mike on pikes peak


Mike Clayton (pictured), a true semiconductor industry veteran, is a valued advisor down through the years to MFG Vision. I interviewed Mike recently.  Like me, I think you the reader will find some valuable insights in his answers.

Q. You celebrated your 80th birthday a couple of weeks ago. First of all congratulations! How and when did you start working in the semiconductor industry?

A.  I was studying Metallurgical Engineering at Carnegie Tech and looking at careers in mining, steel or aluminum production, or metal working (nails, screw drivers, you name it) when luckily the semiconductor industry in that area started to grow.  The IC was invented just in time to save me from a job in what then became a dying industry in Pittsburgh (its much cleaner now.)

Westinghouse had an IC pilot line in Youngwood, PA and built a new IC plant in Maryland where I worked after graduation.  But they had originally hired me in their Youngwood Power Products plant during last two years of college, as a technician, so I could finish up in night school.  That hooked me on high tech manufacturing.

In those days this industry made all its own equipment, and I learned to weld quartz, run 48 hour 1200 C diffusions for high voltage products, then transfer to the new IC pilot line.

So I saw the two extremes of the industry, 200 ampere 1200 volt Thyristors, and the first tiny chips for the Apollo Project and early calculators at the new IC plant.

Q. How do you see the industry in terms of manufacturing today and how do you see it five years out?

A.  The semiconductor industry stopped making its own PolySilicon from Liquid TriCloroSilane, and its own Crystals and Wafers, and most of its own equipment slowly over my first 20 years, and gave up mask making a bit later. Then many of the design teams left the big companies and started up the Fabless industry after Morris Chang created the first really high quality IC chip foundry in Taiwan. SemiWiki.com has much of the industry history on its web site, and they see the future as Fabless with Design Software Companies (Mentor, Cadence et al) and Core IP sellers (like ARM) supplying the tools for hundreds of IC product design companies that have their chips made in foundries.

I see mainland China slowly taking over what is now a Korea-Taiwan-Singapore (mostly) foundry industry with new giant (Global Foundries) coming out of the old AMD Dresden fab merger with Singapore Charter, and with funding from the Prince of Abu Dhabbi, building huge fab in Malta, NY, USA.

So foundries are here to stay, perhaps even in US.  But mainland China will most likely BUY its way into the existing foundry companies by Joint Ventures that start out with lower technology fabs in mainland while they try to keep the bleeding edge products out of there for a while.  The next decade will be interesting.  Qualcomm was first and largest fabless company and is now talking about a JV in China. TSMC is also talking about a JV foundry in China.  We will see how fast the Dragon eats the Tigers…or not.

Q. How will the emergence of big data analytics change the way manufacturing is managed in the next few years?

A.  The job of “data scientist” is now one of the jobs most desired, and it is all due to the semiconductor chip capabilities that led to massive storage capacity, very high speed computing, and global standardized communications based on the capabilities of those chips.   The chip making industry actually was one of the first “big data” users starting with the IBM mainframes in offices and DEC VMS operating systems in factories, running early hierarchical MES (manufacturing execution systems), and then developing internally the SAS-based Yield Management Systems which evolved during the 1990’s into commercial YMS systems.  So it is now possible for the largest to the smallest part of the semiconductor supply chain to purchase well supported systems that meet their particular needs.   I work now days from home, and can link by VPN into any client’s MES or YMS system, to analyze test data vs factory or lab experimental runs for new products, or check yield busts on existing products.  The designers can view foundry results via their early device tests and classify wafers for full circuit testing either at their own site or offshore wafer test, assembly, and final test business without owning any manufacturing systems at all.

Big data analytics can link into the financial and supply chain management systems, and CRM systems through better inter-connectivity standards, so that big and small systems makers as well as big and small components makers can get the benefits of such things a Hybrid Cloud technology to grow their core business without a giant IT department.

IC yield management systems which started on terminals linked to mainframes and moved to PC’s linked to giant RDMS’s, can now can access even more data from web browsers that require almost no IT support for either the client or server side, unless you want that responsibility.  Flexiblility to focus on core strengths.

(Part 2 will be published next week)