Ian, Interesting article, thought you might enjoy...
Close up: Semiconductor Equipment
By Gunnar Miller,
Paine Webber Inc., New York
We believe the prospect of a continued soft market in equipment spending in 1997 precludes taking aggressive positions in semiconductor-equipment stocks at present. What would get us to turn positive? In light of the strong performance of our 54-company universe in recent years (up 72 percent in 1993, up 35 percent in 1994 and up 55 percent in 1995), no one wants to be left on the platform when the train leaves the station for the next go-round. However, we would caution against a "hold your nose and buy them all" approach; equipment stocks are likely to remain the domain of the momentum investor, and levitation by value players isn't likely to be the driver of sustained upward stock-price performance.
Interim Plays
Applied Materials (AMAT: buy): Likely to see sequential margin improvements and appears to be gaining market share even as the equipment slowdown continues. Teradyne (TER: attractive): Will likely benefit from increasing chip-unit volumes, especially as it has introduced new testers for mixed signal, logic and memory. KLA Instruments (KLAC: attractive): A major provider of the in-line inspection equipment that has made better-than-expected yields possible. The following stocks are also rated attractive because they are late/counter-cyclical plays, diversified into areas outside semiconductors, and they can continue to take significant market share: CFM Technologies (CFMT), Mattson Technology (MTSN) and Ultratech Stepper (UTEK). Source: PaineWebber Inc.
Adroit investors should be handsomely rewarded over the long haul. We're not likely to see downswings close to prior five-year lows, given the substantially improved condition of the U.S. semiconductor and equipment industries; however, it's unclear whether the equipment stocks aren't already discounting a more modest and short-lived downturn than seen in previous cycles, when companies posted substantial losses. Here's what needs to happen prior to a group upgrade of these stocks; while most of these signals may seem to be "no-brainers," their early signs may not be obvious:
Sustained (multiple-month and accelerating) uptrend in year-over-year order growth for semiconductors, equipment or both. We would caution that as shipments come down, positive book-to-bill ratios alone will not likely be sufficient triggers; we're talking about the growth rate, period. There is no statistically significant macroeconomic data point that foreshadows an upturn; in the past, GDP and the Purchasing Managers' New Order Index had some predictive value, but these relationships have waned, most likely as a result of increasing globalization. However, even though we only have about five years of data, the year-over-year performance of our equipment-stock index seems to lag year-over-year semiconductor order growth. This may indicate that the market doesn't react to new equipment order information quickly, although the reverse is true for semiconductors.
Supply-and-demand imbalance tightening, with actual increases in apples-to-apples price comparisons (per bit, per gate) rather than stabilization of declines.
Evidence of semiconductor industry underspending. This will probably manifest itself in a bottoming out of worldwide capital spending as a percentage of revenues somewhere in the low 20s, just as it did in 1992. Unless a fundamental reduction is occurring, we believe the median model (25 percent of semiconductor revenues directed toward capital spending) is a good rule of thumb; it's at least as good as the benchmarks used to determine appropriate levels of R&D spending.
Bottoming out of net margins. This will likely happen with chip companies first, indicating they are again positioned for upside operating leverage when the upturn comes. The prospect of improved net incomes and cash flows will likely improve visibility for equipment companies.
Semiconductor stocks start outperforming. The correlation coefficient between year-over-year price performance of the semiconductor and equipment stocks over the past 16 years is 0.88. One cautionary note: In August 1991, Semiconductor Industry Association orders, semiconductor and equipment stock prices turned positive coincidentally. However, while semiconductor orders and stock prices continued upward, equipment orders and stock prices retrenched the following summer.
Strategy Rationale
Despite the beginnings of what later proved to be substantial static-RAM and DRAM price declines during our Asia tour in November 1995, at that time most Asian chip manufacturers planned to maintain aggressive capital spending plans in 1996. For the most part, this was the case, as full-year capital spending for Korean, Taiwanese and other Pacific Rim chipmakers will likely be up 41 percent from 1995, at $12.8 billion. However, most Pac Rim capital spending was front-loaded into the first half of the year, and Japan's calendar 1996 spending (for the fiscal year ending March 1997) looks to be off 7 percent at $11.9 billion.
Nineteen ninety-seven is likely to be different, as Pac Rim chipmakers' capital-spending plans appear to be down 17 percent (to $10.7 billion), and flat to down slightly in Japan. In the wake of our November 1996 Asia tour, we are fine-tuning our worldwide capital-spending forecast to down 8 percent from down 7 percent; although this is not a significant change, we were surprised that all but a few Asian semiconductor makers we visited, notably Macronix, Sony and Hitachi, plan to decrease spending.
We expect 1997 worldwide equipment shipments to be off 16 percent from 1996, a decline borne out by our 1997 covered-universe composite revenues (expected to drop 12 percent, with EPS down 27 percent). Our fab list identifies 126 projects slated for roughly the next five years totaling $110 billion, of which 36 (29 percent) or $34 billion (31 percent) are classified as delayed. If worldwide PC and semiconductor unit demand remain strong, what's keeping the lid on capital spending?
On our Asia tour last November, we confirmed that ramp times to "entitlement" (85-90 percent yield) and die-shrink schedules have accelerated, allowing some companies to enjoy a "phantom fab," or output equivalent to an additional fab contemplated during the last planning cycle. Equipment is also said to have better uptime, improved throughput and to allow for more seamless transitions to smaller linewidths.
The 64-Mbit DRAM will most likely be the next driver of a substantial acceleration in capacity-equipment orders, which could begin in late calendar 1997 and continue into 1998. However, it is uncertain how much of the initial 64-Mbit production will be accomplished through modifying existing facilities to 0.25-micron capability by simply adding more chemical mechanical planarization (CMP), deep-UV photolithography and high-density chemical vapor deposition (HDCVD) and etch equipment.
Leading-edge "technology buys" (small unit orders for evaluating new technologies) in 1997 are likely to continue at levels well below the "capacity-buy" (large unit orders for production) volumes equipment makers enjoyed in the 1993-95 boom period. Specifically, we're concerned that:
Investors view the nascent deep-UV stepper market as bulletproof. We believe unit shipment assumptions in 1997 are high at about 300 units, and that as Nikon and ASM Lithography introduce step-and-scan units, price competition will intensify.
High-density etch and CVD product shipments by the likes of Applied, Lam and Novellus may not offset the significant fall-off from 1996. Novellus' stock appears to have been bid up on the initial success of the Speed High-Density Plasma CVD product, but an order "air pocket" could cause a dip in early to mid-1997.
The same is true for CMP. Proper positioning at the high end will benefit leading equipment players in the next round of capacity additions, but will likely prove insufficient to offset intermediate-term weakness in aggregate equipment orders.
______________________________________________________
Regards, Michael |
