Chips in a High-Stakes Poker Game

Despite the current economic downturn, the long march of technological advancement continues. Electronic devices, from PCs to cell phones to digital cameras, are getting smaller, more powerful, and more energy-efficient. Most of this progress is driven by leading semiconductor companies that design the chips that power these devices. Advancements in chip design, however, are often enabled by advances in manufacturing technology. Thus, third-party chip foundries, which act as outsourced manufacturers for most semiconductor firms, are an integral part of the industry. Let's take a closer look at the major semiconductor foundries and open the playbook on how to invest in these companies.

Industry Overview
Semiconductor manufacturing is like a high-stakes poker game--the entry fee is high, the game changes quickly, and firms either keep up or get knocked out. Chip manufacturing is highly capital-intensive, as firms spend billions of dollars to set up new facilities, called fabs. Fabs require pristine settings that are free from vibrations, temperature changes, or even the smallest speck of dust. Meanwhile, the equipment used in chip fabrication is highly specialized, as these machines are designed to build semiconductors that include billions of transistors on a piece of silicon often smaller than a fingernail. Furthermore, the next wave of smaller, more powerful chips can only be built with the most advanced tools, so new equipment needs to be purchased with each technological transition.

Understandably, semiconductor firms have taken measures to avoid the high fixed costs associated with setting up and running a fab. Many firms, such as  NVIDIA (NVDA) and  Broadcom , have deployed a "fabless" business model, where they design chips but outsource production to firms that specialize in semiconductor fabrication, called "foundries." Chipmakers benefit by avoiding the high "entry fee" for manufacturing, instead focusing their research efforts on design and functionality. Foundries capture a fee for production and become an important part of the supply chain, as chipmakers have become increasingly reliant on foundries to meet their manufacturing needs. Through specialization and working with many different chipmakers, foundries can capture a body of knowledge about certain manufacturing processes that can be leveraged to attract new customers. For all of these reasons, we expect the trend toward the fabless business model to continue.

Despite these tail winds, foundries are hardly assured of generating consistently strong profits. The semiconductor industry is highly cyclical, as the balance between supply and demand constantly teeters in and out of equilibrium. Supply/demand instability occurs because the semiconductor manufacturing business is commoditylike in nature, as chipmakers have the ability to switch between foundries when desirable. Foundries rarely have a defensible position and easily become subject to the swings of the broader industry. In times where chip demand is low, foundries can't recover the fixed costs associated with running their fabs. Yet if chip demand skyrockets and exceeds a foundry's capacity, the firm could lose business to other foundries. In this case, a firm's natural reaction is to expand, but if competition also floods the market at the same time, the foundry space suddenly flip-flops to a bout of overcapacity. In turn, foundries tend to slash prices in order to recover the cost of their new/upgraded facilities.

How to Play the Foundry Cycles
As one can guess, the semiconductor industry is in the midst of perhaps its most severe downturn ever. Many prior slowdowns, including the bursting of the tech bubble of the early 2000s, have been supply-driven as the semiconductor supply chain built up too much inventory and foundries suffered from overcapacity. Foundries have been more focused on managing their production capacity in recent years. Their reward for finally getting it right? A brutal credit crisis and chip demand falling off a cliff.

As depressing as the current environment appears, investors can still make money by playing the industry cycles. Chip demand will eventually turn around, which will lift foundries into the economic "sweet spot." Foundries have refrained from investing in capacity due to the downturn; once demand recovers, this capacity will fill up quickly. In turn, leading foundries can raise prices, as fabless chipmakers are willing to pay more in order to secure production capacity, particularly for the most advanced chips. Although chip demand can often rise and fall quickly, we don't expect this turnaround to happen any time soon, and demand might still decline further before it picks back up.

The key to playing the cycles is to take a contrarian view. Behavioral finance has introduced a concept called "recency bias," where investors tend to put greater emphasis on recent news, irrespective of the probability of that news actually having an impact on the company's future. When foundries put together several quarters of strong earnings, their prospects couldn't be rosier, and the stock price reflects this sentiment, it's very likely that the cycle is close to peaking. Similarly, when it takes a doomsday scenario to arrive at a current valuation, it might be the right time to buy. Either way, the most important consideration is that the industry is cyclical, so a "stick it in the drawer" investing strategy is the wrong way to play the semiconductor foundry space.

