At its November 6, 2006 analyst conference Altera reported revenues of $341.2 million for its 3rd quarter ending September 30th. It predicted 4th quarter revenues would decline 2 to 5%. Then when it reported Q4 revenues at $317.4 million at the February 13, 2007 conference, they were down a whopping 7%. A further decline of up to 4% in revenues was given as guidance for the first quarter of 2007. Should we panick? Is Altera sliding into oblivion? Is this 2001 all over again?
Altera makes semiconductor chips and specializes in FPGAs (Fully Programmable Gate Arrays) and CPLDs (Complex Programmable Logic Devices). All these devices can be contrasted to ASICs, which are Application Specific Integrated Circuit. Supose you have an application, perhaps processing some information inside a cell phone or a hard disk drive. You decide what the semiconductor chips have to do (what their logic structure is) in order to achieve your goals. With an ASIC you then design the layout of the logic that will be in silicon. To create the ASIC chips you go through the standard manufacturing process of using masks, doping, and etching to create batches of chips incorporating your design. FPGAs and CPLDs approach the same problem by giving engineers sets of general solutions that are arrays of possible logic states. You program the FPGA to have the same logic that you would build into an ASIC. So the end products should function identically. Why not use all FPGAs, or all ASICs? There are cost advantages to ASICs once a design is nailed down and if production is on a large scale. But when an engineer sits down to design an ASIC, she could just reach into her parts box, pull out an FPGA, and program it instead. So FPGAs go from idea to finished product much faster. If you want to fix an error, or do an upgrade, often all you have to do is reprogram the FPGA. With an ASIC you have to redesign the circuits, make new masks, etc. Speed and flexibility are what FPGAs are about, but they are also great for prototypes and specialty products with low production runs.
Because of this Altera's competitors are not just other FPGA and CPLD manufacturers; they are ASIC manufacturers as well.
While prices of all stocks fluctuate, there is no way a company like Altera, in the price range it has been in lately (52 week high $22.32, 52 week low $15.54), can be compared to the revenueless Internet stocks of yesteryear. In fact even in the relatively lousy 4th quarter Altera had net income of $99 million, up 43% from its Q4 of 2005. It is in a competitive business, but not that many companies are competing in programmable silicon, so profit margins are healthy.
Altera constantly invests in the future: its R&D runs about $60 to $65 million per quarter. Part of the recent revenue decline is due to customers reducing inventory, getting into cautious positions for 2007 sales. 39% of revenue was in sales to the communications industry, which continues to deploy not just new models of cell phones, but whole new types of products like WiMax and 3G phones and their attendant base stations.
Who competes head-to-head with Altera? Actel, Atmel, Lattice Semiconductor, NEC, Quicklogic, and Xilinx. Actel had $48.2 million in Q4 2006 revenues and has FPGAs as its principal products. Atmel primarily makes microcontrollers, though it does have several FPGA lines; it has been struggling the last few years. Lattice Semiconductor is focused on FPGAs and had revenues of $61.8 million in Q4 2006. NEC is a huge company which derives only a small amount of revenue from FPGAs. Quicklogic’s web site is worth a look: they openly and cleverly challenge Altera and Xilinx on their home page. Their Q4 revenue was $7.7 million, plunging 25% year-over-year. And then there is Xilinx, the main competition. You can find out more about XLNX in my analyst conference summary for this company, which claimed 70% of the global FPGA market in 2006.
The only real short-term threat to Altera’s profitability is a global recession. In the long run their competitors are a threat, but Altera also has the potential to eat into Xilinx's market share. Overall PLD (programmable logic device) usage is certain to grow.