Intel's decision to make chips for other companies, such as Achronix and Tabula, seems strange at first look when you consider that Intel's aim seems to have been for a couple of decades: maximise profit by remaining as close to monopoly status as possible. Everything else is secondary.

Viewed from that perspective, who would wants to trade a gross margin of higher than 60 per cent for a figure that has to be no more than half that? That is where foundries need to be to make their services make sense to fabless chipmakers.

It's hard to imagine the Intel agreement with its foundry customers as being anything less than stringent, especially in what they can discuss in public. Bits leak out but they are doing more to add to the overall mystique of the chip giant's plans than anything else. Intel's moves so far have suggested that it is moving gently and selectively into the foundry business. But as his company launched its first series of Intel-made programmable-logic ICs, Achronix chairman John Lofton Holt hinted that there are many more customers of Intel's manufacturing arm than the trio that has already been publicly announced. This suggests a major move that builds volume but does not do a great deal more. Why go to these lengths?

If the plan helps support higher gross margins overall, it starts to make sense. You could call it the WC Fields strategy: never give a sucker an even break. Intel's management has found that squeezing competitors out of the business or boxing them into a position where effective competition is impossible it can make a lot of money.

The threat to Intel's continued dominance of the chip business comes from a collection of companies that don't even make their own silicon. They contract it out to the likes of TSMC. And they have pretty much established the ARM processor architecture as the standard for personal electronics. This is not good news for Intel's x86 and it's future profits. Architectural dominance has underpinned Intel's position at the top of the chip industry for close to two decades but the company has lost out to a processor family that has delivered, on average better power efficiency. But problems in manufacturing across have opened up an opportunity for Intel to starve out its biggest threat. Holt claims Intel has a head start of more than two years in terms of process technology and it has plenty of money to spend on capacity.

Despite being the third largest producer of finished silicon wafers, the Taiwanese company remains significantly smaller than Intel and number-two chipmaker Samsung Semiconductor. The top two companies are spending heavily on fab expansion in the expectation of pulling away even further.

By offering complementary suppliers access to a more advanced process and use that to make the x86 more attractive to OEMs, Intel can potentially starve direct competitors without slowing down the development of devices that are important for Intel's own silicon. By reducing the available slots to Intel's competitors, the company will make it harder for TSMC to spend to keep up and limit the company to supporting older processes in much the same way TSMC has forced many of its own competitors out of the way.

Samsung represents a threat to Intel's position: it is the main supplier to Apple for anything that does not have a Macintosh logo on it. But one competitor in the foundry business is more manageable than the current situation and even Samsung may find it challenging to keep pace with Intel on logic processes -- the Korean company's speciality remains memory chips.