Intel’s plans kill supplier flexibility

Intel is about to make another huge mistake due to the sort of inflexibility which bought humanity disasters sort of disasters as the charge of the Light Brigade.

Rumours are circulating that Haswell will be the last chip which is not built into a System on a Chip design.

It appears that Intel will only offer mainstream desktop chips in BGA packaging and move to Land grid array (LGA) and micro pin grid array (µPGA) packages.

Practically this means that the age of swappable chips is over, and the new range of chips will have to be soldered into the motherboard as part of a factory process.

Intel probably had the idea from the mobile market, where Systems on a Chip have been around for ages. However, this market gives the channel a warning of things to come, and highlights the danger of Intel’s way forward.

For years it has been possible for resellers to design PC hardware packages for customer use. Making a wide range of different chip and hardware combinations has been the mainstay of the industry. Chips can be placed in motherboards suited to different customers.

Post Haswell, manufacturers will have to stockpile motherboards at their plant. They will not want to do this with too many, because there is a large risk that they could be left with a pile of motherboards that they don’t want. To fix this, they will limit the type of boards that they will supply and commoditise the product.

This will mean that the number of options that will arrive in the channel will be much more limited and impossible to configure.

Instead of having a wide range of boards and chips that the reseller can mix and match, Resellers will have few options to play with.

Fewer options for Resellers will make it harder for them to differentiate between each other and they will be selling increasingly similar products.

On the surface, Intel should not care. After all, it can argue that it is still selling its chips, it is just making sure that they work better.

However, it is this sort of arrogance which is causing it to become unstuck. The idea only works if you have a virtual monopoly, which until recently was the case.

By the time that Haswell is finished, Intel will be facing a lot of pressure from ARM. While it is unlikely that ARM will have all the finesse of Intel chip designs, they will be offering flexibility. Resellers wanting that flexibility will suddenly find themselves ARM or AMD champions.

This is clearly something that Intel would not want, so why have they gone down this route? The answer appears to lie with the success of the mobile market, which is almost entirely SoC.

Their chipmakers can make sure that their chips slide seamlessly under the bonnet of devices with the minimum effort. But this has resulted in products which are more or less have the same functionality. This is fine for the mobile world where the difference between products is usually marketing. Intel might be justified in thinking that the message of the mobile world is the rise to power of the system on a chip and consumerisation. It would be wrong. The rise of importance of mobile is based on a philosophy of flexibility and moving to systems on chips on PCs is the exact opposite of that.

Instead of working out how to limit people’s ability to make products, Intel should be coming up with new ways to enable its channel come up with radical ideas as to how to use its chip. Limiting itself still further could do it damage and cause it to lose ground to those who have more flexible business models, such as ARM.