Latest IC Forecast: Big Demand, Shortages
Semiconductor Engineering
April 15, 2021
Over the last year, the semiconductor industry has seen its share of highs, lows and uncertainties.
In early 2020, the business looked bright, but the IC market dropped amid the Covid-19 pandemic outbreak. Throughout 2020 different countries implemented a number of measures to mitigate the outbreak, such as stay-at-home orders and business closures. Economic turmoil and job losses soon followed.
But by mid-2020, the IC market bounced back, as the stay-at-home economy drove demand for computers, tablets, and TVs. That momentum has carried over into the first part of 2021.
To get some insights for what’s ahead, Semiconductor Engineering talked to Bill McClean, president of IC Insights.
SE: Recently, you raised your 2021 semiconductor forecast. What was it before and what is it now? Why the change?
McClean: Semi market forecast for this year was +12% and it is currently +19%. We also believe that there is much more upside than downside potential to this forecast. DRAM is in short supply and prices are rising steadily in this segment. We started the year at +18% for DRAM and are currently at +28%. Overall, the market forecasts were raised for just about every segment–analog, logic, MCUs, etc. There is increasing strength across the board.
SE: Any thoughts on the chip shortages? When will they end?
McClean: The chip shortages will likely continue throughout this year. Capacity is just not that easily added and takes time. In the automotive segment, there is tremendous demand for devices made on 200mm wafers that do not use leading-edge feature sizes. Most manufacturers are hesitant to add new 200mm capacity, if they can even procure the equipment. TSMC has been attempting to move some of its automotive customers from 200mm to 300mm wafers where they have more capacity available. However, this is not a quick process as the device characteristics can change when moving from 200mm to 300mm wafers as the 300mm wafers are thicker. In the long run this will be a solution, but it is not a quick fix for the current problem. It should also be noted that there is likely a significant amount of double ordering going on right now as customers are panic buying. Moreover, if they are fortunate, they will build inventory to hedge against future demand.
SE: What is your CapEx forecast for 2021 versus 2020? Any thoughts here?
McClean: Assuming a 60% increase in spending for TSMC this year to $27.5 billion, a 37% increase for Intel to $19.5 billion and flat spending for Samsung at $28.0 billion this year, we are forecasting a total increase of 18% this year to $133.4 billion. To the best of our knowledge, Samsung has not publicly released its semiconductor capital spending plans for 2021. IC Insights believes that the company’s outlays this year could be +/-$5.0 billion from our current 2021 estimate of $28.0 billion, a spending range that could by itself swing the total semiconductor industry capital spending growth rate change from +14% to +22% this year.
SE: What are your forecasts for DRAM and NAND in 2021?
McClean: As mentioned, our forecast is for 28% growth for DRAM with a 26% increase in bit volume this year. We are forecasting a 17% increase in the NAND market this year to $64.4 billion with a 37% increase in bit volume. NAND is currently in oversupply and is experiencing pricing weakness that should firm up in the second half of this year. DRAM is expected to show price increases on a per bit basis throughout this year.
SE: What is your foundry forecast? Any issues here?
McClean: Our foundry forecast is for 21% growth this year with much more upside than downside potential. The foundries are pushing forward with price increases, so that even with limited capacity they will be able to significantly increase their growth rates. As noted earlier, there is a significant amount of double ordering and inventory building going on right now in the industry. I always remember the foundries telling me that the worst offenders for double ordering and inventory building were the fabless companies, which represent about 85% of the foundries’ customers.
SE: The U.S. government, chipmakers, and OEMs want to have more fabs and fab capacity in the United States. Any thoughts here?
McClean: This is really a long-term project that will not have any significant impact for at least 4-5 years down the road (better late than never I guess). TSMC’s new fab in Arizona is not expected to come on-line until 2024. Like China, the U.S. government is recognizing the importance of the semiconductor industry about 10-15 years too late. With the combined spending of Samsung and TSMC exceeding $50 billion every year for at least the next three years, it will be extremely difficult for any company to catch up these two in leading-edge logic process technology. As a side note, I think this is the first time in my 41-year career in the semiconductor industry that people are waking up to the fact that the electronic systems industry just doesn’t exist without semiconductors. Those of us in this industry for any length of time always knew this, but the governments and general public are just now realizing this fact.