The 20th annual IEEE Semiconductor Wafer Test Workshop (SWTW) started this evening. Rumor has it that attendance is over 240 this year which is a vast improvement over last year’s 160 or so attendees. At the peak the conference had almost hit 600. Things started off well with a reception where I had the chance to catch up with many industry friends and colleagues.
After dinner, Jerry Broz the General Chair kicked things off with the “Probe Year in Review”. In summary:
- He reviewed key Semiconductor Industry Association (SIA) statistics for 2009 showing how bad the year was. Only three of the top twenty semiconductor companies grew last year – Samsung, Qualcomm & MediaTek. And in 09Q1, fab utilization was less than 60%. More than 50% of the industry capacity is now on 300 mm wafers.
- The bad year for the semiconductor companies translated in to a bad year for probe card industry as seen in the VLSI Research 2009 probe card market numbers released last month. The total probe card revenue was down 29% for 2009. One of the few bright spots was MicroProbe.
- Lastly, he paid tribute to three key probe industry members who passed away in the last year:
- Frank Pietzschmann – Qimonda
- Brett Crump – Micron
- Bill Mann – Conexant and the founder of SWTW
Having had the pleasure to work with each of them for many years, I will miss them. I also know others will certainly miss them not to mention their great industry contributions.
The keynote “What’s Going to Rock Your World – Or At Least Push Your Probes” was presented by Dean W. Freeman, Research Vice President of Gartner Technology, Inc.
They see three hot technologies that will drive semiconductor demand: Cloud Computing, Green Tech/IT, and Social Computing.
Cloud Computing is a return to the old days of mainframes / minicomputers (think Digital Equipment’s VAX systems) where the dumb terminal accessed the centralized computing resource. Therefore, the computers especially the laptops and netbooks we use don’t have to be so powerful. However, we want robust high speed connectivity everywhere on earth. This will push the networking technology as this infrastructure is developed and we have a far way to go. In addition, security will become more critical at the same time driving new devices and technologies.
In Green Tech/IT, renewable energy is still a big driver with the typical applications in solar, wind and energy harvesting (via MEMS). However, conservation will continue to push LED production with greater than 30% CAGR. And there will be more focus on AC to DC conversion along with power transmission since currently only 13% of the power generated actually makes it to the light bulb in your house.
With regards to Social Computing, Dean took a poll and most everyone in the audience was familiar with social media since they use LinkedIn. A fair number of attendees were familiar with Facebook and use it. However, only two of us Twitter and I suspect I am the only who will be blogging about the conference…
Social Computing will drive the adoption of mobile devices and network infrastructure far more than Cloud Computing in Gartner’s view. Younger generations who have grown up using this technology are fully “connected” and this will continue to accelerate the growth of these areas. One issue of concern is that the younger population has grown up learning collaboration from a young age and this tendency to simply contribute and take may strain traditional models of Intellectual Property (IP).
Dean has many additional different sets of statistics and charts that he reviewed quickly. Other highlights include:
- Moore’s law requires three things – deep pockets, building lots of parts (high volumes & hight utilization) and willingness to invest. They are predicting an 8 nm fab will cost $6B however the payback will be $0.13 per transistor. In the end there may be only four or five foundries – Intel, Samsung, maybe Toshiba plus one or two foundries from among TSMC, UMC, Globalfoundries and others.
- The transition to 450 mm wafers is making progress since they see “constructive dialog” between suppliers and potential customers. I.e. the suppliers have moved from “hell no, we won’t build it” to “maybe”. They see prototype fabs in 2016-2017 with production fabs in 2018-2020.
- Through Silicon Vias (TSV) technology is now finally a reality. Elpida is now shipping a part with a stack of 8 DRAM dies on 300 mm wafers at a rate of 10K wafers per month (wpm).
- In terms of overall semiconductor market they see 27% growth for 2010 and the market may finally cross the $300B threshold in 2011. 55% of the growth is due to very high average selling prices (ASP) for DRAM which should see a correction in 2011.
- Semiconductor sales are ahead of “box” (i.e. end device such as computers, laptop, etc.) of sales. Since the current technology of laptops and netbooks is lasting longer, companies are actually building inventory earlier (i.e. in Q1, Q2) and shipping via boat instead of the previous practice of building later (Q2, Q3) and shipping via air for the holiday shopping period.
- Dean provided a sneak preview of CAPEX (capital expenditure) spending of the semiconductor companies. This data is not to be released until Wednesday, so suffice it to say that Samsung had the largest increase as it continues to pick up aggressively increase market share. Foundries (TSMC, Globalfoundries, etc.) are spending heavily in a race to obtain new customers. And Intel spent the least since they are able to reuse existing equipment.