Online information source for semiconductor professionals

I am taking the lead from new Editor-in-Chief at Solid State Technology magazine Pete Singer, who reported the news from their own ConFab event being held this week in Sin City via a newsletter. ConFab is a closed event to the rest of the media but I have posted a link to Pete’s story below. 

The conventional wisdom of a wafer size transition every 10 years is flawed, according to Doug Grose of AMD, wrote Singer in a newsletter yesterday.

Taking a swipe at Intel over its justification for the move to 450mm wafers due to the fact that such a move is a historical and closely tied mantra of Moore’s Law, Grose noted that a wafer transition actually takes much longer if you look at it from an actual start of volume production perspective. Indeed, the timings, according to Grose, are actually slowing if you look at the transition from 200mm to 300mm.

That would put such a transition in the 2016 timeframe rather than the 2012 estimation that Intel, Samsung and TSMC have aligned to.

In his keynote speech, Grose was, apparently, all nice and cuddly towards the equipment suppliers, supporting their serious objections towards the 450mm transition based on the billions of dollars it will cost them, especially when things are so tough right now.

Bless!

But the biggest nail in the coffin for 450mm was his point over actual cost reductions achieved by using larger wafers. Grose claimed that by his calculations, only a 4 percent cost reduction could be attributed to the larger wafer size transition. The rest of the 30 percent cost reductions achieved every year came from die shrinks to yield to overall productivity improvements.

This is somewhat of a smack in the face for Intel and ISMI, which companies have pitched everyone hard - including the trade press - insisting that incremental improvements will not provide the 30 percent-plus cost savings required going forward, and so the wafer size migration is the only course of action.

It should be pointed out that Grose isn’t saying that 450mm is dead and buried, but rather the rush to get there for 2012 isn’t necessary. Looking at the plight of AMD right now, it isn’t too hard to understand where Grose is coming from, is it?

Harking back to last year’s ISMI conference, Grose’s presentation didn’t go as far as knocking 450mm. I did get the impression, however, that there was little conviction on his part that 450mm was a done deal, especially in the aggressive timeframes being pushed.

However, Grose is now questioning the wisdom of the 450mm transition! He was reported as saying that once 450mm is here, he wants to know how the industry is going to reap the benefits from it.

This tells me that with all the efforts of a small manufacturing minority wanting to push the major suppliers down a bottomless pit of investment and questionable ROI, Grose is asking for more solid analysis on whether the move is worthwhile in the first place before jumping on board.

That might not concern Intel at all, but it will concern other borderline manufacturers as well as the whole supplier community.

There are three critical issues remaining unresolved before we will see the move to 450mm wafers: consensus, consensus and consensus.  

Pete’s story can be viewed here.