The Outlook for Semiconductors...Especially in Cars

At the recent SEMICON West event, there was a media luncheon, which sounds rather grander than box lunches with sandwiches. Ajit Manoja, SEMI's CEO and president, kicked off the proceeding with a preview of what was coming in yesterday's press release about the equipment and materials market outlook. Of course, like EDA, equipment and materials are driven by the semiconductor business itself, which, in turn, is driven by end markets like mobile, automotive, and hyperscale data centers. I forget who came up with the lovely metaphor that EDA is the water-skier behind the speedboat of the semiconductor industry. When the boat is running fast, EDA is buoyant, when it slows, we sink. Equipment and materials are in the same boat (to mix my metaphors).

In fact, the semiconductor industry is now so large that it is not just driven by technological change like 5G, but also by how strong the world economy is. So Ajit started looking at real GDP growth (or decline, in the case of 2020). The growth is forecast to be nearly 6% this year globally, and nearly 5% next year, 2022.

Turning to semiconductor, Ajit's forecast is:

• Semiconductor sales are expected to grow 20+% to $550B in 2021 (after growing 10.8% to $464B in 2020)
• Shortages are spreading across the board, mostly in legacy nodes, starting with automotive
• One particular concern is chips for semiconductor manufacturing equipment which is needed to expand capacity and reduce shortages
• Forecasts for 2021 total wafer fab equipment at $80B-$90B growth of 23-38%
• Forecasts for 2021 material are expected to hit a new record high of about $62B, growth of 11%

There are a lot of fabs being constructed or expanded around the world, in both 200mm and 300mm:

• 200mm 25 fabs/expansions expected (5 in Americas, 1 in Europe, 14 in China, 3 in Japan, 2 in Taiwan)
• Total 200mm capacity (this is all legacy nodes) is forecast to grow 18% between 2020 and 2024
• 300mm 60 fabs/expansions expected (6 in Americas, 10 in Europe, 15 in China, 5 in Japan, 8 in Korea, 1 in Singapore, 15 in Taiwan)
• Total 300mm capacity (memory and leading-edge) is expected to grow 48% from 2020 to 2024

Take note of that last percentage: 300mm capacity is going to get almost half-as-large-again as it is today (or rather, as it was last year where we have actual numbers).

Semiconductor market forecasts for 2022 are all over the place, as usual. Forecasts range from UBS at 15.8% to Omdia at just 4.2%. The average is nearly 10%.

The semiconductor equipment market is expected to grow at 8-12% (following over 40% growth in 2021) taking it over $100B for the first time. The total materials market is forecast to grow about 11% in 2021 setting a new high of $62B, and continue with about 7% growth in 2021 to get over $65B.

Summary

Ajit's summary was:

• 2022 electronics end-use markets expected to remain strong driven by emerging applications 
• Semiconductor, equipment, and materials markets are expected to show moderate growth in 2022 following significant growth and record number in 2021
• Potential headwinds for the industry:
  • ongoing trade/export restrictions and geopolitical/economic issues
  • Covid-19 pandemic control and pace of rollout of vaccines still a concern
  • Industry talent shortage becoming more pronounced with current push to expand capacity

Automotive

Later, Bernard Swiecki, who is director of the Center for Automotive Research (CAR, cute huh!) in Michigan, gave a brief summary of what is going on in the automotive market.

First, he pointed out that automotive is going through a transformation with a huge percentage of the US market being electrified vehicles (including hybrids, not just battery electric vehicles). Only SUVs and pickup trucks are bigger. I wrote down 93% but that has to be a typo. So far this year, there has been $35B investment by automakers of which 96% had some element of electrification. For example, Ford is investing $11B in four facilities, two in Kentucky and two in Tennessee, comprising two battery plants, one car plant, and one truck plant. These will only produce under half of the capacity that Ford will need, so there will need to be more investment.

Battery and plug-in hybrid could be 20% of the market by 2030 (it is just 4% in 2021). The current administration is pushing for 50% by 2030 but there is a lot of pushback from the likes of Toyota that this is going too fast. Costs have to come down from the luxury level to where the middle class can afford electric vehicles. This is all good news for the semiconductor industry since there is higher semiconductor content in an EV than in an ICE vehicle. Another change (already in progress) is that the industry will make money not just selling vehicles but selling subscriptions and services, all of which are processor-intensive.

And, of.course, the big potential change is that the industry is rapidly moving towards autonomy. Ten years ago, people felt autonomy was almost imminent, then came the realization that it is harder than anticipated, but now optimism seems to be returning.

But in the short term there is the semiconductor shortage to cope with. As Bernard pointed out:

In Michigan, there are lots with thousands of stored vehicles. They are depreciating just sitting there. All those vehicles have to be brought back in to have missing components added (or sometimes they can be added at the dealership). One example: GM is selling a pickup truck missing an engine control module. They will stay that way forever but get less gas mileage.

In the Q&A, Bernard was asked if there are things other than semiconductors in short supply. He said that the semiconductor industry is taking the heat, but otherwise it would be something like petrochemicals or even magnesium. So other suppliers are happy that orders are down due to the semiconductor shortage so that they can meet demand.

Another question was about automotive manufacturers designing their own chips. Bernard pointed out that this is not a technology that is native to the automotive industry so they need collaboration. They had to get better at batteries before the semi shortage hit, so perhaps that's a blueprint for how an automotive company can work with a tech company. But the automotive product cycle is very long compared to electronics. By the time you design the car, the electronic product might have had two more generations. That was a lesson learned with keyless entry. Self-driving makes this worse so we need to get a whole lot better.

Another aspect, that Bernard did not mention, is a solution involving smartphones to provide the service (maps, music, podcasts, and so on) and just connect to the car through Bluetooth. Of course, that's not what the automotive industry wants, they want to sell you an expensive satnav system. I tried (but failed) to find a graph of sales of satnav systems and see if it confirms my suspicion that most people are not going to buy a navigation system for their car when they already have one in their phone, which is likely to remain up-to-date as they replace their phones over the years but keep their cars for many years so that they will be stuck with a current version navigation system for a long time. For example, in the mid-oughts, I was in Germany in a sales guy's Porsche and the navigation system relied on CDs in a unit in the trunk for the maps (well, in the front I assume since "the engine is in the trunk" as the old joke goes). I am sure that car is still on the road.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.

.