The big news from yesterday wasn't the hard drive warning. The big news really came out of the slide show presentation in the UK on chip demand exceeding capacity in 2012 and the disruptions this will cause.
See topic "2012 Foundary Fab Shortage"
And slides at:
“Semiconductor Market: Changes For The Year Ahead”
One of the surprising things is that this news comes right out of ARMs backyard.
Well, everyone (except Wall Street and the tech media) knows at this point that I believe that ARM has a serious fabrication problem. One that will cause significant disruption to the business. What the slide show added to this picture was important information about the chip cycle in 2012 - primarily that demand will outstrip capacity.
Now keep in mind that these comments are significantly restricted to the fabless/foundry sector of the market. Yes, this is the ARM sector.
The slides demonstrate that the fabless/foundry model is essential broken in that it doesn't provide the means for timely construction of new capacity. And that the impact of this will arrive in 2012.
The production problems with the foundries as die sizes shrink have left TSMC as the single source for a huge proportion of the market. This would be good if TSMC were adding capacity fast enough, which it's not, and if it were a good idea to single source everything at TSMC, which it's not.
"Zero Net New Capacity Growth For 2012 It IS Different This Time (But No-One Believes It, Especially The Fab-Lites & OEMs!!)"
The above from the slides indicates that the ARM world is in denial about the situation.
ARM/TSMC therefore has two very significant problems:
1.) Aging technology as fabrication slows. 28 and 20nm are coming to market very slowly. 14nm is years and years away.
2.) Insufficient capacity due to insufficient capital spending on new fabs.
The potential for disruption in 2012 is large.
And as Semi-Accurate has suggested, this will open a very wide door for Intel who is in the opposite situation from ARM/TSMC. Intel has spent huge and incredible sums to continue to build its fabrication capacity.
This will come to a head in 2012 when those companies who cannot get their needs met at TSMC will find Intel's leading edge technology and capacity extremely attractive.
This will be a sea change in the business and a trend likely to continue as no other company on the planet can match Intel's "every eight quarter" fabrication cycle.
Look for companies like Qualcomm and Apple to turn to Intel in 2013 when 14nm production arrives. But it all starts in 2012...
'Now that's funny. They are "beginning their research". Okay, give them 3 years for research (it took Intel ten) and then 3 more years to plan and build and find the money for the fab. And then another year to integrate 450mm wafers. Yep, about 7 years from now they might start production. But by then it will all be over but the crying and the shouting'
LOL, you are just making stuff up again plucking WAG numbers out of the air. Also building the fab and doing the process would occur in parallel not serially.
"Tablets don't cost $30K so not really relevant as an example of impact damage ;-)."
[Hmmm. But guess what? Tablets have much higher volumes. The potential for disruption is as great or greater...]
"TSMC are also planning 450mm wafers for 14nm so fat chance of any foundry capacity problems in the long run."
[There is zero chance of TSMC having 14nm volume production in 2015 and you know it. And moving to 450mm wafers at the same time will likely delay production by another year or two. We're talking at least 5 years before this ship might come in. And the problems start to ramp in 2012...]
"He also said TSMC will begin research on the 14-nanometer process next year and expects to begin volume production on the node in 2015. They’ll be using 18-inch wafers to process 14-nanometer chips because it helps bring production costs down, which should be music to Apple’s ears."
[Now that's funny. They are "beginning their research". Okay, give them 3 years for research (it took Intel ten) and then 3 more years to plan and build and find the money for the fab. And then another year to integrate 450mm wafers. Yep, about 7 years from now they might start production. But by then it will all be over but the crying and the shouting...]
'Look for companies like Qualcomm and Apple to turn to Intel in 2013 when 14nm production arrives. But it all starts in 2012... '
Intel will not be producing ARM competitor chips. You sure have a vivid imagination ;-).
Brian M. Krzanich, the Senior Vice President/GM, Manufacturing and Supply Chain at Intel addressed the capacity issue at his UBS presentation.
