Future Horizons' Malcolm Penn: This is not 2001

By Suzanne Deffree, Managing Editor, News -- Electronic Business, 11/4/2008

Conventional wisdom says "cut back in a recession." Malcolm Penn, chairman and CEO of Future Horizons, disagrees when it comes to the chip industry. Indeed, Future Horizons, a global semiconductor industry analysis company started in the 1990 recession and expanded during the 2001 bust, this week expanded again, launching its "Start Up Support" program Web site. Penn discussed with Electronic Business via e-mail the company's October report and how this financial-crisis-driven industry downturn differs from other industry downturns, what will cushion the industry's blow, why semiconductors are attractive to investors right now, and how Moore’s Law works in. What follows are excerpts of that conversation.

EDN: It's clear the economy is heading into a recession, but the big fear in the semiconductor industry is that are heading into another 2001 situation. Are we? Why or why not?
Penn: With Q4 clearly now the start of the industry slowdown, the big industry fear is "are we/aren’t we heading for a replay of 2001?" The perceived consensus (wisdom) is that we definitely are, but there is also an opinion that feels this is not the case. Future Horizons is of the "no, we aren’t" opinion. We have drawn this conclusion by analyzing exactly what causes chip market recessions.

Looking back over the 60-year industry history, downturns are triggered either by demand- or supply-side issues. In either case the common denominator is excess capacity, caused either by over investment (making capacity overshoot demand) or a demand slowdown (whether through an inventory burn or recession) making capacity exceed near-term demand. Recovery from an inventory burn is the fastest, excess capacity the longest, a recession somewhere in between.

The 2001 slowdown was unique in that it was triggered by both demand- and supply-side issues, with the industry experiencing a simultaneous collision of dot-com inflated demand euphoria, a 9-11-driven economic slowdown, a resultant massive inventory burn just as a huge amount of excess capacity was coming on stream. Everything that could have gone wrong did go wrong, the so-called perfect storm; this destructive combination has never happened before. As a result, one fab in three was redundant and the investment binge meant the problem would get worse before it could get better. It takes a year from turning the tap off for the added new capacity to stop flowing. This is why excess capacity-driven recessions take longer to recover from.

Today we have no serious overcapacity in place (pre-slowdown utilization rates are in the 90% region), capex was cut back 12 to 18 months before the slowdown hit, meaning net new capacity started to slow ahead of the economic recession induced demand reduction, and IC ASPs [average selling prices] were in the middle of a cyclical upward trend. Prior to the economic shenanigans, there was little sign of inflated demand – IC units have been running at or below there 10% per year long-term trend line – and no serious excess inventory in the supply chain, although it has to be said this is the industry’s Achilles’ heel.

For sure, the global economic recession will reduce near-term chip demand and trigger an associated inventory adjustment, but demand will soon find its new (reduced) status quo, which will certainly be lower than the current installed capacity causing utilization rates to fall further than they would normally do so in the seasonally slow first half of any year, but this will all be taking place against a background of simultaneously dramatically slowing capacity. For once, by pure luck, the industry "anticipated" a global demand slowdown and cut back on capacity 20-plus months ago. This will make the 2009 effect much milder than would have been otherwise.

"What makes 2009 different from the 10 previous downturns is that this time it’s happening during a period of prolonged fab capacity restraint."

EDN: In your opinion, how strong is the coupling between the "real" economy and the semiconductor industry?
Penn: There obviously is a strong coupling, but it is not absolutely fool proof. Looking back over the past 47 years (from 1960 to 2007), there have been seven instances where the chip market grew in value when the economy declined and two occasions when the value declined despite a growing economy. In unit terms, the WSTS data only goes back 22 years but there were three times in this period when units grew in the face of a declining economy and five times when unit demand fell as the economy grew. Needless to say there was no correlation between the chip value and unit economic growth disconnects. 2008 will be another disconnect, with units tracking the economy but market value not, whilst 2009 will see all three declining together. Given the current pump priming, the economy should be well on the way to recovery by 2010 – as with 2003 – with units and value both trending up, as well.

In general, however, these instances are nuances. … If the economy collapses it will take the chip market with it; the only uncertainty is by how much and how fast.

EDN: You state in the press release on the Future Horizons October report that the economy is "not quite king; inventory, excess capacity and ASPs also play a role." So how do inventory, capacity and ASPs look going into 2009? Will these three factors in any way help ward off a depressive downturn in the chip industry? Why or why not?”
Penn: I think I’ve answered this in my answer to question 1, aside from a commentary on the role of ASPs. ASPs are a mix and technology issue, with a dose of industry behavior for good measure. There is basically an underlying five-to-eight year cycle modulated by regular – but unpredictable – mini disruptions of unpredictable magnitude and impact.

Moore’s Law is the basic driving factor here, with economic influences, excess (under) capacity and price wars (increases) triggering the disruptions. Moore’s Law allows us to either reduce the cost (price) of a device or make a more complex device for the same cost (price). In practice we do both simultaneously. If you get the balance right, you always have new products (at a higher price) coming on stream to offset the impact of declining prices on the old parts. Volume kicks in, as well; "the cheaper they are, the more you sell," so total market value can grow even with declining prices, providing the units are growing faster than the decline.

