Intel has a dilemma. It needs to enter the mobile phone market but it is not sure if that market will generate the revenues its needs.
Last week’s foundry deal with TSMC is Intel’s latest attempt to unlock the mobile phone chipset market.
It needs to do this because the microprocessor market is changing.
Intel has dominated the microprocessor market for a quarter of a century with the x86 microprocessor architecture on which the PC market has been based. It generated enough money to invest in wafer fabs needed to keep the PC makers supplied with chips.
But the definition of what constitutes a PC is changing, and along with it the processor and source of supply is also changing. PCs are now internet terminals and increasingly seen as handheld devices.
Intel needs to sell low-power processors which will run in netbooks and smartphones.
Crucially, it needs to sell processors to mobile phone companies. And that is something it has found difficult to do. The TSMC deal is aimed at getting its Atom processor designed into the system-on-chips (SoCs) used by mobile phone companies.
Is this the start of a plan to scale back all Intel’s chip fab investment, or just designed to address the mobile processor market?
Only time will tell.
But does Intel’s deal with TSMC suggest that one day Intel will be forced to do something it has never done and license its processor architecture to other semiconductor companies?
It could be argued that this is something it will have to do to make any serious impact on the smartphone processor market, which is dominated by ARM and its licensing strategy.
But there is more than just a wish to protect its intellectual property that keeps Intel from following ARM down the licensing route.
Intel is 70 times bigger than ARM and it is doubtful whether a company of Intel’s size can sustain itself from licensing IP rather than selling processors.
With the PC market slowing, Intel knows the growth it needs is likely to come from mobile phones.
To get into phones it must change its business model, but an IP licensing route cannot create acceptable returns.
That is Intel’s dilemma.