Intel (NASDAQ:INTC) announced that it would not install the equipment in its upcoming semiconductor manufacturing plant, known as Fab 42, at any point in the near future. The reasoning behind this should be obvious to anybody following the company: Intel has had a very serious problem growing its unit volume shipments largely on the back of two main drivers:
- It's gotten far more limited traction in the mobile market than expected back when the company first started building the fab.
- PC unit shipments have been on the decline rather than growing, as Intel had expected several years ago.
Why stop building this fab?
While semiconductor manufacturing rival Taiwan Semiconductor (NYSE:TSM) has had problems meeting the demand for its leading-edge wafers, particularly at the start of a new node, Intel has seen a demand problem. Without meaningful exposure to high-volume mobile chips today, and with the company's next-generation LTE-Advanced basebands (as well as upcoming integrated baseband and apps processor) built at TSMC during 2014/2015, Intel hasn't seen wafer volumes increase.
So, Intel had two options. The first option would be to let the shell of the factory (i.e. the building) sit idle without any equipment in it at this time. This lowers capital expenditures and prevents Intel from being hit with further excess capacity charges. The second option would be to move the equipment in, and pay for it, but have the factories sit idle. That would waste a ton of money keeping the lights on and the workers employed, building nothing. It's a no brainer.
How Intel can better utilize capacity
The first problem that Intel faces is that its next-generation, likely high-volume smartphone parts known as SoFIA will be built at TSMC during late 2014 and most of 2015. These chips will be transitioned to Intel's own 14-nanometer process in late 2015, but until then, the vast majority of Intel's incremental mobile shipments won't even be from Intel's own manufacturing plants. It hardly makes sense to build more capacity.
That being said, once Intel does move these components in-house, not only does TSMC lose this business, but Intel's utilization of current factories will get a whole lot better. This could serve as a major gross-margin tailwind, particularly since even a modest share of the smartphone market -- about 10% -- would translate into more than 100 million units per annum. That would significantly increase Intel's wafer scale, even if the mobile chips are smaller than their PC brethren.
This isn't doomsday
Yes, it's absolutely a problem that Intel doesn't see the need to grow capacity, since it implies that the number of chips sold that were built within Intel's factories isn't growing. This is a problem that needs to be addressed, and will only come from Intel's mobile efforts. Intel has all of the ingredients to really succeed in mobile -- and to do so brilliantly, at a good margin structure. But it needs to have the right designs.
Intel got closer in the 22-nanometer generation, but at 14-nanometer, Intel could finally have a decisive lead and could really gain a lot of share. If Intel succeeds at 14 nanometers, then by the 10-nanometer generation, Intel could win this war pretty decisively -- and it could finally need to fill Fab 42.