By Thomas Smith, CFA We at Standard & Poor's Equity Research see the semiconductor industry's current cyclical expansion broadening in 2004-05. We forecast global industry sales (in dollars) to rise 30% in 2004, and 20% in 2005. And as that happens, we expect shares of Intel (INTC
; recent price, $29), the world's largest chipmaker, to outperform the S&P 500-stock index over the next 12 months. Intel carries Standard & Poor's highest investment recommendation of 5 STARS, or Buy.
The industry and Wall Street both seem to agree on a robust advance in the near term. The Semiconductor Industry Assn. (SIA) midyear forecast update, released June 9, projects global industry growth (in terms of dollar sales) of 28.6% in 2004 and 4.2% in 2005, before a decline of 0.8% in 2006. We have seen general agreement on Wall Street that 2004 should be a boom year, with growth in the 25% to 30% range.
As the divergence in our forecast vs. the SIA's projection indicates, we see a much wider range of estimates for 2005, given that some observers anticipate a quicker end to the industry upturn than others. Our view is that double-digit growth can continue into 2006. Why? Chipmakers, with the notable exception of Intel, were reluctant to spend on new plants during the recent downturn. The hesitance of industry players to build modern plants should delay the onset of a new bout of overcapacity, in our view.
BIG FISH. Also, a broader mix of electronics end markets being served by chipmakers might help avoid a precipitous collapse of demand. The fact that the industry's fortunes aren't tied so closely to a dominant product or "killer application" -- the role played by the PC in recent cycles -- may mitigate the size of the next industry downturn, in our view.
Intel has a commanding market position. In terms of overall sales, it is larger than its next three competitors -- Korea-based Samsung, Texas Instruments (TXN
; S&P ranking, 4 STARS, or accumulate; $25), and Japan-based Renesas -- put together, according to 2003 revenue data compiled by market researcher IC Insights. Intel's products include semiconductors, circuit boards, and other semiconductor components used in computers, servers, and networking and communications products. Sales of microprocessors represented 73% of Intel's total in 2003.
Its key customers include other technology giants, such as Dell (DELL), which accounted for 19% of 2003 sales, and Hewlett-Packard (HPQ), responsible for 15%. In 2003, 72% of the chipmaker's sales came from outside North America, including 40% from Asia, 23% from Europe, and 9% from Japan.
HOW IT RUNS. Since early 2004, Intel operates in two segments: Intel Architecture Business and Intel Communications Group. Architecture's products include the Intel Pentium 4 desktop processor family. In 2004, Intel began manufacturing these processors at 90-nanometer linewidths on 300mm wafers. In 2003, most microprocessors and chipsets were produced at 130 nanometer (0.13 micron) linewidths. The Intel Celeron processor family aims to serve the lower-cost end of the PC market. The Intel Pentium M processor and 855 chipset family serve the notebook PC market. The enterprise market is served by the Intel Xeon processor family and, for higher computing power, by the Intel Itanium processor.
The communications group offers wired Ethernet products for networking, various wireless connectivity products, communications infrastructure products, and microcontrollers. It also makes flash memory, which stores data even when power is switched off and is used in mobile phones, PDAs, and other portable electronics devices.
Intel continues to shift wafer manufacturing to the 300mm wafer standard, and it expects to have three plants producing with 300mm technology by the end of 2004, with another under construction. Two more should make the change after 2004. The larger wafer holds over twice as many chips. Thus, with many costs associated with processing a wafer fixed, the efficiency jump from having greater surface area on the wafer is substantial.
REVENUE WATCH. Intel's early adoption of the new wafer size, vs. the prior standard of 200mm, has helped it to be ready to face the heavy demand we expect in the projected 2004-05 industry expansion, in our view. We think that being an early adopter has created more time for Intel than some competitors to fine-tune the new processes and bring wafer-fabrication yields and efficiencies up. We see improving gross margin in recent quarters related in part to improved fab yields -- a measure of usable chips per wafer processed.
Recent Wall Street estimates called for Intel's 2004 revenue to grow 14%, below our estimate of 17.4% and the latest trade association estimate for 17.7% growth in PC-processor revenue. We believe Intel's revenue will likely surpass Street expectations in the next 12 months.
Capital spending was under $4 billion in 1998 and 1999, and then jumped to $6.7 billion in 2000 and $7.3 billion in 2001, before easing to $4.7 billion in 2002 and $3.7 billion in 2003. With the once-a-decade shift to a new wafer size now completed, we expect depreciation of $4.6 billion in 2004 to be above capital spending of $3.7 billion, and for the two to be about equal in 2005, at $4.8 billion. We believe this contributes to an attractive cash-flow story for Intel.
SETTING THE TERMS. Historically, Intel has maintained about an 80% share of the PC processor market, where it competes mostly with Advanced Micro Devices (AMD
; 3 STARS, or hold; $16). Lively competition on technology and price exists between Intel and AMD. We believe Intel's larger scale gives it an advantage in research and development, as well as in production capabilities, financing, and marketing. Scale, in our view, gives Intel the ability to respond to occasional market-share advances by AMD. The SIA puts the overall global microprocessor market at $27.4 billion in 2003, and it sees $32.3 billion for 2004.
