Introduction
Actel Corporation is a niche player in the FPGA industry. An FPGA, or Field-Programmable Gate Array, is a semiconductor logic device that can be configured by the end user after its manufacture (as opposed to its distant cousin and competitor, ASIC, which is hard-wired). The four players in the space are: Xilinx and Altera, which have ~85% market share and focus on SRAM/static-memory-based FPGAs; Lattice, which also focuses on SRAM-based chips; and Actel, which specializes in antifuse- and flash-memory-based FPGAs. We think the key things to understand in the Actel thesis are:
- FPGAs are not afflicted by the curse of the semiconductor industry, namely, short cycle times because FPGAs mostly don't go into consumer products - a chip that is designed into an elevator, for example, is designed in for years. The typical lifespan of an FPGA is over a decade, and even then, product revenues decline only gradually.
- Actel does not compete head-on with the "Big 2" (Xilinx and Altera). The vast majority of FPGAs are designed based on SRAM memory. Xilinx and Altera focus exclusively on SRAM-based FPGAs, as does Lattice, which competes directly against the economies of scale in the SRAM duopoly. By contrast, Actel does not dedicate any resources to SRAM-based chips. Within the antifuse-based and flash-based FPGA arena, which Actel specializes in, there are no direct competitors.
- We think the best way to understand Actel is as the sum of the following parts:
- Antifuse-based Mil/Aero FPGAs - the crown jewel of Actel's businesses, there are almost no alternatives to Actel's antifuse FPGAs for the applications that presently use them. The result is fat margins, growing sales, and customer captivity.
- Non mil/aero antifuse-based chips - the legacy Actel business. For most end-users here, there are few alternatives to Actel's chips, but the number of end-uses is not growing significantly. Still, this segment generates high margins and enjoys solid customer captivity.
- Flash-based FPGAs - billed by management as the future of the company. Actel has spent, we estimate, close to $200M in R&D on Flash-based chips over the last five years and the company continues to spend significant sums on Flash R&D. The segment has only recently started to generate noteworthy revenues, and presently is a significant cash user.
Antifuse-based Mil/Aero FPGAs
Antifuse chips posses high levels of security, low power consumption, and most importantly, little susceptibility to radiation-induced disruptions. The key disadvantage that antifuse has is that it's one-time programmable, i.e., once the user configures the chip, it can't be reconfigured. SRAM-based chips, by contrast, are reconfigurable continuously and have very high performance capabilities, but they are power-hogs and are susceptible to radiation-induced disruptions. Military and aerospace applications are frequently mission-critical and the end-user can't risk a radiation-induced disruption (you can't simply replace satellite parts, for example). For the most part, then, Actel has no FPGA competitors in this space. In fact, for applications where the performance capability of an SRAM FPGA is required, the SRAM chip itself (designed by Xilinx or Altera) is controlled by an Actel antifuse FPGA.
Presently, the vast majority of logic applications in mil/aero use hard-wired ASICs rather than FPGAs. Aside from the significant disadvantage of ASICs not being programmable by the end user (they have to be configured in the fab at the time of manufacture), the long term trend in the ASIC business has been, and continues to be, a rapid increase in startup and design costs. This has led many applications that would have used ASICs in the past, to consider using FPGAs instead, and indeed FPGAs are winning designs and market share from ASICs at present.
To give you an idea of the runway Actel has in mil/aero, consider the space market (mostly satellites). In 2007, the market for logic semiconductors in space was ~$200M, with $30M being Actel FPGAs and the rest ASICs. Actel expects to be able to grow significantly in this space (+15% CAGR over the next five years), as the total market and Actel's market share grow. The opportunity in military applications is even larger (75% of the $600M market is ASICs and Actel has 5% market share).
Actel has generated ~$82M in TTM revenue. Based on our conversations with management, we estimate that operating margins in this segment are ~25%, resulting in EBIT of ~$20M. A 6-8x EV/EBIT multiple on operating earnings here seems extremely conservative given the durability of the business, its likely double digit growth rate, and the utter lack of competition when it comes to mission-critical applications. For the sake of comparison, Altera and Xilinx sport a 10x EV/EBIT multiple at present. At 6-8x EBIT, we're looking at a segment worth ~$150M (that's an EV/Revenue of less than 2, compared to Xilinx and Altera's >3 valuation at present).
