Headwinds which will likely continue into 2016…
ASML reported revenues of 1.55B Euros with EPS of 0.75 Euros more or less in line with expectations. Orders were the weak spot, falling to 904M Euros versus the previous Q2 orders of 1.523B Euros. The company guided Q4 revenues to be down about 10% to 1.4B Euros below current flattish expectations. The company largely blamed foundry weakness as the primary culprit but we are concerned that they will also see memory slowing much as other equipment companies have reported.
EUV stumbles & delays yet again…..Intel likely part of the pushout
The company now expects to ship only 4 EUV systems in 2015 versus the prior expectation of 7 systems. We had predicted this push out yesterday after hearing Intel’s comments on their call which all but named ASML as the equipment company.
The key issue here is why?
ASML said that customers were being cautious and that was the reason for the delay. However, Intel said that the tool(s) needed to be reconfigured for higher output (ie; they either weren’t making spec or didn’t make required milestones).
Intels CEO, BK said “So you have to remember, there is this lag, and that’s why, as we looked at the tool, actually, we are making an adjustment on the efficiency of that tool, basically the number of units per tool out. In order to get more capacity, when that tool is really required”
It sounds like he is saying that the units per tool (read that as wafer throughput) is too low and needs to be higher. Interestingly he also added that when the tool is really required which would indicate its not yet required so we can push it out and wait for the latest and greatest version of the tool when its ready for prime time….sounds a lot like it may not be required for 10nm (as Intel has stated they have a non EUV process flow for 10nm). This is not the first time that Intel has laid the blame on ASML as we have previously pointed out.
Immersion tools slipping?
We would point out that the number of immersion tools shipping is slipping over the last year. While upgrades and service increase may point to customers finding other ways to get more out of existing tools it could be that customers don’t want to invest more in new, expensive immersion tools while waiting for next generation EUV. Its likely we are seeing more equipment reuse in the near term.
EUV system up time still an issue…
ASML said that at “certain select customers” EUV had up time of 70%. which sounds a lot like if you averaged the whole customer base that up times would likely be well below 50%. So even if you can do 1000 wafers per day, the up time cuts that in half to 500 wafers per day (or less) suggesting we still have a long way to go in improvements
Memory risk…
So far, ASML has not seen as much of a drop off in memory spend as others in the industry have reported . Memory went from 38% of orders to 52% of orders in the current quarter. Given concerns about memory pricing and the overall stability of the memory market we think this high exposure adds risk. Orders also fell from 41 systems in Q2 to 32 systems in Q3 with foundry falling from 38% to 23% and IDM (likely Intel) fell sharply from 39% of orders to 10%.
The stock…
As expected the stock had a negative reaction. We think there may be more downside as investors digest the EUV issues and memory risks. The company did not say a lot about 2016 which makes us a bit nervous about the outlook for next year if memory does indeed slow. We think there could be potential downside to $80ish.
Robert Maire
Semiconductor Advisors LLC
Next Generation of Systems Design at Siemens