I know that cfoe and others here have spent a lot of time in developing projections of QCOM's future revenues/earnings. I've been doing something along the same lines, but with a somewhat different objective, and more on my efforts later.
Right now, I'd like to get some feedback from anyone who has any ideas on the apparent decline in QCT chipset ASP's, which is beginning to bother me. I thought I had a decent handle on this important aspect of QCOM's core business, but the last two earnings reports have raised some questions, and while there's been some reported analyst discussion of handset ASP's (on which the royalty stream is based), I've seen nothing from them on chipset ASP's.
First, here's the chipset sales and revenues for the last five Q's in which 1x chipsets have been shipping:
FY CSM 95A/B 1X QCT Revs Blended ASP
Q2 01 1.0 15.0 1.0 $363.741 $22.73 (1)
Q3 01 3.0 12.0 2.0 $333.115 $23.79 (2)
Q4 01 4.0 9.0 4.0 $336.881 $25.91 (3)
Q1 02 1.0 9.0 6.0 $359.144 $23.94 (4)
Q2 02 2.5 6.0 8.0 $343.815 $24.56 (5)
"CSM" here is the way QCOM reports, i.e. the millions of "equivalent voice channels" supported by the number of CSM infrastructure chips shipped, and not the actual number of chips. I include it for completeness, but it turns out to be an irrelevant (to revenues) number. The (n) at the end of each line is just for ease of reference in what follows.
After the Q4 results last November, I did some work on the data to get a handle on the ASP of 1x chips versus the $22 per chipset QCOM was apparently getting for IS 95A/B chips. Using the data in lines (2) and (3), I came up with ASP's of $21.97 and $34.78 for 95A/B and 1x chips respectively. You can round these off to $22 and $35 for practical purposes, but I left them as calculated since they're on a spreadsheet. In this calculation, I ignored the revenues from CSM chips based on two posts from S100 (may he rest in peace, I also valued his insights and comments) and cfoe back in February:
Message 17125162
Message 17128871
With the $22 and $35 in hand, I checked to see if using these results gave me the revenues in (1) - and they did, almost to the $. So I was reasonably confident going into the Q1 02 report that QCT revenues would be around $406 million, given the guidance from QCOM as to shipments. Shipments came in as predicted at 9 and 6 million for 95A/B and 1x chips, but revenues were only $359 million - see line (4).
So what happened? 2 million additional 1x chips shipped and only $22 million additional revenues works out to about $11 per 1x chip, way below my $35 estimate. Not possible, I thought. Could it be that the CSM revenues were in fact significant, and the Q1 02 shortfall was due to the decrease in CSM shipments? With 4 Q's worth of data - lines (1) through (4) - and only three variables, it was straightforward to extend the analysis to calculate the revenues/chip for all three categories - CSM, 95A/B, 1x (thank goodness for spreadsheets and their ability to invert matrices and solve sets of linear equations!).
Negative result on the CSM revenues. S100 and cfoe were correct, no matter which combination of three data sets I used, CSM chip revenues came out to be so small as to be effectively zero in comparison to the MSM revenues. If QCOM accurately reported the numbers of MSM's shipped instead of "rounding" to the nearest million, it might be possible to get a realistic number for CSM chip ASP's, but they don't, and it introduces little error to fold those revenues into the MSM ASP's.
