SAN JOSE -- Set against the backdrop of strong recovery for the semiconductor industry, Gartner analysts gathered here Thursday to present their predictions for this year through 2014.

Bryan Lewis, research VP for semiconductor research, started the morning with a whammy of a question that tends to make most industry execs cringe: Although things looking pretty good currently, is there a correction coming?

Based on a newly created index, Gartner believes the semiconductor industry has run beyond true system growth, which was expected, but it leaves the question open as to where true demand lies, and what happens after that. "If we don't see any slowdown in the back half of the year, what happens? You end up with a pretty big gap between what systems sales growth is and what chip sales growth is," Lewis pointed out. "Our current projections are that system growth right now is running at about 12 to 15% revenue growth on an annual basis and chips, our current forecast is 27% but our upside scenario at 33%. So you ask yourself have we gone over true demand at this point? We've done our inventory replenishment; we've gone to true demand and with capacity being as tight as it is, is there some double ordering going on out there?"

Laying the groundwork for its forecast, Gartner made some key assumptions. First, PC production unit growth in 2010 is now expected to be 22%, up from 20% in its Q1 update, while processor ASPs (average selling prices) are firming and revenue now expected to grow 15.5%, up from 10% in Gartner's Q1 update. Second, rising DRAM prices coupled with strong PC demand is leading 2010 DRAM revenue growth to 78%, up from 55% in Gartner's Q1 update. Third, cell phone production unit growth in 2010 is now expected to be 14%, up from 12% in the Q1 update, while application-specific cell phone chip revenue is expected to grow only 11% due to major ASP pressure from companies like MediaTek. Also, second half 2010 growth is expected to be below seasonal norms as semiconductor sales align with electronic system sales.

Specifically, in Q1 Gartner forecasted revenue for 2010 would grow 19.9% to $276 billion. Now, Gartner predicts 2010 revenue to reach $290 billion, up 27.1% over 2009. For 2011, the company predicts the semiconductor market will grow to $307 billion, up 5.7% from 2010, and in 2012, Gartner expects revenue of $310 billion, just 1% higher than 2011. Further out in 2013 and 2014, Gartner predicts the semiconductor market will reach $325 billion and $347 billion, up 5.1% and 6.6%, respectively.

Lewis noted the $14 billion difference between Gartner's Q1 and Q2 forecasts. "Where did it come from? In the past when things have gotten better it's primarily been due to PCs and cell phones. PCs and cell phones of that $14 billion was only about 25% this time around. There was automotive and LCD that added more to the forecast but nothing too dramatic there. The rest of the applications were a huge chunk of it, meaning that all application areas have improved and in all regions with the exception of Europe most recently. The general story is that as the tide came up, all applications and all devices went up with it, with PCs and cell phones the big drivers. Still, as the numbers come up in 2010, you have to ask yourself if system demand is running lower at some point, we need a correction and does that run into 2011?"

"The short answer is yes," he continued, "we did lose a bit of ground in 2011 and 2012. You can see we expect 1% growth, and this has to do with overcapacity in the industry on a production level and especially in DRAMs."


ASPs strong ... for now

A key trend is ASPs and the system unit growth numbers. "The big downturn in 2009 was a unit problem. System growth dropped off hugely and there was some ASP pressure, but it was minor in comparison to past years. It was really a system problem that drove everything down as we had the financial crisis, everyone lowered their IT budgets and stopped spending as much so units went down primarily. Now we are looking at with capacity being so tight that you see the unit growth is still running pretty strong at a good 15% clip but then ASPs increased too because capacity is so tight and you can get more for your product out there because it is supply and demand oriented. But going into the outer years, we start to see how that rolls off. In general ASPs firmed up across the board ... but we will swing back to overcapacity in 2012, especially on memory. With all the capacity that is coming online, we could see some overrun on the capacity side," Lewis explained.


2014: Market drivers range from smart phones to solid-state drives and blue laser DVDs

Looking forward to 2014, Gartner expects the biggest market drivers to carry more than 15% compound annual growth rates and more than $8 billion in revenue. Key applications include smart phones, PCs of all kinds, LCD TVs, entry-level servers, mini-notebooks and media tablets, solid-state drives, IP set top boxes, blue laser DVDs, automotive applications such as infotainment and powertrain modules for hybrids and electric cars, as well as industrial applications including LED lighting.

Overall, Lewis said chip growth is running way too hot relative to system growth and as a result there will be a modest chip correction in the back-half of the year and into 2011 as semiconductor growth re-aligns with system growth.

 
Capital equipment spending strong, but won't beat 2007 high

Next up on the Gartner agenda was Dean Freeman, research VP for semiconductor manufacturing, who noted that similar to the buildup in the semiconductor space, the equipment side of the industry has experienced the same dynamic. He predicts worldwide semiconductor capital equipment spending will surpass $35.4 billion this year, a 113.2% increase from 2009 spending of $16.6 billion, but now the question is if the recovery overshoot will send the industry into oversupply in 2011. Freeman asserted that the current cycle will peak in 2012 but will not surpass the last cycle in spending, the high reached in 2007.

"Historically, every cycle from an equipment perspective has been stronger than the one before but we are seeing things like more equipment being recycled, some companies reducing their capital intensity. As a result, we think that while it is a very strong growth period, we're not going to see the phenomenal growth than we've seen in the past," he said.

Freeman pointed out that the bulk of the spending is due to the move down the technology roadmap: DRAM is shifting from 55 to 45 nm, while NAND flash is being driven down to 22 nm. The top five capital spenders are Samsung, TSMC, Intel, Global Foundries, and Hynix Semiconductor.

He recommended that semiconductor equipment manufactures should prepare for a period of slower growth in the second half of the year as the semiconductor industry shift from technology to capacity buys in the second half and into 2011. Equipment companies need to be prepared for this shift, as well as for the capex peak to occur in 2011 if memory companies further accelerate capital spending to keep up with Samsung.


Big semiconductor companies learning from start-ups

Finishing up the day, Jim Tully, VP and distinguished analyst for semiconductor research, discussed how larger semiconductor players should work with start-ups, and observed that they could learn how to operate more efficiently, how to motivate their R&D engineers and how to do things faster. This sounds wonderful, but it never happens, he said, because of the bureaucracy inside large companies.

Because that won't happen, he asserted that instead big companies should get start-ups to do R&D for them. And while this may sound like a strange idea, Tully said it is happening. In the last two calendar years, total investment by large semiconductor companies was $756 million, which is an average of $14.3 million per company. This indicates that sources of funding are shifting from VCs to vendors, with one quarter of investments now from vendors.

This is currently concentrated from a handful of large vendors - Intel, Samsung, Qualcomm, Renesas, Toshiba, and ST - but he recommends that other large vendors should explore investments.

And, because of the benefits, Tully expects to see more of this activity going forward. With fewer start-ups to invest in, he believes the industry is reaching a crisis and that by 2015, there will be a "land grab" for start-ups by the big players. As such, big companies should stake their claims now, establish relationships with start-ups by direct investment with a seat on the board but also with IP licensing/cross-licensing deals, as well as becoming customers of those small companies.