Xilinx announced their results today and had their conference call this afternoon, which I listened to. For them this is 1Q fiscal 2015 which means you have to be careful since there is a big difference between talking about fiscal quarters and calendar quarters. Xilinx’s conference calls are interesting for a couple of reasons. Firstly, of course, they are the leaders of the FPGA market so it provides insight into that market. Secondly, Xlilnx FPGAs are one of the process drivers for TSMC and so there is some insight into exactly where the various processes are in the ramp to volume.
Revenue was $613M, down 1% from last quarter. However, profitability was up with net income of $173M with gross margins of 69.1% (which is obviously higher than expected since net income is up on lower than predicted revenue).
The reasons for the variance given on the call were primarily China LTE deployment, and aerospace & defense.
- wireless revenue was slightly down due to lower than anticipated 28nm sales to China LTE base station manufacturers due to delays in the deployment of 3rd phase LTE rollout there. LTE deployment outside China was healthy and there were higher than expected shipment of 40nm and 65nm parts
- aerospace & defence was weaker due to the timing of some programs moving into September, which should see a rebound next quarter
- wired communication sales (routers etc) were up and higher than anticipated driven by enterprise networking, optical (OTN) and access
- industrial, scientific and medical were as expected.
Going forward they are forecasting a slower 28nm ramp than previously, again driven by the slowdown in China LTE deployment and slower than forecast ramp in the wired communication. However, Xilinx are confident that 28nm will be the most successful node in their history as it continues to ramp and design wins go into volume over the next several years.
Kintex UltraScale, the industry’s first 20-nanometer device been in the market since November and according to their largest customer is providing them with more than six months advantage over you-know-who. During this last quarter Xilinx started shipping Virtex UltraScale the industry’s first and only 20-nm high-end device.
In the questions there was some more color on the issues in China LTE rollout. Last quarter a lot of product was shipped but due to incomplete kitting (other parts not being available, nothing to do with Xilinx, they were not deployed, leading to a slowdown in this quarter which should work out in the next couple of quarters). China Mobile (the biggest operator in China, or indeed the world) was expected to deploy 4-500K base-stations and so far have only deployed about 300K.
Xilinx reckon their share of 28nm products was ovder 70% last fiscal year, and should be in the high 60% in this quarter at around $130M. The target for the (fiscal) year is $600M.
They are looking at possible belt-tightening but are committed to carry on their 20nm and 16nm development. Xilinx’s strategy remains to get new technology out as fast as possible.
Moshe was asked if they see risk in TSMC’s schedule and whether they would consider changing foundry vendor. Of course the correct answer is NFW but what Moshe actually said was:We are delighted with the progress of TSMC. As best we can tell, they’re on schedule and they have numerous other users of the technology who actually, in this case, will even be ahead of us. So, there really is no issue, in our mind, on the availability of the FinFET from TSMC. And, if anything, we’ve heard significant delays with other players.They’re doing 16 in two cycles. They’re doing a FinFET and they’re doing the FinFET Plus version, and we’re going to be using the FinFET Plus version. We’re benefiting from all of their development at this point in time.
Xilinx said they are on-track to start taping out 16nm designs in Q1 (calendar) 2015.
Transcript of the conference call at SeekingAlpha is here.
More articles by Paul McLellan…
TSMC Unveils the World’s Most Advanced Logic Technology at IEDM