The semiconductor cycle is widely accepted to run 4 years and is generally considered to be at its peak in its current cycle. This common knowledge makes some of the reaction to the sector results coming through odd.

For example, Ultra Clean Holdings (NASD: UCTT) disappointed analysts last Monday and has since been knocked back over 20%. This is not simply company specific (the ridiculously heavy cost of Sarbox here claims another victim; and the VC firm that IPO'ed the company has moved on) for the financials are attractive despite a tiny customer base (cf below).

No, this is a reflection of ambitious investor expectations based upon faith in the miraculous powers of liquidity. That the Ultra Clean management commentary has to confirm that its growth is moderating (albeit with some optimism of new market offsets) in order that investors take fright is remarkable.

Ultra Clean is, it is true, a new-boy-stealing-market theme, not a pure call on the sector. But well over 80% of its sales are to three mature semiconductor companies: all-pervading giant sector generalist Applied Materials (AMAT) and smaller specialists Novellus Systems (NVLS) and Lam Research (LRCX). Ultra Clean plays largely on the cost side of its clients - it offers subsystem outsourcing without having itself to carry the capex requirements of component manufacture. Nonetheless, it is usually difficult to expand in a cyclical trough: its fortunes are greatly tied to those of sector proxies AMAT, NVLS and LRCX.

These three, especially the under leveraged AMAT, appear to have battened down the financial hatches and prepared themselves for a storm. While this is marvellous for a subcontractor like Ultra Clean who depends on healthy clients, it is also an acknowledgement of the severity of the last serious trough, triggered in December 2000 when the TMT sectors crashed shrinking the semiconductor industry by 30%. Comfortingly, though, Citi has produced research suggesting the next semiconductor trough, like that of 2005 (if you blinked you missed it), will be short and shallow.

Citi argues that the end-user TMT sector sells short replacement cycle products (mobiles, set-top boxes, playstations etc) and is in a sweet-spot driving semiconductor demand. Further, the TMT sector has matured itself to the point where it actually tracks what it is their customers want as well as running a tight operational ship. That is, they are good semiconductor industry clients unlike in late 2000 when, collectively, they were the Flaky Bunch.

Citi also provide supporting data suggesting bottoming trough trends are already in place for capacity utilisation, inventories and memory pricing. Which is interesting once the 4-year cycle is dissected. A paper by Tan and Mathews produced last January would class these Citi factors in two of three sub cycles of the 4 year circle itself: annual seasonality and the “true” two year industry specific cycle. The third factor - the ghost at the feast – is the global economic cycle.

And so, as with much in the world of finance today, it boils down to soft or hard US landing coupled with, or not, decent decoupled non-US demand. Benign (or more likely, delayed) ghostly denouement and firms like Ultra Clean et al, in an industry prepared for trouble, are already attractive (given the lack of subtlety of gimme liquid investors). On the other hand, a few more consumer shocks like that in the subprime US and what appears to be the beginning of a property tumble in Spain and prudence will win.

Sources: Semiconductor Industry Association; Alfonsa Velosa, Semiconductor Manufacturing: Booms, Busts and Globalisation; Hao Tan and John A Mathews, Semiconductor Industry Cylces

Bookmark and Share


  1. Anonymous // 4/27/2007 12:31:00 AM

    If we could agree on a definition, we could bet on that outcome, settling up in food and drink in, say, an impoverished Barcelona.

    aka CB

  2. RJH Adams // 4/27/2007 05:15:00 PM

    All for it - set out your defs.

    Just have to decide which way to bet...


Related Posts with Thumbnails