10.18.23
Total memory spending (the company doesn’t break revenue out by memory type) rose to €1.27B from €897M in the previous quarter. Lithography spending doesn’t show when the memory downturn started. It peaked in Q4-22, declined slightly the next quarter, then fell nearly in half the quarter after that. I think this is because memory makers push out scanners as their last move in reducing capex spending. Net bookings for memory declined sharply, from €1.40B in Q2 to €520M in Q3. That level of bookings is below any single quarter’s revenue for all of the history of ASML. Some of that must be timing. Still, it is a measure of the depth of the memory downturn. Memory fab utilization of scanners remains reduced and has not yet inflected upward. Hopefully this remains the case for at least a couple of quarters after ASPs turn up. Wennink gave a history lesson on why he believes 2025 will be a strong year for them. The severity of a downturn historically determines the strength of the upturn that follows. The last two years have been historically bad for memory, so the coming upturn – which he signaled would begin in 2024, else this downturn will have lasted 2.5 years – will be historically strong. They are seeing early indicators in pricing of the memory market turning up, both NAND and DRAM. For memory investors, we hope to see ASPs increase rapidly for as long as possible before memory producers start increasing their orders for lithography equipment from ASML, because a longer delay means a longer and stronger upturn. WFE companies are the end of the bullwhip in the memory cycle. If the memory pricing increases visible now prove to be the beginning of the upcycle, I hope it will be at least three quarters before orders to ASML from memory companies are seen. When they do come, I also hope they are gradual. Memory makers need to hold the line on adding capacity to support strong pricing for as long as possible. They have a lot of cash losses to make up for.
-S. Hughes (cyclical long MU)