Non-Mil/Aero Antifuse
While there are also applications in the non-mil/aero space that require a chip that's not susceptible to radiation-induced disruption, the macro trends for this segment are not as positive as in mil/aero simply because the opportunity for enormous market share gains from ASICs is non-existent (ASICs never really had market share in these applications in the first place). Still, the antifuse chips in this segment have high margins, require little R&D to maintain, and have no competition.
In the trailing 12 months, Actel generated ~$81M in revenue in this segment, and we estimate operating margins here in the 20% range, resulting in EBIT of ~$16M. Given the multiple and comps we looked at earlier, we think 5x EBIT is a fairly conservative valuation to put on this segment (we fully admit that no- or slow-growth businesses deserve lower multiples). At that valuation, this segment is worth ~$80M.
Flash-based FPGAs
The flash-based technology that Actel has been developing over the last 8 years seemingly combines the best of the antifuse and SRAM worlds. Flash-based FPGAs are highly secure, have little susceptibility to radiation-induced errors, are reprogrammable at all times, and - extremely important in today's world - use very little power (orders of magnitude less than SRAM-based chips).
Unfortunately, flash-based FPGAs at present do not have nearly the performance capabilities of SRAM-based chips, so Actel's not really going to be winning significant market share from Xilinx and Altera at the upper end of the market (where they generate the vast majority of their revenue). Still, the opportunities in flash are enormous for Actel and can be summarized as follows:
- Market-share gains on Xilinx and Altera in the "value-FPGA" area, where FPGAs are needed due to their reprogrammability and low costs relative to ASICs, but extremely high performance is not required. The flash chips use significantly less power, are more reliable, and are more secure in terms of reverse engineering the intellectual property on the chip than any SRAM-based chip. The total market size of value-FPGAs in 2008 is ~$2.5B. Actel has gone from ~0% of this market in 2005 to a $60M annual run-rate in the recent Q3, as it wins new designs against the Big 2.
- The value-FPGA pie continues to grow. Similar to the trends in mil/aero, but at a faster pace, new designs in applications that have traditionally used ASICs are being won by FPGAs. Historically, FPGAs were uneconomical in high-volume applications due to high per-unit costs. But as the start-up costs for developing an ASIC continue to rise, the per-unit costs of an FPGA are starting to become more attractive, and with an FPGA, the end user also gets the benefits of reprogrammability. The other major weakness of FPGAs vs. ASICs - power consumption - is precisely the area where Actel's flash technology has a big edge over the Big 2's SRAM-based chips. So as FPGAs take share from ASICs, Actel should win a disproportionate piece of that share.
The big question is how to value the flash segment. We estimate that flash has had TTM EBIT of negative $40M on ~$53M in revenue. We think that GAAP EBIT, though, is not the best way to analyze this for a few reasons. First, of the ~$45-50M in R&D that has been spent on flash in the trailing 12 months, almost none of it relates to the two product lines that are now being sold in volume. Given the extremely long cycle times for FPGAs, this year's R&D pertains mostly to products that will generate revenue in 2-3 years at the earliest. If the R&D that has been invested in the past in flash had been capitalized for each flash product and then depreciated over the useful life of that product, we estimate on that basis that flash would actually be profitable. In addition, because the flash product lines are relatively new, they have not gone through the typical cost-cutting process that semiconductor devices go through on their way to achieving optimal gross margins.
While it wouldn't be audacious to value the flash business as the sum of the last five years' worth of R&D (on the assumption that that number would equal "replacement cost"), we think that valuing it at half that number, or ~$100M, is fair and properly weighs the probability that flash isn't all it's hyped up to be. That assigns a value of less than 2x revenue to a semiconductor business that has had a 6.5% compounded quarterly growth rate over the past 2 years, with no technological equals, and significant potential to become a disruptive technology.
Risks & Conclusion
The major risk here would be that the flash technology turns out to be a dud and that the company continues to pour resources into it. In that scenario, we see little downside considering the $70M enterprise value the market is giving Actel - it is surely worth more than that to an acquirer in a field that has historically been rife with M&A. But the stock could be "dead money" in such a scenario while the well-capitalized balance sheet should be useful in mitigating any permanent impairment of capital. The other major consideration is that Actel's management does not break out their results in a similar fashion as we have in this write-up. Thus, it requires a lot of digging and speaking to management to come up with reasonable estimates for sales and margins of the different parts of the business as we see them, and the market may not come around to this valuation without better publicly distributed information.
On a sum-of-the-parts basis, we think Actel is worth ~$330M plus the $130M in cash that is presently on the balance sheet, for a total market value of ~$450M or ~$17.50/share. The stock presently trades at ~$8.