It will be quite interesting to see whether the so-called fab shortage does happen in 2010, given that there might not be enough capacity to handle demand... (Are Indian fab backers listening?) In the meantime, I am awaiting Malcolm's responses. So, watch this space!
Executive overview
"On the face of it, April's data showed a 7.6 percent decline in total semiconductor sales versus March 2009 but, after adjusting for the five-week long March, this translates into a whopping 15.5 percent growth. The corresponding numbers for total ICs were minus 9.3 and plus 13.4 percent respectively. This is the strongest April month-on-month growth since April 1996. That's the good news!
Given the still delicate state of the global economy, this growth is not however directly driven by increased end-user demand, instead it is purely a correction to the steep Q4-08/Q1-09 inventory declines.
In other words, the markets clearly over-reacted to the September 2008 global economic collapse, sucking the supply chain dry, paving the way for this counter-balancing period of inventory replenishment. Historically, we can expect this to last through to Q3-09. Beyond that, growth will depend on the underlying end-market demand.
What then does this mean for the 2009 outlook? Whilst much of the current industry tittle-tattle focuses on the 'green shoots of recovery' debate (are there/aren't there?) and/or the 'shape of the downturn' (V, U, W, sharp, stretched, extended, etc), we prefer to take a more sober look at the underlying trends.
As mentioned in our May Global Semiconductor Report, at minus 24.2 percent growth, Q4-08 was a little worse than our 22.5 percent January IFS forecast, whereas Q1-09's 15.5 percent fall was slightly better than out minus 18 percent number.
The counterbalancing overall effect of the two was to put the 2009 market slightly ahead of our official minus 28 percent forecast, to minus 25.3 percent. In our book, this does not constitute a 'forecast revision' given that basic forecast assumptions and analysis had not changed; it was merely 'finetuning the number'. So far, so good!
Q2-09, however, might well be different in that it now looks to be coming in with the growth in the 4-5 percent range, versus our -2 percent January estimate. If true, it would represent a material change to our 2009 forecast, improving it from -28 percent to -21.3 percent, assuming that the second half of the year rolled out as planned, and removing most of the down-side risk potentials. Our verdict on this will be presented at our upcoming July 21st Mid-Term Industry Forecast Seminar in London.
In the meanwhile, here is a snapshot of what is actually going on in the markets. The first quarter was clearly a difficult time for the industry, the combined effect of the global recession on top of the normal Q1 seasonality weakness. Not quite a knockout punch, but a real double-whammy. Based on a reasonable cross industry sampling, the overall result was a net 12 percent fall in electronic equipment sales versus Q1-08.
Aside from government/military -- the only sector to grow -- every market and geographic region was negatively impacted, with Japan and Taiwan/China the worst hit, the latter however being the first to show a rebound.
Source: Future HorizonsLooking at the key mobile and PC industry sectors, both of these have been hit badly by the discretional consumer spending slowdown, with Q1-09 phone and PC unit sales down 16 and 20 percent respectively versus Q1-08.
Given the magnitude of these declines -- all markets, all sectors, all regions, all customers, consumers and enterprise -- the industry and chip market exited the first quarter in remarkably good shape relatively speaking. That is, not to say that it will be plain sailing hereon out, far from it, but that the industry has clearly weathered the worst of the storm, bloodied but (mostly) not beaten.
The 12 percent, 16 percent and 20 percent OEM, mobile and PC sales declines are key industry benchmarks in that they represent the absolute worst-case full-year scenarios. Our January 2009 forecast called for 8.2 percent, 15 percent and 22 percent declines here, respectively, which are all well in line with the way the market is unwinding. Given that we expect the Q1-09 12:12 declines to improve as the year rolls out, the downside risks to our forecast are clearly diminishing.
From an economic perspective, our 8.2 percent electronic equipment industry decline was based on the then IMF's World GDP growth forecast of plus 1.8 percent. This however has been subsequently revised down three times, first to +0.5 percent on January 28, then to -0.5 to -1.0 percent on March 13, to the current forecast of -1.3 percent in the April 2009 World Economic Output Report.
Intuitively, one would worry that these downward revisions ought to force a corresponding revision to the electronics equipment market; in reality this does not seem to be the case.
Which in turn beggars the big unanswerably question: "to what extent are these downward revisions to world GDP growth the cause or effect of the electronic equipment industry decline and what impact will their downward revision, and for that matter subsequent recovery, have on the 2009 and beyond electronic equipment industry absolute growth number?"
The bottom line answer? No one really has the faintest idea; moreover, it is impossible to calculate. While anecdotally and intellectually, there is an obvious link between GDP and electronic equipment industry growth rates, the electronic equipment industry represents only 2-3 percent of total world GPD. In contrast, at their peak, the financial derivatives markets totalled 120 percent of world GDP!
