Power Semiconductors: Learning from History

Kevin Parmenter, Director, Applications Engineering. TSC, America



­When I first entered the semiconductor market it was dominated by big players who could afford to put up their own manufacturing and fabs. Some in leadership had an arrogant attitude that seemed to imply: “look at my great manufacturing processes that these customers are screwing up with their orders and requests.” Not surprisingly, these companies no longer exist.

Another misstep by some of the giant, vertically integrated Japanese conglomerates was spinning off their semiconductor businesses. They learned the hard way that your manufacturing (fabs) does not matter unless you are better than all other open market fabs, that owning architectures didn’t matter, and that sometimes value migrates from the semiconductor companies to the end customers. Thus, open standards, such as ARM cores – and using largest and most advanced contract fabs in the world – mattered more than their ability to control the ecology and they were wiped out. In other words, value-added is having a bunch of smart people using the best available global technologies and innovating. It turns out manufacturing outsourced gets better results than the highly trained professionals at traditional semiconductor companies did – whoops.

So, what does this mean for power semiconductors? We are amid a transition from silicon power semiconductors to the wide bandgap (WBG)-silicon carbide (SiC)-gallium nitride (GaN) revolution. As these technologies go mainstream, the conventional wisdom by some semiconductor companies is to take control and create customer dependencies. If they take the opportunity to transition as the microcontrollers and cores did, customers will get more value and everyone wins.

Today, traditional silicon semiconductor companies are pushing silicon farther than ever before. They are innovating in packaging, which is now as important as the die itself, regardless of the semiconductor material. Innovation is also taking place at the module level. A recent report from Yole indicates that the SiC market will be a multibillion-dollar prospect with incredible growth, capacity expansion and supply chain integration. This begs the question: do you have to own everything in the manufacturing process to be in the WBG business? With today’s constrained capacity, customers who never considered WBG due to perceived risk are now using these parts simply because they are available. Will they return to silicon when the great semiconductor famine of 2021-2(3) passes? Time will tell.

WBG companies are also innovating packaging – often with no packaging! Although silicon isn’t going away, the future of WBG might look like the fabless SOC, memory, and microprocessor market if contract manufacturing offers WBG manufacturing capacity with innovative, leading-edge processes. Companies with smart people and capital can design power semiconductors and have them fabricated wherever they like, and they can choose the best available technologies – without the boat anchor of “my fabs.”

What are the lessons for the power semiconductor company of the future? Opportunities abound if you make the right decisions. Punishment by the market awaits if you do not learn from history.