It has been a good year for Xilinx (XLNX) investors, with the stock recently remaining near the 52-week high of $46.54 hit on July 18, 2013, and almost 52% above the 52-week low of $30.25 on July 24, 2012, almost a year ago.
I've found that Xilinx products are poorly understood by many investors. In general Xilinx makes PLDs, Programmable Logic Devices, like rival Altera (ALTR) and others. CPUs and many other semiconductor chips are programmable, but the difference is in the nature of the programming. CPUs and programmable ASICs (application specific ICs) have instruction sets burned in during manufacturing. They can't change their instruction sets, but are programmed by software directing which instructions to execute. In contrast PLDs consist of arrays of elements (logic gates and memory) and the ability to configure the connections between elements after manufacturing. An engineer can make two originally identical PLDs do two very different things by configuring them differently. Unlike CPUs which usually accomplish a wide variety of tasks (spreadsheets, photo touchup, databases), PLDs can be programmed to do one thing repeatedly and very rapidly, like deciphering a cell phone signal. ASICs can do that too, but every time you make a change you have to redo the manufacturing process.
The most notable financial peculiarity of the PLD space is high margins compared to many semiconductor companies. Xilinx had a GAAP gross margin of 69% and an operating margin of 33% in Q2 2013 (its first fiscal quarter 2013). Competitor Altera had a gross margin of 68% and operating margin of 26.6% in the quarter.
Checking the basics, on a GAAP basis revenue was $579.0 million, up 9% sequentially from $532.2 million, but down 1% from $582.8 million in the year-earlier quarter. Net income was $157.0 million, up 20% sequentially from $130.6 million, and up 21% from $129.8 million year-earlier. Diluted EPS (earnings per share) were $0.56, up 19% sequentially from $0.47 and also up 19% from $0.47 year-earlier. [See Xilinx Fiscal Q1 2013 press release]
That looks good against the major competition, since Altera's revenue was $421.8 million, up 3% sequentially from $410.5 million but down 9% from $464.8 million in the year-earlier quarter [See Altera Q2 2013 press release]. This could represent a major competitive disadvantage, although Altera management has stated they believe Xilinx's current advantage will turn into a negative in 2014 mainly due to Altera's new alliance with Intel (INTC) and the nature of the addressable markets at the new process nodes.
Guidance is for revenue to be up 0% to 3% sequentially, which is disappointing. The September quarter should be seasonally stronger than that. However, growth of the newest, 28 nm (nanometer) products is expected to remain strong.
One reason for strength in the June quarter was demand in aerospace and defense, which is not likely to increase much in the September quarter.
One potential upside is from wireless LTE technology deployments by China Mobile. So far Xilinx has shipped only prototypes, and the bidding process between OEMs and China Mobile has not closed. So while shipments to this market could increase in late Q3, revenues are more likely to start ramping in the December quarter, or possibly in 2014. [per Xilinx analyst call, July 17, 2013]
I believe that rapid changes in technology, including segmentation, favor increased use for FPGA (field-programmable gate array) technology, and other types of PLD, over time. Both Xilinx and Altera seek to remain on the cutting edge of PLD/FPGA development, and they both have their fans in the OEM world. In 2014 we will see a battle of process technologies, but keep in mind that overall design is usually more important unless there is a big difference in process technologies, which there won't be. Also note that while 14 nm, 16 nm, and 20 nm parts may be introduced in 2014, much of the industry is still transitioning to 28 nm parts.
I also like the potential increases integration of other IP with PLDs, as we see in the Zynq parts, which include ARM cores along with the programmable logic.
I have a generally favorable impression of Xilinx, but with its closing price on July 23, 2013 of $45.83, barely off the 52-week high of $46.54 (as opposed to a 52-week low of $30.25 one year ago today), and a P/E ratio of 24.3, I am more inclined to watch the stock than to buy it. It does have a dividend of near 2.2%, which would provide some insurance against a price decline, which I like. The company also has a buyback program.
As I continue to study PLD products and end markets I will be looking for less obvious growth opportunities versus ASICs (Application Specific ICs, or regular, pre-programmed chips). The rivalry between Xilinx and Altera is important, but with both companies claiming their new products will win them market share, I am cautious about making bets on the claims this far in advance.