With recession, cutbacks clearly hit the electronic equipment industry early, as both enterprises and consumers hit the 'stop spending' button relatively quickly. In the case of the 2008 downturn, for 'relatively quickly' read 'instantaneously fast'. The impact on the chip market is immediate, aggravated by the associated component and WIP inventory burn, with overshoot inevitable.
Inventory levels clearly stabilised during Q1-09 and are being rebuilt in Q2, most probably targeting an electronics equipment production level 12 percent down on this time last year, i.e. in line with first quarter actual. This being the case, the industry will have adjusted much faster than in previous cycles, with today's inventory imbalance levels already peaked and much more in line with the 2H-06 'course-adjusting' excess than the post-dot com bubble burst 2001 flood.
To summarise, the downside risks to the 2009 market are clearly abating with our 13 percent third quarter growth forecast still looking reasonably robust, given the current inventory rebuild plus a touch of seasonal strength. Likewise, it is still credible for this to be followed by a seasonally weak 3 percent forth quarter growth given the normal end of year inventory clean out.
With the global economic recovery then starting to gain traction in 2010, a 'normal' quarterly (-2 percent, +2 percent, +14 percent, +3 percent) 2010 growth pattern would be reasonable, yielding a 2010 annual growth of around 17 percent, well in line with our '15 percent with lots of upside potential' January 2009 forecast.
While the slightly revised 2009 quarterly growth pattern would call for an (upwards) formal forecast revision to the actual 'growth number', the underlying market analysis and assessment presented at our January 2009 Forecast Seminar will not have materially changed, either for 2009 or 2010.
Market summary
We tracked the worldwide and European 12/12 industry growth rates for ICs, Opto, and Discrete Devices from January 1998 to date. These show the current month as compared with the same period 12 months ago, and are a useful industry momentum indicator. We also show 15-month rolling worldwide and European sales by major product category.
Industry capacity
Overall MOS wafer fab capacity decreased by 7.8 percent in Q1-09 versus Q4-08, from 2110.4k 200mm equivalent wafer starts per week to 1945.1k. Whilst no category was immune, the decreases were highest in the 200mm and below wafer sizes, and at 150nm and tighter feature sizes.
These cutbacks add to the previous quarter's 1.6 percent decline and compared with a 1.7 percent quarterly growth this time last year. Whilst some of the decline can be attributed to closing of older lines due to the recession, for the most part they are the direct result of the deliberate slowdown in capital expenditure that began mid-2007, well before the current recession started.
The 300mm wafers now account for 50.1 percent of the total MOS capacity, up from 48.2 percent in Q4-08 and 40.7 percent from the same period last year. 300mm wafers now account for over half the total capacity, with 200mm in second place at 37.7 percent, down from 39.1 percent in Q1-09 and 45.7 percent in Q1-08.
At 733.5k wafer starts per week, Q1-09 200mm capacity continued its absolute value decline, from 826.0k in Q4-08, a fall of 11.2 percent. 200mm capacity is now down 22.8 percent versus the same period last year.
Even advanced capacity (i.e., 0.08 micron and below) declined, primarily due to the DRAM firm's bankruptcy problems. Only the 120-159nm category, escaped with capacity up 1.4k wafer starts per week, or 0.6 percent.
Despite sizeable capacity decrease, Q1 utilisation rates plummeted still further to 57.2 percent, from 87.5 percent in Q3-08 and 68.4 percent in Q4-08. The comparable figure for Q1-08 was 90.7 percent. Advanced IC capacity, i.e., 0.08 micron and below, also fell to 69.9 percent (from 84.4 percent in Q4), whilst 300mm and 200mm wafers checked in at 72.8 percent (Q4 = 83.2 percent) and 44.0 percent (Q4 = 55.5 percent), respectively.
The fall in Q4-08/Q1-09 utilisation rates was a result of massive order cancellations and demand collapse triggered by the September 2008 Lehman Brothers ollapse and was much faster and deeper than the 2001 dot-com driven recession. Q1 is however expected to be the trough, with rates climbing back fast in Q2 and thereafter.
Given the significant cutbacks in capex since mid-2007, we expect to see utilisation rates trending back to the 90 percent 'full capacity' threshold much faster than in previous recessions, accelerating the supply-side recovery dynamics by at least four quarters.
Fab shortage waiting to happen?
Looking ahead to 2010, demand should start to accelerate in line with the anticipated global economic recovery, tightening the capacity screw still further. And, with 2010's capacity fixed by 2009's spend, at a currently estimated $20 billion, this spend represents barely one third its 2000 $60 billion peak. That means, in round numbers, 2010's new capacity will be only 40k 200mm wafer starts/week minus any capacity closures.
We have said it before and we will say it again. There will not be enough 2010 capacity in place to meet demand... this is a fab shortage waiting to happen!"
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.