Mr. Richard: It's interesting. I think we are on a roller coaster. Why I say that is that if you go back and take a look at the most significant downturn in semiconductors when the dot-com bubble burst, in five quarters units went down 32%. That's pretty bad. In this downturn, units went down 36% in two quarters. And effectively, the only way that that is possible to happen is if you're just taking inventory out of the system, and clearly when we had the financial meltdown last fall, everybody in the supply chain purged inventory, and then what happened is everybody woke up in February and realized they had no components and started to expedite refilling that pipeline. There have been very strong Q2 reports and it looks like a V-shaped recovery. Intel (INTC) was up 12% sequential in Q2. The last time that happened was 1988. They shipped 850,000 microprocessors that quarter. They were one-tenth the size in revenue, they were taking share from mini and mainframe computers and they grew 66% year on year. Clearly, Intel is not that kind of growth company today. That was 21 years ago. It's far more mature, it doesn't move that fast. So again, I think we have one more quarter of this snap-back and we'll likely overshoot, simply because customers want more until they have too much and then they don't want any.
TWST: That's typical, isn't it?
Mr. Richard: It's very typical, but the velocity of the correction is the
highest I've ever seen. This is happening very fast, and if you look at end-
demand it is likely going to be down 5% to 10% this year. Semiconductor demand
fell 36% in units, 35% in revenues in two quarters. So clearly there was an
under-build. In the trajectory we are on, if we have the kind of Q3 that is
shaping up, the industry is going to be down 10% year over year in units. It is
likely that Q4 will be seasonally weaker than normal and actually could even be
down. The numbers after this earnings cycle are likely, for Q4, to be wrong and
too high. So that's the short term.
If you look longer term, the foundries are full. At the leading edge, the memory
guys, if they want to stay in the business, are going to have to reinvest in
technology, and I think as you get into the second half of next year you are
looking at what I would call a super cycle. I say that for two reasons. The
first is that there has been underinvestment in semiconductor capital equipment.
Effectively over the last 30 years in cap equipment, the average cap ex to
revenue has been 25%, and one could argue that there has been wasteful
investment and that ratio is too high, so 20% is probably the right number. It
basically would match the depreciation cycle of the equipment. So if you go and
you take a look, last year cap ex to revenue was 16%, which is a low number;
this year it's an even lower number, about 11%. So we're going to have to have
some catch-up spending.
At the same time, nobody is putting in any new factories, no new 300mm plants.
There is plenty of 200mm capacity but not 300mm, and for memory, 200mm is not
cost effective and can't be used. In the foundries, customers are migrating from
130nm, which can be done on 200mm wafers to 90nm, 65nm, and 45nm, and there is
not enough of that kind of capacity. The foundries will have to start to invest;
there are no empty shells, so either you are going to have to convert an 8-inch
plant, which could take a year and a half, two years, or build a new one, which
would take a year and a half. The industry is going to have a structural
shortage of capacity or ability to grow it rapidly. So that really constrains
the supply side.
When you look at the demand side for semiconductors, you've got five groups that
buy semiconductors, and I am going to bucket them in this way - governments,
corporations, emerging market consumers and developed economy consumers.
Stimulus packages are going to drive governments' spending. The second group is
corporate. Corporate has held back on investing in IT and PCs in particular, and
the average age of a PC is getting up to over three years, next year it will be
four. Nobody is adding in new PCs. You've got the Windows 7 transition in
October; consumers won't buy in front of that, corporations will wait a couple
of quarters before they move to Windows 7 and that will drive an upgrade cycle.
In the back half of next year it will just be cheaper to go buy new hardware
than upgrade the existing system or actually even to maintain the existing
system. So I see a big upgrade cycle on the enterprise side coming probably by
the back half of next year and corporations also have to spend on improving
their networking capability and will need to add storage.
On the networking side, you've got a couple of phenomena. The first one is that
the netbooks and global carriers are going to subsidize the sale of netbooks
much like cell phones. That is going to drive the next 500 million to a billion
users to the Internet in the emerging economies like India, South America and
China. It will be the first time these users will have access to broadband
through new 3G networks. Lower incomes will be able to buy a netbook at a
subsidized price and then pay a monthly fee for the bandwidth. It will be a
great deal for the carriers, it will be a good deal for the consumers and I
think that really accelerates the need to add to these 3G networks. It is a
killer app. You're going to broaden the penetration of IP and that's going to
drive the global carriers to spend.
The second driver of carrier spending is the use of video. Video on the Net is
increasing at a rapid rate, sites like Hulu and YouTube, but also people want to
watch Internet video on the TV, although the cable companies are resisting that
move. Cisco (CSCO) talks about TelePresence, it's another application; Facebook
talks about Video Chat. There is the move to IPTV where the phone companies
basically use an IP network, an Internet type of network to transport video into
the home, and that's point-to-point, not a broadcast, and all these things are
going to accelerate the need for bandwidth. So the wireline carriers are going
to have to add capacity to support that and video traffic. Video traffic is
going to go from roughly a third of the Internet's traffic to half in the next
few years. This is also going to drive carrier spending.
So now we have three big spenders spending and then the emerging economies -
back to the netbooks - I think that that's where the economies are most robust.
You're going to have the ability of people to get online consistently for the
first time and that's going to drive a lot of demand for netbooks. So they're
going to be spending on that end. As their standard of living improves, it will
drive demand for TVs and other types of consumer electronic.
In the developed economies, quite frankly, I think the consumers are bankrupt
and they're not going to be spending, but there are enough silicon drivers in
other areas. As semiconductor demand picks up, the industry is going to get
caught short on manufacturing capacity. There will also be a multi-year
accumulation of inventory in the channel to assure delivery and that could drive
25%, 30% growth, much stronger growth than we've seen in many years. It could be
multi-year while industry struggles to add capacity.
So it's a bit of a roller coaster. Right now, I think we're sort of at the top
of one of the bumps and Q4 and Q1 could be tough quarters, but by the second
half of the next year we should be coming into a very strong and multi-year up
cycle and hopefully the global economy will cooperate with that.
Tickers included in this excerpt: ALTR, ATHR, INTC, SLAB
For more information call (212) 952 7433. The Wall Street Transcript does not endorse any of the comments made by interviewees, and does not make stock recommendations.