EE Times: One of Exar's markets is analog, specifically the big growth area of power management. How will you compete with much larger companies in that space that have their own fabs to optimize products?
DiNardo: The idea that real men need fabs is a 20-year-old story, and we are not going to fight yesterday's war.
An esoteric process here or there is certainly valuable in some areas, but the analog world has changed. Today you have better design tools, models from foundry services are much improved, and processes from many foundries are as good in aspects necessary to deliver high-performance products as anything Maxim, TI, and ADI have internally.
EE Times: What foundries do you use?
DiNardo: We had our own fab from a 2007 Exar merger, but we closed it and moved the equipment to China, where we are collocated with [a small fab called] Silan that uses the equipment to build some products for us.
We use Globalfoundries and TSMC for digital chips and three analog foundries -- Dongbu in Korea, Plessy in the UK, and National Japan Radio Corp. in Japan. You have to find the right recipe in analog, and you get spread a little wide, but you can find what you need.
EE Times: Do you have any insights from your work in China about why the country has not lived up to its hopes to be a big chip foundry?
DiNardo: Quality and IP are probably the two things that got in way. The aura of IP issues stymied what could have been hockey stick success for building chips in Mainland China. I haven't run into any IP issues, but that cloud was probably their biggest stumbling block. There are cycles of learning in quality, too -- there are quality foundry services in China now, but they had to go through cycles of learning.
I just came back from 10 days in China, in Wuxi and Hangzhou, where we have a design center with 35 people. It's a pretty vibrant place. People are talking about a hard landing with GDP growth at 7 percent, but we'd kill for that in the US. There are some cities still clocking 7-10 percent GDP growth and some industries still on fire there.
EE Times: Tell me about your market for connectivity products in what you call the Industrial Internet.
DiNardo: Its everywhere from washing machines to tractors to sensors for shipping containers. I've heard there are as many as 3,000 sensors in a tractor.
I looked at investing in a company installing sensors on refrigerated shipping containers handling produce. People lose 15 percent to spoilage when they ship blueberries, for example, from Chile to the dock in San Francisco. Sensors in the containers measure the carbon dioxide generated when vegetables and fruits decay. They trigger ozone generators and pumps that help stop the decay, reducing spoilage to maybe 3 percent.
The amount of electronics content [in industrial systems] is massive. It's a perfect place for a mixed-signal company, because it's about acquiring real-world data, then amplifying, filtering, and digitizing it to get it on the Net.
We are avoiding consumer electronics. We have little desire to be in next iPhone. For a company our size, that's a whip saw, and you can fall out of bed pretty quickly. So we chose very sticky markets. In industrial, you get designed in and stay designed in for three to five years or longer, and you get paid on value, not just the cost of the silicon you provide.