 

[baryon]

Sounds fab!

 

[d0nK]

Cheaper prices for us then.

Oh no.. wait.. we're talking about Apple here. This just means even bigger profit margins for them

I'm surprised they didn't already make their own chips. Commodore made such a success of it back in the 80's. You'd have thought Apple would have been a bit more clued up by now.

 

[donutbagel]

No it isn't, it's only "better" because Americans pump them full of fat and sugar.

Depends where you go to eat.

 

[asthamapheo]

i'm waiting to see what's going to happen next.

 

[paulrbeers]

I had the same thought. Apple doesn't push even close to enough product to make owning a fab a good idea. It would be like a person buying a fully automatic tire balancer for 20 grand because they want to change their own tires once every 2 years.

You are kidding right?!?! Apple has sold over 600 million iOS devices to date since 2007. At current rate they are selling almost 200 million PER YEAR (remember they sold "few" iOS with the original iPhone compared to the iPhone 4, 4S and 5). The A# chips are in the AppleTV, iPad Mini, iPad2, iPad 4, iPhone 4, iPhone 4S, iPhone 5, iPod Touch 5G.

Intel in 2011 sold 330 Million processors:
http://www.wired.co.uk/magazine/archive/2012/10/features/the-empire-strikes-back

Intel has 12 fabs (granted not all of those making micro-processors).

So yes, Apple can definitely keep 1 Fab busy making A# processors.

If you are Apple and you are about to release your big top secret processor (say the A8 in 2014), wouldn't you rather control the manufacturing and let other contractors make your "older" processors (i.e. A5, A6, A7) since by then the cat would be out of the bag anyway?

I'm not saying they are going into the FAB business, I'm just saying there are many reasons I could see them doing it.

----------

For the same reason keeping them fabless for the other components, it makes sense to them to remain fabless for chips. Design themselves and let someone else build them, so they can negotiate for lower prices, make contracts for the quantities they need, and make sure they have them in time.
Running a fab is an unnecessary waste of energy. They won't be able to keep up with the demand they need, while substaining unnecessary costs in areas they are strangers to.

While true, you could easy farm out your "older" processors, but make your new A# processor at your own Fab. Thus increasing security around your newest designs.

 

[BigPrince]

Who says they didn't buy the report?
That doesn't mean they're allowed to quote it so that the rest of the internet doesn't have to buy it.

Well when I read the part that said "With the majority of the report behind a paywall, details remain scarce..." to mean they didn't or they could of used better wording to imply that the content could not be quoted directly...

 

[JoeIsInTheCloud]

Everything old is new again: http://en.wikipedia.org/wiki/MOS_Technology

 

[mdelvecchio]

So much for innovating. Apple just buys out start up places and rebrands it.

get real dude. seriously... buying a fab plant is a perfectly normal way to own control of your needed chip supply.

 

[thasan]

You made your own car, eat your own grown and raised food, and you built your own house too, right? So much for innovating.

eeeh.... even if he/she grows own food...god/nature already designed it unless its a gm food

 

[MegamanX]

Hope this means we get slightly cheaper prices due to direct manufacturing. But no high hopes here

lol you are dreaming there. You know as well as I do that Apple will pocket the savings.

 

[hleewell]

More interesting is the choice of partner. Everyone is guessing the Taiwanese, but my guess (based on nothing much!) would be ST. Their Crolles fab seems to be second only to Intel in their power/performance tradeoff.

STMicro and Apple together? That would be somethin'

 

[beaniemyman]

what ever you do, apple, please start making quad-core devices.

 

[firewood]

If you had a business, would you rather pay 18$ an hour to unionized workers or 2$ an hour in China?

The chip fabs in mainland China are currently nowhere near state-of-the-art around 20 nm. And even SMIC pays more than peanuts to the engineers and techs in their lead fab lines.

One guess is that Apple outright purchased a brand new fab from TSMC or UMC conditional on them running it. (between-the-lines: TSMC might sell one instead of "dedicating" one for a nice enough $sum.) It take $Billions and several years to get a leading-edge fab up and running with good yield.

 

[kkamann]

I thought TheOtherGeoff made a lot of sense when he posted this response on the AppleInsider comments on their same story. He said:

If anything, they will own the initial production of the AXx line of chips and custom ASICs needed for the next phone/pod/pad/mac, using the virtual fab designs to keep the design in house, then make their initial runs in house as well. Once established, they will push the proven etches and design to the market of fab facilities for volume and longevity.

