USA & BANGALORE, INDIA: While a majority of analysts at a recent panel discussion on global semiconductor outlook predicted semiconductor growth in the range of 6-11 percent during 2008, some other panelists predicted 2008 to be flat year or a year of negative growth.
There were fears of a possible recession in 2008, along with concerns surrounding consumer spend that could be hit by higher oil prices and the US mortgage crisis.
This panel discussion was organized last week by Semiconductor International, USA. This is the full report.
Semi forecasts mixed for 2008
Amid concerns of a possible recession in the US economy in 2008, analysts at a recent Webcast hosted by Semiconductor International, were divided in their forecasts for the coming year. A majority predicted semiconductor growth to be in the range of 6-11 percent during 2008, while some others predicted 2008 to be flat year or a year of negative growth.
Anne Craib, director of Market Research, International Affairs and Finance, Semiconductor Industry Association (SIA), said the global economic situation needed to be factored in, as well as its impact on consumer demand.
She said: "Semiconductor demand is driven over 50 percent by consumer demand currently. That is something we should increasingly be aware of. Areas like gas prices and the home mortgage market are things that we previously would not have paid much attention to that we have had to take into account in our forecasting this cycle.” She was confident of the semiconductor industry reaching 7-8 percent CAGR during 2008.
Steve Szirom, President, InsideChips.com, added that many economists were predicting recession in 2008. He said: "The demand-supply balance should be somehwat better than this year. We may have a demand driven recession." He adopted a pessimistic view for 2008, predicting -8 percent growth.
DRAM weak, NAND bright
Gary Grandbois, principal analyst for iSuppli Corp., noted: “We have reduced our forecast to 7.5 percent for 2008 and think it might go lower than that. We think it’s going to be a negative first half. Certainly in the DRAM area, it’s looking very poor. We think it will improve in the second half, almost mirroring 2007, but giving us a far weaker year in 2008 than we’ve expected.”
Richard Gordon, Managing Vice President, Semiconductors, Gartner Dataquest Research, agreed with Grandbois, adding that DRAM would see a negative side in 2008. "Our forecast is 6 percent for 2008, and it doesn't factor in the US recession," he said.
While the DRAM market has been predicted to be negative next year, analysts see a positive market for NAND in 2008. New applications, such as WUSB (wireless USB), increase in cell phones, higher content in portable media players, etc., are likely to drive growth.
Commenting further on the outlook for 2008, Moshe Handelsman, President, Advanced Forecasting Inc., noted that 2007 would be the peak of current IC cycle. "From that point on, the underlying demand for semiconductors will decline and decline in 2008. We are negative about 2008," he added.
Carl Johnson, Executive Director, Research Infrastructure, concurred that the industry had become much more global. "We now have to look at the mortgage debacle, etc. Consumers will be very tight in first half of this year." He added, "I would say, next year's going to be flat."
Mike Cowan, an independent semiconductor industry analyst, said the growth would be about 8.15 percent during 2008. "The dynamics of the market and the industry will change month-to-month as well," he quipped.
Capex likely to dip in 2008
Regarding capex in 2008, Carl Johnson of Research Infrastructure, expects the next year to be bumpy as far as capital spending is concerned. "We're in a downturn right now. Foundries, who are investing lot more money in older process generations, and that is a function of some of the other older IDMs and fabs, are actually shutting down and saying, 'we can go over to the foundries and process wafers for less than what we can do it on our own'. We are seeing lot of consolidation within the fab space. Mid-level players are consolidating. The customer base is clearly narrowing."
The cost of designing some of these leading-edge devices, and getting them to market, and then following it up with another product, if you don't want to be a one-product guy, is a real challenge. That is limiting the number of players that are going into the mega fabs. So, the field is narrowing in 65nm, and 45nm, and as we get to below 45nm, the field is going to get much, much narrower.
According to him, capital spending is likely to be down in 2008. "I am predicting 10 percent down next year. There's also going to be a great consolidation in the devices manufacturing community, and also in the capital equipment community. We are seeing a number of M&A activities in the capital equipment business. It will also go into the supply chain business."
Gartner's Richard Gordon said the research firm was forecasting capex to be down by -15 percent in capex in 2008, and that includes -30 percent in the DRAM sector. He added: "Looking at the individual companies in the DRAM space, I won't be surprised to see that go even lower. So, -15 percent in capex can get even worse as 2008 unfolds. We will see it coming back. But, it will take a while for demand to catch up with supply."
