Analyst Robert Maire has previously pointed out in ASML’s earnings note that for a long time, their view of the semiconductor industry is to look at things through two different components of supply on the one hand and demand on the other.
To this end, they argue that the “first round” of declines in the current downcycle has been largely supply-driven, as the industry frantically builds capacity following the Covid and supply chain crisis. The industry built and built, apparently recklessly giving up, until we “surpassed” the supply we needed, and now find ourselves in a state of oversupply, starting the current down cycle. At the same time, the demand side has softened, perhaps with global macroeconomic concerns, and the downward pressure on demand has accelerated.
In their view, the “second phase” of the current chip cycle is likely to be driven more by weak demand than by oversupply in the first half of the year. This can be worse because supply issues tend to be easier to solve than demand issues because if you’re a chipmaker you can control supply, as we’ve seen in the memory market with Micron and Samsung pulling capacity and products off the line to support pricing. The problem is that the industry can do little to stimulate demand for chips. Lowering the price of automotive chips does not stimulate demand. The reason for making such a judgment. They predicted it from ASML’s financial report.
The industry has long believed that no chipmaker in its right mind would cancel or delay a lithography tool for fear of returning to the end of a very long production line later and in a worse position, they said. But that fear seems to have dissipated, as there was a noticeable shift in ASML‘s order book, with tools being rolled out and other customers pulled in to fill what had been vacant positions. Quite another point, chipmakers such as TSMC, Intel and Samsung, if they are willing to delay key lithography tools, will surely feel that demand will not improve anytime soon. This suggests a deeper and longer-lasting recession in the semiconductor industry than is currently or previously thought. You wouldn’t put off new lithography tools if you thought the industry would “bounce back.” This has very ominous implications for the industry as a whole, as it overshadows the notion of a quick recovery.
Robert Maire reiterated his view after today’s earnings review of Lam research. Lam’s memory business is down to lows we haven’t seen in a long time, as memory is 32% of the overall business and foundry is 46%, he said. A larger portion of the memory business is likely to be services, which also declined sequentially as new tool shipments to memory customers likely fell relatively close to zero.
Lam made it clear that memory capex has not yet bottomed out, as they commented that memory spending is expected to “fall further.” The company estimates that memory capex has fallen by 50% as memory makers continue to reduce bit growth and bit output in an attempt to support falling prices.
We seem to be fairly clear, given the trajectory and momentum that memory will not recover before the end of the year, and when it does eventually, the strength of the recovery will be weak and slow. Memory makers will have a lot of existing capacity that can be brought back online before they buy a new piece of equipment, or even consider expanding or adding a new fab. In addition to idle memory manufacturing equipment at fab shutdowns, there are technology curtailments that add capacity without buying new equipment. At the end of the day, memory makers can easily survive a year or two based on current demand trajectories without any additional spending due to idling of equipment, declining production volumes, and potential technology changes.
The bigger problem is that in the current loss-making environment, after another year, perhaps only Samsung will have the financial ability to survive. And whenever the current down cycle ends, it certainly won’t be memory that will lead the way.
Post time: Apr-28-2023