End User Report: ON Semiconductor

By Ed Sperling
System-Level Design sat down with Daniel McCranie, ON’s chairman, to discuss what drives the company, what it considers as critical to staying profitable, and future competitive threats. What follows are excerpts of that conversation.

SLD: Why has ON been making so many acquisitions? Is it a need to get big or to shift direction?
McCranie: You’re right that ON is extremely acquisitive. If you look back, once you started talking about $35 million for a stepper you could no longer put your own fabs up unless you’re a memory company or a processor company like Intel. Then, with rapidity, companies started to move to the foundry model. Even the big guys started to go ‘fab lite.’ That process took 21 years. Looking back you have to wonder why it didn’t happen sooner, but they had their own process people and their own development people. It’s the same in the IP industry. Big IDMs have their own IP people. But from Oct. 1 2008, to March 31, 2009, the semi industry dropped 45% in revenue. That really got everyone moving.

SLD: And they’re not hiring the people back, right?
McCranie: Not the same people. They’re hiring architects and software engineers.

SLD: What’s changed in terms of how ON is looking at the world?
McCranie: ON traditionally was a transistor and circuit company. They had transistors and low-end circuits. In the last six or seven years the company moved up to making small SoCs. After that, there were an increasing number of outside acquisitions. ON has what I consider derivative competencies. They are world-class in things like response to customer requests and overall product cost. What they’ve been doing over the past half-dozen years is looking at those pieces of silicon that are being poorly built, either because the company making them didn’t have good quality or good cost metrics. The results speak for themselves. It’s a combination of going upstream with more complex integrated circuits, buying IP outside, and taking their competencies and really improving the gross market of the product.

SLD: Is it about cost or time-to-market?
McCranie: ON is rigorous about its spending and cost model. But it is now moving rapidly to DDCT—Design, Develop and Cycle Time—which is reducing time to market. Over the past several years the company has moved the time from concept to qualified silicon down significantly.

SLD: Is there a move to advanced process nodes?
McCranie: No, and part of the cost is based on trailing processes. What’s changed is you see a lot more architects being hired and more design managers moving into solutions centers where you engage with a customer directly.

SLD: What becomes the competitive edge for a company like ON?
McCranie: There is always a potential for cost problems with Asian competitors. So far that hasn’t manifested itself because of the constant changes. ON has a high percentage of its business through the consumer channel, which has very, very short product lifecycles. ON has been able to reduce this design-development cycle time aggressively and crank out new products at reasonable margins. To date, the competition hasn’t been able to track that design-development time as fast. That’s always the problem. Once Asia gets to the point where they can conceive, design and develop these products in the same time frame, it could be a problem.

SLD: We’re starting to see quality problems in some high-profile end-product companies like Toyota. At what point does the pendulum swing toward quality vs. price?
McCranie: If your fundamental culture is quality and customer-centric—and that is ON’s strength; the company had to develop innovative new products and rapid development of new products—and you manage to keep that all together, then you shouldn’t have a quality problem. We look at quality metrics continuously. That could be a big differentiator vs. the onslaught of Asian competitors.

SLD: What’s the biggest challenge in getting chips out the door?
McCranie: We had to create aggressive IP re-use and get the mask right the first time—not the second or third time. You have to be mindful of the cost, but if you have robust IP and you have all these other processes in place you can get a chip out the door fast. It’s always profitable to build a lower-cost chip if you grind and grind and grind, but most guys are going for a good cost while getting the chip out faster.

SLD: How does not being at the bleeding edge of Moore’s Law affect IP re-use?
McCranie: ON does not chase Moore’s Law. As a result, ON’s IP requirements are not at the most advanced node. But the company’s IP requirements are non-trivial. You still have needs at 180nm and 130nm. There’s still a hard requirement.