Semiconductor Cycles: Industrial and EVs
Plus my new taxonomy of cycles! Where I think each end market is in the cycle.
This earnings season, Industrial semiconductor demand has been the biggest incremental softening in end market demand. That’s not surprising. I have been talking about the FIFO (First-In, First-Out) cycle, and the only two remaining segments that have not had a meaningful correction are Industrial and Automotive. We are now seeing the beginning of Industrial weakening.
In contrast, smartphones, memory, and PCs are firmly on the other side of the cycle.
Most people watching global markets have noted the softening in demand in the industrial economy. And I think this is just another data point for what many people already suspected: we are in a global industrial recession.
Semiconductor Commentary So Far
I will create a stylized narrative on what has been said about the industrial sector on semiconductor earnings so far.
Macom is seeing a negative near-term outlook in Industrial markets.
Demand continues to be weak in telecom and to a lesser degree, in certain parts of the industrial markets and we remain negative on the near-term outlook for these markets.
NXPI sees better consumer demand but weaker communication infrastructure markets. They had their trough quarter in Q1 and have managed the cycle well.
We see the trends, including especially content growth to be pretty similar to Automotive. In our Consumer IoT and Mobile business, after over a year of weak demand, we see an incrementally improving environment. Finally, we do believe the weak demand in Communication Infrastructure and Other
Magnachip is seeing reduced orders in industrial, driven by higher inventories.
However, industrial markets, which has been an area of strength for us over the past several quarters, slowed by double-digit percentages in Q3, as our customers reduced orders to better manage their inventories.
Microchip, one of the companies with the highest industrial exposure, is seeing broad-based weakness.
The weakness is very broad-based across the different geographies, across the different end markets, with perhaps the one end market, which continues to have reasonable resilience is aerospace and defense, as you might expect. But it's extremely broad-based at this point.
Allegro Microsystems is starting to see inventory corrections and guided down.
The continuing macro trends are starting to have an impact on overall demand, leading industrial OEMs to become more cautious.
Looking into the third quarter, we expect to see a sequential decline in industrial sales as OEMs trimmed their production schedules and managed their inventory.
This quarter, Silicon Labs, arguably the weakest result, discussed how the industrial end market is collapsing.
The Industrial and Commercial business ended at $121 million, down 17% during the same period last year. All 3 product groups in I&C declined in the quarter with the broad industrial category experiencing the largest decline
Wolfspeed called out Chinese power and industrial demand, particularly inventory corrections.
We are seeing some softness in industrial and energy areas, particularly in China and Asia. Right now China represents about 20% of our total revenue, again, primarily in industrial and energy space. But in conversations with customers, that's really just kind of I think inventory timing, we're still seeing growth there. I'm still very heavy demand from automotive customers from China as well. So we're really just working through the revenue -- sorry, the inventory timing if you look out into the industrial side.
For Monolithic Power, industrial power was the biggest segment that declined quarter over quarter.
Third quarter 2023 industrial revenue of $42.1 million decreased 15.3% from the second quarter of 2023 due to lower sales in security and industrial meter applications. Third quarter 2023 revenue was down 28.2% year-over-year. Industrial revenue represented 8.9% of our total third quarter 2023 revenue compared with 11.9% in the third quarter of 2022.
Furthermore, Monolithic Power discussed the extremely poor visibility.
I think what we're experiencing here is a unique business cycle and that right now, our end customers are unwilling to commit beyond a fairly narrow window of -- expecting lead times within under 10 weeks for delivery. So that makes, as we said earlier in the comments, the predictability really hard to call right now.
STMicroelectronics talked about how the industrial segment was the reason for the midpoint of the guide to be below expectations.
The $100 million difference at the midpoint relates mainly to the industrial end market in Asia where the level of orders materializing towards the end of Q3 to learn of our Q4 backlog has been below our expectation. We confirm our 2023 net CapEx plan of about $4 billion.
While Renesas talked about increasing inventory in industrial in particular, inventory at the company level is going down but going up at distributors.
Next slide, please. So this is sales channel inventory. In all of the segments, basically, in terms of the WOI has increased Q-on-Q. The auto slight increase. Industrial and infrastructure increased. In both segments, it was a little over 9 weeks. And overall, the inventory level is 9 weeks plus. In terms of industrial infrastructure, the inventory in yen terms has increased quarter-over-quarter. That is that sell-through decreased more than expected industrial and infrastructure segment to -- in terms of mass market sales.
Meanwhile, Texas Instruments continues to see record-high inventory and ramping capacity and now beginning to see industrial weakness.