So I was forced to conclude that QCOM was discounting the selling price of either the 95A/B chips, 1x chips, or both. Why would the blended ASP of 1x chips be decreasing so early in the commercial cycle, especially with no real competition, and with presumably higher priced 1xEV-DO chips shipping into Korea to boot? Looking back on the ASP's for 95A/B chips over the years, I came up with the following numbers:
Year Quarter # 95A Chipsets Revenues Chipset
(Millions) (Millions) ASP
1999 Q1 5.0 $193.315 $38.66
Q2 9.0 $263.411 $29.27
Q3 11.0 $318.324 $28.94
Q4 14.0 $358.372 $25.60
Year Total 39.0
2000 Q1 14.5 $352.395 $24.30
Q2 11.0 $279.187 $25.38
Q3 15.0 $338.132 $22.54
Q4 11.0 $268.989 $24.45
Year Total 51.5
2001 Q1 15.0 $330.054 $22.00
Based on this historical data, I decided that QCOM was unlikely to be discounting the ASP on 1x chips, and that the Q1 02 shortfall was due to selling off 95A/B inventory. What was the discount? Easy to figure from the line (4) data for Q1 02 that they were selling 95A/B chips for $16.72, around 24% discount to the previous price of $21.97. Use the rounded $35 for the 1x chips and you get $16.50 for the 95A/B chips, a nice round 25% discount from the previous $22. Bit tough on Nokia I thought, just when they thought they'd got 95A/B chips right, QCOM whacks them with a price cut!
So, on to Q2 02. Using the $34.78 and $16.72 ASP's and QCOM's updated guidance of 8 million 1x and 6 million 95A/B chips, I projected $378.5 million in revenues ($379 million if you use the "rounded" $35/$16.50). I expected to get confirmation that my discounted 95A/B chip ASP was correct, but QCT's revenues came in at a disappointing $343.815 million. Again, it's not CSM chip sales, they were actually up from Q1 02 (see lines (4) and (5) in the first data table).
So, a couple of possibilities again. Either:
(1) QCOM had cut the ASP for 95A/B chips even further; or
(2) QCOM was also now discounting the 1x chips.
Easy to figure (1) - if this was the cause of the lower revenues, 95A/B chips are now being sold for $10.93, while the mean ASP for the 1x's is still $34.78. Call it $11 and $35 respectively. That's around a 50% discount on 95A/B chips over the last two Q?s. Is this it? QCOM is stuck with some inventory of the MSM 3xxx's that they are dumping off so as to switch over to the 1x and 1xEV-DO MSM 5xxx's as quickly as possible? If so, even tougher on Nokia! I also wonder which customers are still buying IS 95A/B MSM-3xxx's? I don't believe they support the R-UIM's that are needed in China, so they must be going into other end markets that don't need 1x, or which need a "cheap" basic CDMA chip. I do recall reading some comments somewhere recently that QCOM expected to be shipping 100% 1x by the end of 2002.
If (1) is the answer, then the 10 million 1x and 5-6 million 95A/B chips projected for Q3 02 will result in QCT revenues of $402-$413 million ($405-$416 million if you use my rounded ASP's), a 17-21% Q/Q increase, which would be pretty impressive. I haven't cranked these revenues through my earnings model yet to see how they stack up versus the QCOM projected $0.19 - $0.21 EPS, but that's the next step.
OTOH, if QCOM is maintaining the same apparent discount for 95A/B chips as before - around 25% for an ASP around $16.50 - then they must already be reducing prices on the 1x chips, to result in a blended ASP around $30.50, about a 13% reduction from the prior Q1 02 $35 price. If (2) is the answer, then QCT revenues in Q3 will be $388-$405 million, still a 13-18% increase, but less than for (1). Also, if the ASP's for 1x chips are already being cut in price, then we could see more reduction in 1x ASP's over the course of the year, with a corresponding impact on projected QCT revenues.
If 1x ASP's are already coming down, despite being in the early stages of the product cycle, despite the release of 1x and 1xEV-DO chips with more functionality, despite the fact that there doesn't appear to be any real competition for QCOM in 1x right now, and despite the fact that 1x handsets are not being sold "on the cheap" as yet and so could easily absorb a $35 chip ASP, then I don't like the implications for QCT revenue growth going forward, which will depend on increasing unit sales volume rather than stable and higher chip ASP's, which I'd previously assumed would be the case for at least three or four Q's.
I'd appreciate any thoughts anyone might have on this subject, or on the analysis presented in this post.
David T. |
