Enabling

LOADING
Photo Title 1Photo Title 2Photo Title 3Photo Title 4Photo Title 5Photo Title 6
SSIA Summit 2016 - Predicting the Next Wave of Semiconductor Growth - From IoT, Deep Learning to Autonomous Driving

It's well known that Semiconductor industry is highly cyclical. The mean growth rates appear to be slowing down from >10% in the years 1960-2000 to ~3-4% in the last 15 years. So an apt question to ask is "WILL SEMICONDUCTOR INDUSTRY GROWTH SLOW LIKE OTHER MATURE INDUSTRIES?"

Ruchir's presentation argues that future semiconductor growth is still healthy. Some of the factors pointing to this are:

  • Cost per transistor is decreasing at the same rate as always (30% per year reduction)
  • Semiconductors continue to absorb existing applications. One example is solid state drive. The cost per bit of Solid State Memory is predicted to cross that of Rotating Media in 2018.
  • Manufactured area of silicon is growing at the same rate as always. Yet IC silicon revenue/area has been constant & stable. This extrapolates to growing IC revenue in the future.
  • Demand for semiconductors continues to grow. The low growth we are seeing is due to the effect of captive semiconductor revenue reporting.
  • New applications for semiconductors, like IoT, continue to emerge

Historical approaches to semiconductor revenue forecasting has been inaccurate, in some cases off by close to 100%! Ruchir believes that there is a reliable way to forecast future growth of semiconductor markets and it's based on the Gompertz Curve (or the S-Curve). One can apply the Gompertz curve in the life cycle of a product or industry. The Gompertz curve is applied in various domain as follows:

  • Traditional semiconductor markets like PC desktop, PC notebook, cellphone, smartphone and internet usage.
  • Semiconductor manufacturing: 200 and 300 mm wafer shipment, 193 nm stepper technology.
  • The Gompertz curve shows that total transistors and Silicon transistor are sWll in infancy stage!
  • Emerging IoT opportuniWes: IoT Wearables, IoT Fitness Trackers, IoT Medical Wearables
  • Automotive market – autonomous driving.

The big semiconductor growth in IoT are attributed to three major areas:

  • Data center. The main semiconductor content are big digital SoCs for servers, networking and storage.
    The insanity appetite for data center storage is driven by social media. YouTube users upload 400 hours of video every minute, ~ 1 petabyte of capacity daily. As the cost per bit of solid state memory falls below rotating media, future storage in data center will be installed with solid state drive.
  • Gateways. The growth of smart peripherals will also drive the corresponding growth in data collection hubs or gateways.
  • IoT nodes. More and more IoT nodes are emerging. Examples are smart meters, urban infrastructure and IoT edge devices. The semiconductor contents needed are: sensors, actuators, imagers and transmitters.

Ruchir concludes his presentation with interesting insight on how the semiconductor industry makes money.

 

jerome1

Article contributed by:

JEROME TJIA

Executive Committee Member, SSIA

Head of Development Centre, Infineon Technologies AP Pte Ltd