He said the cost of having over-capacity might be a half a billion dollars whereas the cost of under-capacity is BILLIONS of dollars. He said Intel is extremely good at managing its capacity and deferring it until they need it. But they would much rather have too much as too little due to the extreme economic cost of having too little.
ARM is not good at managing its capacity. It can't be. It's fabless. Therein lies the rub. They don't have fabs so as not to have the huge costs. The downside is the inability to bring new capacity on in a timely manner. Those Silicon South West guys in the UK say it's going to bite them in 2012. And as Intel has commented the cost of under-capacity is billions.
Take a look at their Nissan example whereby the lack of a $10 IC cost Nissan $90 million in sales.
And from another slide:
The Shape Of Things To Come?
OEM Company 'A' Places Module / Box Order On CEM Company 'B' Whereupon CEM Company 'B' Places Component Order On Chip
Company 'C', Who In Turn Places Wafer Requirements Order From Foundry Company 'D'. Job Done, Sit Back & Relax
Foundry Company 'D' Suddenly Cannot Supply Chip Company 'C' With Enough ICs; Forced To Allocate Available Production
Chip Company 'C' Receives Only Partial Delivery & Cannot Supply CEM Company 'B' With Enough ICs; Forced To Allocate What's Available To Its Customers
CEM Company 'B' Cannot Supply Enough Modules Or Finished Units To OEM Company 'A'; Forced To Allocate Available Units Across Its Customer Case
OEM Company 'A' Suffers Shortfall In Supply, At Best Missed Sales, Worst Case, A Full Line Stoppage
So, what is the fix to the problem? Well, there really isn't one. They might have something in three years. In the meantime the only fix is to go to someone who has planned well and might have excess capacity. That would be Intel...
"Intel will not be producing ARM competitor chips. You sure have a vivid imagination ;-)."
You can argue it with the Semi-Accurate author if you want.
I think his point is that when they can't get production out of the foundries they won't have any other choice.
That's why I'm saying the projected absence of capacity in 2012 is big news. Did you go look at the slide presentation?
If so, tell me where you think their logic breaks down...
Seems to be:
"Welcome to the Silicon South West network - the organisation that provides regular networking events, news and national and international promotion for the region's microelectronics cluster."
"Simon Bond is CEO & Founder of the Silicon South West network which promotes the region’s electronics design cluster particularly around start-ups. He works as Director of Innovation and Networks at Bath Ventures, the University of Bath’s IP commercialisation group and is a founder member of the SETsquared business incubation team in partnership with the universities of Bristol, Southampton and Surrey.
Simon also runs the openMIC – the Mobile Innovation Camp network for application developers; the Assisted Living Action Network for assisted living technology developers; and is Vice Chair of the Low Carbon South West network which is focused on the development of new low carbon businesses; Prior to joining the University of Bath he worked for 18 years as a communications & public affairs professional with companies including AT&T, BT, Huawei, Marconi, NYNEX, Lucent Technologies, & Telewest and was founder of the broadcast TV start-up City Television."
IMO, INTC and investors are still digesting the reduction in forecast and what it means.
Note that Intel sells a CPU to customers who must then integrate with other components for their finished product. Many of those components, I would imagine, are also subject to the same availability problems.
There have also been some fairly large option trades that could put pressure on the stock price.
Intel draws bearish trade at resistance
"If that's the real news why is INTC down hard again and ARMH relatively flat when not up."
Um, because what Wall Street knows about fabrication you could fit on the head of a pin.
As a matter of fact what Wall Street knows about the Intel story you could fit on the head of a pin.
Wall Street still doesn't get the Cloud, big data or the IoT, much less fabrication. That's why I endlessly post on these subjects hoping by osmosis that the word might get out and into the world of the technologically brain dead.
Other than the Semi-Accurate story and the UK presentation, have you ever seen a media article talk about the longer term implications of ARMs fabrication problems?
Even on this board, people don't understand the fabrication dynamics. I admit it's complicated. But it's necessary to any fundamental grasp of what is going to happen financially in the next few years.
It's like ARMs version of subprime. A major dynamic that no one wants to recognize until it's too late.
Intel will be the beneficiary. I just thought a few people might want to know...