EDA industry in catch-up mode
The EDA industry is said to be lagging behind the semiconductor industry at the moment, and is in the catch-up mode, according to Gary Smith, President, Gary Smith EDA.
Commenting on the outlook for the EDA market, Smith said the EDA industry is in a lttile unusual position. He said: "The market's been flat for the past four years. Tools for 65nm, 45nm silicon design have also been delayed." The R&D was not put in because of the recession. "Right now, we are in a position of lag in the market," he added.
EDA tools cover two process generations. The industry is just starting to introduce 65nm and 45nm tools. That generation is being called the DFM generation tool. Smith said: "It is even more important to the semiconductor industry as we run into manufacturing problems that they are relying on design tools to solve, rather than on semiconductor equipment." That's a major shift in the market!
EDA to grow 7.8 percent in 2008
According to him, the industry is now now into a pretty good growth area. "We were 11 percent last year, 10.2 percent to come in this year. We will be a bit down next year at 7.8 percent," he forecast. This has been attributed mainly to the EDA industry's lag in the market. "Some are moving to 32nm. And certainly, a lot of work is being done in 45nm," he added.
Smith noted: "The EDA industry is in the catch-up mode. We will lag them. We are expecting the downturn to really hit us in 2009. However, we are not an industry that goes negative often. No matter what you guys do, you still have to design something. So, when you go into recession, typically, the way you get out of recession is you generally design your way out!"
DFM, ESL growth drivers
Among the growth drivers is the DFM (design for manufacturing) issue, which is increasingly getting more complex. There is said to be a move to restrict the design rules that is in place now for 45nm. "We are going to see major changes in 32nm; that'll have impact on tools," he added.
The other issue is parallel computing that has become a major task for the EDA industry. "With signal threading, we can no longer handle designs over 100 million gates. Of course, at 45nm, you can do a 100mn gates. That rewriting process is another issue that is also slowing out down. That's a full three-year re-write," Smith said.
Further, EDA is also starting to move up into the ESL. The electronic-system level (ESL) is going to shift the EDA market more into the systems market, and serve less on its dependency on the semiconductor world.
New fabs in India, China
There have been a lot of announcements made regarding new fabs, especially in places such as India, China and Brazil.
Gary Grandbois at iSuppli said: "Brazil is a better example. India just announced that they are building new fabs. What we saw at the turn of the century is that the industry split into two areas -- one traditional components manufacturer and second is the SoC manufacturer. Those are the companies that need leading edge fabs."
According to him, the cost of R&D was going out of sight for process development. "We're also seeing consolidation of research groups. We expect that come down to five consortiums or less. All companies can afford do their own process development once the basic process has been developed."
There are going to be different types of fabs. With globalization, lot of countries may decide they want to have a fab. Brazil announced one. "You're going to see them all over the world. The market's going to change," he added.
Anne Craib from the Semiconductor Industry Association said: "If you look at the cost structure, it costs over $1bn to build and operate a fab in the US. The question is where is the fab going to be located? The US companies will continue to be major players. Again, the question is: where is it going to be economically feasible? The interest is outside of the US."
New elements likely in 32nm
On the subject of integration of MEMS, 3D, etc., Carl Johnson from Research Infrastructure said, "A very large topic with the design community is big change in computer architecture -- the big change is the multicore -- that's the biggest driver now."
Jim Feldhan, President, Semico Research Corp., noted that the industry is going to hit limits with silicon processing at some point of time. "We have to bring in new elements. In 32nm, there'll be only a handful of companies who can push real hard there and can afford it."
Push to 450mm fabs unlikely?
Finally, is there be going to be a push to 450mm fabs and how's the impact going to be like?
iSuppli's Gary Grandbois, said that curently the industry was expecting that it would need to abandon silicon in 2020. "If we don't have any silver bullets by then, we are going to use nano stuff to augment CMOS. We don't know what that's going to be."
He added: The issue is: are we going to use silicon at all in 2020? If we start developing equipment for 450mm, we are not going to have that very soon. What they have to consider, what is the payback for that move?"
We would love to hear from you on how you see the semiconductor industry going forward in 2008.