Some good info from SEMICON West

SEMICON West in San Francisco was its usual busy place this July. There were 711 exhibiting companies there to show you everything to do with semiconductor design and manufacturing. And, 309 of those companies had a new product they wanted to introduce you to. You can still garner SEMICON West information at www.semiconwest.org/.

There were also dozens of interesting talks (more than 30 hours worth) at SEMICON. One of the more interesting ones I was able to hear was by Bob Johnson of Gartner (www.gartner.com), talking about the outlook for semiconductors and semi equipment.

Gartner thinks worldwide semiconductor revenue is on pace to reach $336 billion in 2014, a 6.7% increase from 2013, and they expect a 4.9% increase in 2015. One of the things Bob pointed out is that while both DRAM and NAND memory are expected to have good growth in both 2014 and 2015, Gartner expects DRAM to decline some from its 2114 $43 B in sales starting in 2016, while NAND will continue to grow and will out sell DRAM in 2016. This growth is driven mainly by SSDs. The market for NAND is expected to approach $48 B by 2018, up from about $31 B in 2014, with more than half this coming from SSDs.

Courtesy: Gartner

From the above chart Bob presented, you can see the semi revenue from Traditional non-smart cell phones is almost gone in 2018, and that’s with there still being around 200 k of those devices being sold. The premium and basic smart phones will have taken over by then. But, they think the revenue from those smart phones will be up only a tad bit in 2015, and then stay pretty flat through 2018 – as the market is pretty well saturated. But revenue is not thought to be declining for this area. Unit volume for the combination of premium and basic smart phones in 2018 it expected to be close to 1.5 billion.

Bob also said he expects the traditional two year per node pace of Moore’s Law will continue to slow down and that only a few high volume/high performance applications will be able to justify the costs of 20 nm and beyond. He expects 28 nm will be a long lived node.