backtop


Print 18 comment(s) - last by ianweck.. on Jan 16 at 12:08 PM

It's upgrading other Chandler, Arizona facilities instead

Intel is putting a halt to the opening of its advanced chip factory in Arizona and will instead upgrade its other facilities in the same area. 

According to a new report from Reuters, Intel's new chip facility -- called "Fab 42" -- is being put on ice for the moment due to its lack of "capital equipment" needed for creating the cutting-edge chips. 

Intel spokesman Chuck Mulloy said that the facility has been built and has the basics like heat and air conditioning, but still doesn't have the main, expensive manufacturing tools needed for the chips. 

Fab 42, which is located in Chandler, Arizona and was expected to be a $5 billion project, was meant to be a facility for creating Intel's most advanced technology, like 14 nanometer microchips with over 100 million transistors on the head of a pin. 


[SOURCE: xbitlabs]

But the Reuters report suggests that Intel's PC sales have decreased due to the emergence of mobile devices. This development has hit the tech company's wallet, making it difficult to complete the new Arizona plant. Global PC shipments fell 10 percent in 2013, according to Gartner, and Intel said it expects its revenue to be flat in 2014 compared to last year.
 
However, it looks like Intel will put some money into upgrades at its older facilities in Chandler, Arizona. Intel has traditionally made its 22 nanometer chips in these factories, but they're being upgraded to also make the 14 nanometer versions, as this is likely cheaper than outfitting an entirely new building with the necessary equipment. 
 
In addition, Intel hired over 1,000 new employees for the new plant since its construction started in 2011, and "many" of these new hires will work in the older plants now. 
 
Earlier this month, Intel CEO Brian Krzanich revealed some big plans regarding Intel's move toward wearables -- which will be powered by Quark chips -- and that it wants to power 40 million tablets sold this year alone. 
 
In November 2013, it was announced that Intel created a new division for internet-connected devices. 

Source: Reuters



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

450mm delay is the real reason
By The Von Matrices on 1/15/2014 12:01:38 PM , Rating: 5
The reason for all this is that ASML (the biggest company producing semiconductor processing equipment) has halted production of machinery to handle 450mm wafers.

Intel (and many other companies) want to move to 450mm wafers (from 300mm) to reduce costs. Since you occasionally also need to retool some facilities to produce smaller (14nm) chips. It makes sense to do these simultaneously so that you reduce downtime at the facility.

Regarding why Intel constructed a new fab, the issue is that when you switch to 450mm wafers, basically everything in the existing 300mm facilities need to be replaced. In addition, the building itself may even need to be renovated, requiring it to be offline for an extended period. This is undesirable since the old equipment could still be generating revenue. If you just construct a new facility, you have no downtime at all (you can continue to have the old equipment in the old facility build chips).

Regarding why Intel is leaving the new facility empty, with the 450mm equipment delay, Intel is stuck at 300mm wafers for another generation. It makes no sense to buy soon to be obsolete 300mm equipment to fill the empty building. The best choice is what Intel did - for now upgrade a few existing 300mm facilities to 14nm transistors and leave the new building empty until 450mm wafers and 10nm transistors come around.




RE: 450mm delay is the real reason
By ianweck on 1/15/2014 12:41:16 PM , Rating: 2
300mm equipement won't be obsolete any time soon. Intel continued to produce chips on 8 inch wafers for years after the 300mm tools came out.


RE: 450mm delay is the real reason
By The Von Matrices on 1/15/2014 2:55:18 PM , Rating: 2
I agree, 300mm technology is not obsolete in the technical definition. However, the semiconductor world is focused upon cutting costs to remain competitive, so it will quickly become economically obsolete.

Investing in a new 300mm plant to make chips today makes zero sense when two years later your competitors can undercut your prices using 450mm wafers. Keeping existing technology running is a different matter entirely and makes much more economic sense.


RE: 450mm delay is the real reason
By FaaR on 1/15/2014 3:02:51 PM , Rating: 2
Any word on what is causing the production stoppage of 450mm equipment?


RE: 450mm delay is the real reason
By ianweck on 1/15/2014 5:54:40 PM , Rating: 2
If you think very many companies will be ramped up on 450mm wafers in two years you're crazy. It takes two years just to build a fab, let alone install and qualify the tools. Especially if their number one supplier (ASML) of backbone tools (scanners) has stopped production, which is news to me. I'd like a link to that if you have one.