Excess capacity tends to cause firms to drop selling prices, i.e. by buying the business or attempting to stimulate demand, easily doable given the typically low IC device variable cost. This is especially true in a regime of falling prices … what looks like a cheap price today so often turns out to be a great deal tomorrow. Then there are the price wars, as with the recent MPU one between Intel and AMD, which is more to do with hurting a competitor and/or reminding them who’s the boss. These can be dangerous, though, as once individual device prices collapse, they never recover causing an ASP decline. This in turn only recovers when the higher complexity replacement devices start to kick in.

Falling prices are not always bad, indeed in the long-term it is these declines that have caused the chip market to grow but, unlike memory, there is no price elasticity in PC MPUs – you don’t buy two PCs just because they are cheap. In contrast, halving the memory price drives the "free memory upgrade" promotions and at the same time sets the new minimum PC memory expectation level.

EDN: Are there any current trends that suggest the chip industry will grow during this economic downturn?
Penn: Unfortunately, no. Slower unit growth and an ASP decline are inevitable for 2009, but because this is against a background of prolonged under-investment and unprecedented global economic pump priming, the decline should be short lived and the rebound quite fast.

EDN: How will this downturn differ from other downturns in the semiconductor industry?
Penn: I have personally experienced nine out of the 10 industry downturns (I only missed the first one in 1960) and each one has been different, even though the fundamental cause was the same (excess capacity); it all depends on how my so-called four horsemen of the semiconductor apocalypse (the economy, unit demand, fab capacity, and ASPs) inter-react.

"For once, by pure luck, the industry "anticipated" a global demand slowdown and cut back on capacity 20-plus months ago. This will make the 2009 effect much milder than would have been otherwise."

In 2001, all four horsemen were in concert and the combined effect was a 32% year-on-year decline. In 1996, the problem was just over expansion – the economy and unit demand were both still strong. As a result the market declined only 8.6%. In the 1981-1982 economic recession, there was no previous over-investment and the market actually managed to still keep on growing, albeit a low single-digit levels; performance in the early 1990’s recession, even though it was very deep and long, was actually even better, with three out of the four years with high single-digit growth.

In general, though, growth tends to be negative in these periods, so the balance of probability for 2009 is a small (low single-digit) negative growth. It’s by no means yet a certainty, though; there is still a possibility for low single-digit growth. What makes 2009 different from the 10 previous downturns is that this time it’s happening during a period of prolonged fab capacity restraint. That will clearly have a moderating (dampening) influence.

EDN: Are you concerned at all about the credit crunch hindering merger-and-acquisition activity in 2009? What affects, if any, would that have on the semiconductor industry's ability to create innovative technologies and grow revenues?
Penn: Yes, I am, and I think that this fiasco has clearly discredited the past debt-driven business strategies endemic in the recent world. What makes semiconductors so attractive is the fact they are heavily cash flow positive. Companies with cash will find lots of rich pickings ahead; those that don’t will be struggling, indeed, constantly "counting the pencils" (aka cutting back on R&D and trimming product lines) in an effort to stay afloat.

EDN: What semiconductor markets (wireless, storage, etc) will take the biggest hits next year?
Penn: Unlike 2001 when fixed telecom bore the brunt, we do not see any one sector grossly out of kilter and thereby more at risk next year. Of three four big ones (PCs, mobile, and automotive), each should hold up reasonably well, the first two due to the impact of the emerging markets and the latter due to the unstoppable increasing levels of electronics penetration.

EDN: Are there any semiconductor markets that you see extra opportunity for in the next year, brought on by the sour economy?
Penn: Probably industrial – everything from lighting to motors, with power saving the underlying driver.

EDN: What about the outlook for 2010?
Penn: Given the slowdown and capital squeeze, capacity investment will be even more reduced through at least Q3 2009, which means most of 2010’s new capacity will be squeezed even further. Unit demand, in contrast, should start to pick up by second-half 2010, meaning a capacity undershoot is inevitable in 2010-2011. This will take 18 to 24 months to correct (and overshoot again) making 2011-2012 the next double-digit growth years. Far from 2009 being a re-run of 2001, the dynamics are looking more like a replay of the 1990’s recession.

EDN: Any further food for thought?
Penn: Conventional wisdom (and short-minded investors) says "cut back in a recession"; in the chip business this is the last thing to do. The orders are still there and the world will go on growing (despite the best efforts of the global financial community to inadvertently kill it!). History and classical economic and business theory says the opposite; now is the time to accelerate the business.

According to Harvard Business School, "it can be demonstrated that brands that increase advertising during a recession, when competitors are cutting back, can improve market share and return on investment at lower cost than during good economic times."

Similarly Gordon Moore postulated the same doctrine for R&D: “You never get well on the old products; you only get the increased revenue by moving on to the next generation of products. So it's very important you continue R&D investment across the bottom of the cycles. This is something that tends to be counterintuitive to people used to operating in other industries where you cut your cost, which means often cutting development, during the recessionary periods and build them up again during the others. In semiconductors you can't do that; you have to keep developing the new stuff all the time. In fact, you even need to accelerate the development of new stuff across these negative periods.”

We [Future Horizons] fully agree, hence the need to set profits aside in the good times for the inevitable market bust. The same doctrine holds true for starting up a new company, although investors tend to do the opposite, freely lending cash for umbrellas when the weather is fine, only to ask for them back the minute it starts to rain. Hard-to win customers and market share gains secured in the bad times last a lot longer than those in the free wheeling booms.