In market-share skirmishes, Intel is generally the heavyweight that sets the industry's agenda, as it did in March, 2003, with a $300 million budget to coordinate a marketing launch of Centrino wireless technology for notebook PCs. Recently, we believe that an early move to introduce 64-bit processors has won AMD some headlines and customers, mainly in the server processors, but we view software availability to prompt rapid adoption of the 64-bit processors as lacking for now. Given time, we believe Intel can develop an adequate response.
Beyond processors, Intel is active in chipsets for various computer markets and is trying to diversify into communications, where one of its biggest products is flash memory. The flash-memory group lost market share in early 2003, as Intel successfully instituted a price hike. However, by spring 2004, the communications operations had been reorganized from two segments to one, and results have been more stable for this group so far in 2004.
SHIFT IN FOCUS. The SIA tallies the flash-memory market at $11.7 billion for 2003 and estimates the broad flash category will grow by 49%, to $17.5 billion, in 2004, driven by demand for portable electronics applications that require flash memories, including wireless phones, PDAs, and notebook PCs with wireless functionality.
We view the challenge of diversification into the communications segment, where the markets are more fractured than in the PC arena, as a central one for Intel's long-term growth. In our view, Intel has a good opportunity to compete well in communications, but it faces a tougher situation than in PCs. Over our 12-month time horizon, however, we think the key driver for Intel shares will be the company's performance in PC processors, given the large share of sales that computing still represents.
We believe the shares are attractively valued compared to Intel's growth opportunities, especially considering its good long-term record of profitability despite the strongly cyclical nature of the industry. Intel made money through the recent downturn, which was the industry's toughest, and its
return on equity (ROE) has topped 30% in boom years. Debt-to-equity is modest, near 2.4%, and Intel has about $13 billion in cash and equivalents, which we believe offers adequate financing for expansion.
ROOM TO RUN. We looked at potential price appreciation for the stock from a number of viewpoints. The shares recently traded near 23 times our 2004 earnings per share (EPS) estimate of $1.27, which is at premium to the multiple for the S&P 500 but attractive vs. many of Intel's peers. Applying a slightly higher peer-group multiple of 25 to our 2005 EPS estimate of $1.70, we see potential price appreciation to $42 over the next 12 months.
By applying another measure, the average annual historical high price to tangible book ratio of eight times the present tangible book value of $5.36, we get a potential price appreciation to $43.
The shares presently sell at about six times trailing 12-month sales, which is in the middle of a wide range for chipmakers. Applying the present price-to-sales multiple to our forward 12-month sales estimate of $5.62 indicates price potential of $34. However, we believe it's likely that multiple could expand to seven times if sales come in above Street expectations, as we project. This would indicate price potential above $39.
Overall, given our belief that the chip industry cycle has legs through 2005 and our expectation of strong sales for Intel in the second half of 2004 based on seasonality and a need for businesses to replace aging PCs, we recommend buying Intel shares. We blended our valuation assessments detailed above to arrive at a 12-month target price of $42.
CHARTING DEMAND. Our intrinsic value for Intel shares based on our
discounted cash-flow analysis is $33.25. This indicates the shares are selling at a discount to their intrinsic value, by our calculations.
The main difference between our operating EPS estimates of $1.27 for 2004 and $1.70 for 2005 and our respective Standard & Poor's Core Earnings estimates of $1.07 and $1.50 is stock-option expense that we estimate at 20 cents per share for both years. Intel may have gains or losses on investment or asset sales that could affect the S&P Core Earnings calculations, but for estimating purposes, we assume these have zero net effect. In our opinion, Intel's reliance on stock-based compensation is higher than most companies, but in line with its chipmaker peers.
The primary risks to our recommendation and target price revolve around fluctuations in demand for and supply of semiconductors. Demand is influenced by global economic growth patterns and the presence or absence of consumer and business need for new electronics applications. Our estimates of economic growth and demand for key drivers such as PC and cell-phone sales could prove incorrect.
PLANNING PROBLEMS. Supply is affected by industry capacity to produce semiconductors. At present, chip capacity utilization is above 90%, which we believe is contributing to an upward trend in average selling prices. But this intense level of plant utilization might fade sooner than we estimate. Historically, when a wave of new wafer plants comes on line in an attempt to raise supply to match demand, the new supply tends to surpass demand, which dampens average selling prices of semiconductors.
Demand and supply imbalances have been a key factor in what has historically been about a four-year boom-bust cycle for the semiconductor industry. It takes almost two years to build a new wafer plant, and we believe this long lag time in bringing on new capacity makes planning for a smooth match of supply and demand difficult.
Predictions by others and by S&P for the timing of phases of semiconductor industry expansion or contraction may prove incorrect. One reason is that inventory levels at original equipment manufacturers and distributors are often difficult to gauge, masking the pace of ultimate consumer demand.
Additional risk factors include competition, the usual risks of financing and operating wafer plants, the risks of operating internationally, above-average share-price volatility, and reliance on stock-based compensation. Analyst Smith follows semiconductor stocks for Standards & Poor's Equity Research Services