During the quarter, automotive growth continued and industrial weakness broadened. Similar to last quarter, I'll focus on a sequential performance as it's more informative at this time. First, the industrial market was down mid-single digits, with weakness broadening across nearly all sectors. The automotive market continued to grow and was up mid-single digits. Personal Electronics was up about 20% off a low base. And next, communications equipment was down upper teens.
I’ll leave it finally with Infineon, which says that Industrial and automotive are strong, while consumer, PC, and smartphones are weak—a rather backward-looking assessment.
These divisions remains very dynamic. In addition, some product sectors where, as expected, we find the market environment normalizing. And they remain good. These include classic automotive components, industrial drives. Other fields such as consumer applications and personal computers and smartphones, the demand remains weak. We find a reduction in inventories in these sectors and the demand among end users needs to recover.
So what are we to do with this new information? I think it’s time to be cautious about Industrial.
Industrial Weakness in the Big Picture
I want to zoom out of the tail-end of the bullwhip and look at the big picture. It’s one thing for industrial to be weak in the micro (semiconductors), but where is this showing up in the big picture? So far, I see it in three ways. Chinese industrial production is tanking, Solar is weakening, and EV inventories are rising.
Let’s start with Chinese industrial production. Multiple companies called out Asia, or in this case, China. This is a chart from Bloomberg Intelligence, and sometime in October, refinery utilization plummeted.
China has been in a prolonged contraction, and it has worsened. Now, no one knows what the future holds, and given China is about to pursue a new round of stimulus, who knows where this ends up? The near term is not good, however.
Residential Solar in the United States is having issues. Enphase is the leading residential solar microinverter company, and their results have fallen off a cliff. This is another leg of the industrial weakness.
Utility-scale solar projects are not as weak but have had some slight pushouts.
But last, I want to talk about Automotive, which has been the darling of semiconductors for so long. It’s no secret that EVs are a large part of the automotive content story. It’s a doubling of content over an ICE vehicle. The problem is that we are starting to see real EV inventory problems.
Multiple sources confirm this. Within the US, there is just a low amount of EVs that are moving compared to traditional ICE vehicles. This is from CarDealershipGuy on Twitter.
Meanwhile, EV inventory is skyrocketing in the US. Days of inventory are rising, and financial incentives are rising as a percentage of the sales price.
This is mostly US-centric, and Europe and China are a bit stronger. A large part of this in the US is the rise of interest rates, making car financing unaffordable. But even in China, we see the first pockets of inventory overbuild. This is from SMIC’s most recent earnings.
Besides, the inventory of automotive-related product is now in relatively high level after being in short supply for 3 years, which has caused major customers to be alert to the market correction and to tighten their orders placing rapidly. After more than one year's ups and downs in the market, customers have experienced a shift from aggressive expansion 2 years ago to tough defense this year.
I think it’s time for the Automotive to turn and follow Industrial quick deceleration.
Let’s turn now to the cycle. A few years ago, I was in the camp that semiconductors would get a bit less cyclical on average. Multiple end markets drove this, each having its own inventory cycles, and on the net, it would smooth out the results.
This has come true. While PCs and smartphones had a massive correction, Automotive and Industrial stayed strong. However, each of the constituents of the entire semiconductor market has not gotten less cyclical; aggregate end markets are not in lockstep.
Automotive and industrial technology have been relatively benign, while PC, Smartphone, and Memory have had some of the worst cycles in recent history. Now it’s time for them to take leadership and for Automotive and Industrial to turn.
This is my attempt to split the semiconductor market into end markets and evaluate each end market’s cyclicality. It’s a simple framework: revenue and inventory. Revenue is falling or rising, and inventory is falling or rising. Those are the four quadrants.
The worst place to be is rising sales and inventory rising faster, which inevitably leads to over inventory and revenue declines. The best place to be is when inventory starts to fall; that’s the early cycle. Stock markets appreciate this and will bid up stocks before inventory troughs for this reason.
Today, we can see this dynamic in multiple end markets. I consider this by looking at days of inventory and revenue change quarter over quarter.
Taking this framework, we look at where each company is in the cycle. Automotive and Industrial are clearly in a market that is not as correlated to the others, while PCs and Smartphones are the biggest improvements this quarter. I added arrows to indicate where each end market is going as well.
I’ll be posting this chart from time to time as markets develop. I’ll try to add sizing (sorry, I won't do it in my first version!) and be consistent. This is the best way to visualize the cycle so far.
I have more behind the paywall for information about which companies are best positioned and how I think it’s time to play the cycle.