 

The Cast and Crew
Individual foundries remain subject to the ebbs and flows of the industry, since they can't control chip demand or the capacity decisions of competitors. However, certain firms can excel by taking technological leads, efficiently utilizing their facilities, and gaining scale advantages. By far, the strongest foundry, in our opinion, is  Taiwan Semiconductor (TSMC) (TSM). According to Gartner's April 2008 data, TSMC has 44% market share in the foundry sector, nearly triple that of its next closest rival. The company earned $10 billion in revenue in 2008, and its gross margins and earnings have far exceeded the results of its competitors. We believe the company's strong performance--and narrow economic moat--is due to its scale advantages that no other foundry can match. TSMC has the resources to invest in the latest equipment, giving the company an advantage in making cutting-edge chips. TSMC leverages its technology leads into attracting business from top chipmakers. Specializing in advanced chip production also allows the company to better maintain pricing and achieve above-average gross margins, as older semiconductors quickly become commodified and customers will only pay a premium for new and advanced chips. TSMC uses these earnings to reinvest in even more advanced equipment, and the cycle continues.

 United Microelectronics (UMC), also based in Taiwan, is the clear-cut number-two foundry, with $3 billion of revenue in 2008 and 15% market share. Chinese foundry  Semiconductor Manufacturing International (SMIC)  and Singapore-based  Chartered Semiconductor  are the other foundries with annual sales over $1 billion. The rest of the foundry field is littered with countless smaller players that either specialize in certain fabrication niches or remain satisfied with making older chips that lag in technology. While UMC, SMIC, and Chartered have scale advantages over the remainder of the industry, each firm significantly lags TSMC in operating efficiency and financial returns.

Foundries also face outside threats. Perhaps the largest comes from  Samsung (SSNLF), an integrated device manufacturer, or IDM. IDMs tend to be large firms that design their own chips, but also have the size and scale to manufacture their semiconductors in-house. Samsung has dipped its toe into the foundry space, essentially renting out some of its excess capacity to fabless chipmakers, and giving itself the ability to exit the foundry arena if it suddenly needs to apply this capacity internally. We believe Samsung looms as a significant threat to smaller foundries, as we anticipate that chipmakers will transfer business to Samsung from lesser players. Perhaps the first step in this direction came from programmable logic chipmaker  Xilinx , a previously loyal UMC customer that recently announced that it would use Samsung to make its latest set of chips. Other IDMs, such as  IBM (IBM) and  Texas Instruments (TXN), could potentially take a similar approach. Another long-term threat is the GlobalFoundries group, which was recently spun off from  Advanced Micro Devices (AMD), a firm that shifted from an IDM to a fabless model.

The Future of Foundries
Ultimately, the greatest threat to the industry is the current economic downturn and the probable shakeout of the sector where many foundries may disappear. The dramatic falloff in chip demand has made it nearly impossible for foundries to remain profitable in the current environment. Many foundries will likely be unable to break even on a gross margin basis (i.e., before R&D and SG&A costs) this quarter. Certain foundries, particularly SMIC and Chartered, also have heavy debt loads that may hamper their ability to stay in business through the downturn. We would not be surprised if these two firms, as well as many other smaller players, fell by the wayside in the current recession. Merger rumors have also been swirling amid the downturn; a somewhat far-fetched idea, in our opinion, is the pairing of UMC and Chartered. Perhaps the biggest wild card and possible lifeline for weak foundries could come from Asian governments. Countries such as Taiwan and China have shown a strategic interest in the foundry business and may rescue struggling firms, regardless of whether they deserve such a helping hand.

We believe that UMC will remain healthy enough to make it through the downturn but will likely retain its role as a second-tier foundry. The most likely winner in the foundry industry will be TSMC, as we believe they are best suited to navigate through the downturn and outpace the competition on the rebound. The company's balance sheet is sturdy enough to handle both its debt obligations and its upcoming R&D investment needs. Thus, we expect that TSMC will expand upon its technological leads--and economic moat--by improving upon its scale and continuing to be the go-to manufacturer for cutting-edge chips.

Nonetheless, we don't expect foundry industry conditions to improve any time soon. These already beaten-down stocks may get cheaper still, and further sell-offs might make for some opportunities to "buy low and sell high" down the road. However, we anticipate that the gloom and doom surrounding the industry will someday turn to glowing optimism at the height of the inevitable upturn. Semiconductors will continue to drive technological innovations, and foundries will remain an important part of the supply chain. A quality foundry like TSMC should emerge as a stronger player from the downturn, and when the chip sector finally rebounds, investors that understand the chip manufacturing cycles have a great opportunity to profit.