This has 2 distinct advantages... One, all the initial proofing of chips is done under tight controls... no leaking of chip designs (ala Samsung) early in the process, 2nd, this allows Apple to drive the design independent of what the fabs are able to build at the time they bid it out... they can buy state of the art processes and prove the concept, and then drive the proven solution to the 3rd parties at less risk (which means lower cost... I mean if TMSC was asked to do something they never done before, they'll say XXbillion and we'll be ready in YYmonths, +/- 10%. Apple can now say... 80% of that, and +/- 2%, because we are giving you proven etches and processes).

I see a pattern of designing and building out a couple million chips in house, and then negotiating hard what, 6 months prior to launch of a new product with new chips in it, saying we're giving you everything, except the design then 2 months prior, the chip designs are shipped for volume launch (the 1 month after main launch), which at that point the Apple chip foundry starts tooling up for the new chip cycle.

This likely cuts 6 months exposure of IP, and keeps the Samsungs and Qualcomms of the world in the dark until product launch, then they play catch up. It also allows Apple to design chips independent of what the chip fabs can currently make (or plan to make), and then avoid the 'that's a nice design, but can you do it using our process, because we don't know how to do what you want to do" compromises)

Or it can all be a rumor.

 

[Djtrackie]

you can't fully trust a source who's name is SEMIaccurate.

 

[G4DP]

Hey, I own plenty of Chinese knockoff products that are clearly ripped straight from the original. It doesn't make them worse. In other news, American German food is better than German food.

Take it you've never been to Germany then.

If Apple buy into a fabrication facility they are losing one of the main excuses they have if things go wrong.

They can't blame anyone else. Like they were able too when the POWERPC series stalled. Funny offer a far better server grade CPU than intel could dream of.

 

[hundleton1]

I still thing buying Amd would have been grate move, buying Intel would have been awesome but they would never get away with that

 

[damir00]

As a semiconductor guy, this seems like a very bad move. It will all but guarantee they will always be at least one, if not two, generations behind the leading edge.

Just cut a check to TSMC or etc and buy allocation.

 

[Azathoth]

I thought TheOtherGeoff made a lot of sense when he posted this response on the AppleInsider comments on their same story. He said:

If anything, they will own the initial production of the AXx line of chips and custom ASICs needed for the next phone/pod/pad/mac, using the virtual fab designs to keep the design in house, then make their initial runs in house as well. Once established, they will push the proven etches and design to the market of fab facilities for volume and longevity.

This has 2 distinct advantages... One, all the initial proofing of chips is done under tight controls... no leaking of chip designs (ala Samsung) early in the process, 2nd, this allows Apple to drive the design independent of what the fabs are able to build at the time they bid it out... they can buy state of the art processes and prove the concept, and then drive the proven solution to the 3rd parties at less risk (which means lower cost... I mean if TMSC was asked to do something they never done before, they'll say XXbillion and we'll be ready in YYmonths, +/- 10%. Apple can now say... 80% of that, and +/- 2%, because we are giving you proven etches and processes).

I see a pattern of designing and building out a couple million chips in house, and then negotiating hard what, 6 months prior to launch of a new product with new chips in it, saying we're giving you everything, except the design then 2 months prior, the chip designs are shipped for volume launch (the 1 month after main launch), which at that point the Apple chip foundry starts tooling up for the new chip cycle.

Product launches are not that "in the dark" - there are a lot of other things tha tneed to be in place (scheduled production, component allocation), and the knowledge that the next iWhatever within the next 12 months is a given.

One does not simply move a 20nm from one fab to the next especially if they don't use exactly the same technology (I've worked at a major chip supplier. I know)

My guess is that it's some sort of consortium buy-in.

 

[Glassed Silver]

What is this macrumors, a freshman psych paper? Buy the report if you're going to quote all of its pertinent information.

I get the sentiment, but can understand that some folks don't want to support paywalls.
I myself for one avoid them like the plague.

Now, start the moaning about how wrong that is and that folks need to earn their bread and me being such a "freeloader"

Glassed Silver:mac

 

[ConCat]

My favourite line from the article:

SemiAccurate has a mixed track record​

I wonder if Eric chuckled a little as he wrote that.