Mannerisms

Now TSMC is saying it will have its first 45nm production wafer out in September with volume production in Q108, and Intel is saying it will start its production ramp on 45nm in Q4.
Who is first? Difficult to say.

What one can say is that TSMC has a more difficult task than Intel. Intel has only to bring up one process, a high performance process, for one customer, itself.

TSMC, on the other hand, has to develop several types of 45nm process e.g. high-performance, low-power, etc and has to make sure it can be used by a vast variety of customers.

This regular rite of chest-beating is an interesting one because Intel is now the only independent IDM doing all its own R&D and manufacturing, and TSMC is by far the largest foundry.

With every other IDM now engaged in a consortium for its R&D, while outsourcing a proportion of its production to foundries, it is going to be interesting to see for how much longer Intel can enjoy its solitary splendour.

If TSMC or UMC ever got an edge over Intel on process technology, then the game would be up. It would be impossible for Intel to compete effectively against an AMD which had access to a better process.

If Intel can keep ahead of TSMC with a proprietary process which gives it a demonstrable edge in processor performance, then it can maintain its independence of the consortia and the foundry industry.

Which is why, as each new node approaches, the Intel-TSMC PR war breaks out for bragging rights on who gets there first.

But it's more than just about bragging rights. It's about whether the foundries will get to totally dominate the semiconductor industry.