The semiconductor industry is either being very smart, or uncharacteristically cautious, but it's not responding to full capacity by ordering a swathe of new manufacturing equipment.
Although the semiconductor industry was operating at virtually full capacity in Q3, according SICAS (Semiconductor International Capacity Statistics), nonetheless orders for new semiconductor manufacturing equipment were flat, October on September, according to SEMI, the trade body for the makers of semiconductor manufacturing equipment.
This is good news for the semiconductor device manufacturers because it means that no one is rushing to add the new capacity which could boost supply and keep prices down.
If there is no significant rise in manufacturing capacity, then lead times will grow longer, allocation will become more widespread, and prices will rise higher.
This is not good news for the industry's customers but, for an industry which, for a decade, has seen little growth in its market by value, though a considerable increase in unit volume, a period of high prices is needed to replenish its coffers and re-establish itself as a high-growth, high-margin, high-value industrial sector.
The SICAS figures show that capacity is very close to full. In Q309, the foundry industry was running at 91.9% of its capacity. 300mm wafer production was running at 96% of capacity.
Leading edge processes - those below 60nm - were running at 93.5% of capacity in Q309. Processes from 80nm to 60nm were running at 95.7% of capacity and, at 80nm - 60nm, capacity utilisation was 90%.
At such a juncture you would expect the device manufacturers to be ordering new manufacturing equipment like crazy. But they're not.
US-based members of SEMI had $756.2 million worth of orders in October 2009, down on September orders of $758.9 million.
After a period of a steadily increasing book-to-bill ration for semiconductor manufacturing equipment - from May's 0.73 to 0.80 in June, 1.06 in July and August and 1.17 in September - the ratio dipped slightly in October 1.10.
So, either the device manufacturers know something which no one else knows, or they're being over-cautious, or they've finally realised that it's smart to keep supply running slightly behind demand.
Hi David ... it's neither I suspect just the knee-jerk reaction of a brow-beaten industry that has temporarily (?) lost its way.
The logic goes like this.
Q3 wafer loading (i.e. Q4's IC shipments given the typical 3-month production cycle time) was great but this is the seasonal 'peak'.
Q1's IC shipments (Q4 wafer starts) will be seasonally down on Q4 so if we managed to get by in Q3, albeit flat out, we'll have enough capacity for Q4.
Q2's sales (Q1's wafers) are typically flat to slightly up on Q1, so that should be OK as well and this then leads us straight into the holiday season so any investment decisions can be postponed until after we get back.
Q2 (Q3's IC shipments) will grow nicely slopping up any remaining spare capacity but this will go largely unnoticed due to the seasonal holiday lull.
It won't thus be until Q3-2010 that the proverbial will hit the fan ... back from the hols facing a seasonal Q3-Q4 growth challenge with a fab capacity cupboard that is exhausted and nothing new in the pipeline.
By then it really is too late to do anything. The smart money would be building new capacity. now.
Thanks, Malcolm, that's a fascinating explanation. To me it seems so strange for the semiconductor industry to be hanging back on expanding capacity, it's usually just the opposite - over-expanding!