By The Von Matrices on 1/15/2014 7:36:07 PM , Rating: 2
The major ASML customers disagree on the 450mm specifications, so ASML has "put it on the back burner" until an agreement can be reached.

A formal report is due by ASML in February. The information below came a few weeks ago:

http://www.bits-chips.nl/artikel/asml-zet-450-mm-o...

Part of the issue is also that the advantages of 450mm production are diminishing and it seems like Samsung is getting cold feet:

http://450mm.com/blog/2013/12/19/diminishing-case-...


RE: 450mm delay is the real reason
By ilt24 on 1/15/2014 5:45:42 PM , Rating: 2
quote:
Intel continued to produce chips on 8 inch wafers for years after the 300mm tools came out.
Intel is still running a 200mm (8") fab in Hudson MA, although they did announce they would be closing it at the end of this year.


RE: 450mm delay is the real reason
By ilt24 on 1/15/2014 5:57:57 PM , Rating: 2
While the shell of Intel's new fab in Arizona will handle 450mm wafers, they were going to start with 300mm wafers on a 14nm process. At the time they decided to go ahead with that factory, Intel's best case senario didn't have them making production 450mm wafers until 2016.

The reason for the change is a reduction in demand. Intel decided to build that factory in late 2010 when they, and others, beleive the market for PC processors would continue to grow. That didn't happen, and the market has actually shrunk and while they now plan/hope to sell lots of atom processors into the tablet, phone and internet of things markets, the die are much smaller so they aren't going to fill the wafer demand they thought they would need when they gave the go ahead for that factory.


Intel/Apple
By melgross on 1/15/2014 5:52:57 PM , Rating: 3
It's too bad Intel didn't do that deal with Apple that Ottolini regretted so much. Apple will need close to 300 million SoC's this year, at an estimated average price of $22. That's $6.6 billion of chips in one year, likely increasing by 20%, or more per year as time goes on.

Now, that going to be spread between Samsung and TSMC. A shame. Apple going to Intel would be much better for both, and would hurt both of their rivals.




RE: Intel/Apple
By ianweck on 1/15/2014 5:58:01 PM , Rating: 2
Competition is good, and I see no reason to give Intel yet more fab business. The more fab facilities that exist the better, in my opinion.


RE: Intel/Apple
By melgross on 1/15/2014 6:12:09 PM , Rating: 2
Well, personally, I'd rather give business to more companies here.


RE: Intel/Apple
By ianweck on 1/16/2014 12:08:56 PM , Rating: 2
As would I. So buy something made by Micron. Or TI, or Global Foundries. Samsung has production in the U.S. as well even though they are Korean. Lots of ways to support the local economy.


RE: Intel/Apple
By ilt24 on 1/15/2014 6:14:44 PM , Rating: 2
I think even had Apple originally went with Intel's chip for the iPhone, they would have moved to their own processors as their volumes grow.

Using your numbers(300M * $22) that's $6.6B to buy chips vs. buyinf 300K wafers @ $6K each for $1.8B and then paying $2/die * 300M for assembly and test gets them to a total of $2.4B if they design their own chips and outsource manufacturing. That's a $4.2B difference.



This makes sense to me...
By retrospooty on 1/15/2014 11:06:12 AM , Rating: 2
There are already so many here and it doesnt look like chip sales will be increasing by any large margin. Why build yet another one?




RE: This makes sense to me...
By Argon18 on 1/15/14, Rating: -1
By retrospooty on 1/15/2014 11:19:18 AM , Rating: 4
Ptetty sure Intel did the ROI analysis on it already. That was the root of the descision. They are going to upgrade old FAB's rather than build another new one (BTW they just built a new one)


Bad for the industry
By Sahrin on 1/15/2014 11:12:05 AM , Rating: 2
Intel spends an ungodly amount of money on R&D; any slowdown for them represents a transfer of revenues from Intel to other companies, many of whom spend *less* on R&D than Intel.

Discounting all other factors (eg, Intel bribing device makers not to use competitor's chips), Intel doing well is generally good for the industry because they spend like a drunken sailor on R&D, which pushes the process node forward.




RE: Bad for the industry
By retrospooty on 1/15/2014 11:14:00 AM , Rating: 3
This isn't about R&D at all... it's manufacturing capacity. They have a TON of it and don't need more.

R&D is unchanged.


"It looks like the iPhone 4 might be their Vista, and I'm okay with that." -- Microsoft COO Kevin Turner














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki