The senior analyst has material positions in INTC and AMD, and does not have material positions in KLIC, AMAT, FEIC, KLAC, LRCX, NVLS, RTEC, or TER.‡
SEMICONDUCTOR CAPITAL EQUIPMENT
We Could Be Seeing The Early Signs Of A Cyclical Recovery
Shekhar Pramanick, Ph.D., Fayad Abbasi
As we noted in our May 25, 2001, First Call note entitled “Upgrading Semiconductor Equipment Group To Strong Buy As New Cycle Could Be On The Horizon,” we believe that there is a four-step logical progression for a semiconductor cycle recovery:
1) end-market stabilization,
2) a semiconductor revenue-trend reversal,
3) semi-equipment order recovery, and
4) the emergence of a new semi cycle.
We believe that we are now seeing the first signs of a semiconductor revenue-trend reversal, with positive datapoints from the semi end-markets.Although these remain early datapoints, we believe they could signal the beginning of a new semiconductor upcycle.
First, Semiconductor End-Market Datapoints Are Clearly Positive, In Our View. Over the past few months, semi end-market datapoints have clearly been positive:
Packaging House Unit Improvements For Several Months. We have seen unit improvements from several of the packaging houses over the past several months, including ASE Inc. (five consecutive months of improving chip units, including a roughly 11% sequential increase in November from October), and Siliconware Precision (five consecutive months of improving chip units, including about 2% sequential improvement in November from October). In addition, both Amkor and ASAT have had positive comments about chip unit improvements recently.
We Have Started To See The First Signs Of Wire Bonder Orders. Kulicke & Soffa Industries (KLIC—$18.10; rated Hold and high risk) recently received orders for about 230 wire bonders from Siliconware Precision and Advanced Semiconductor Engineering (ASE). What is notable about these orders is that we believe they are for additional capacity. Wire bonders sold last year (8028-S), when run at slower speeds, are capable of pad pitches below 50-micron, based on our checks. The new wire bonders (8028-PPS) are more efficient at those fine pad pitches. We should also note that Kulicke & Soffa was one of the first companies to see the slowdown (in the May/June 2000 timeframe), and wire bonder orders have historically been a good leading indicator. Based on our checks, wire bonder upgrade at ChipPac could be imminent. Also, ASM Pacific, a division of ASM International, received orders for its fine pad pitch wire bonders from four companies.
Revenues At The Wafer Foundries Have Been Modestly Improving. Both Taiwan Semiconductor Manufacturing (TSMC) and United Microelectronics Corporation (UMC) recently reported sequential increases in revenues, with TSMC seeing monthly revenues up 7% in November over October (the fifth month of sequential improvements) while UMC saw monthly revenues up 2% in November (the fourth month of sequential improvements). More
notably, TSMC guided December-quarter utilization for its leading edge (0.18-micron and below) of 80%-plus, a substantial improvement over the September quarter. We are also seeing improving wafer starts through the month of December (and the company suggesting January could be up over December), with wafer outs expected in the January/February timeframe. TSMC is already stating March-quarter revenues could be up 10% from December-quarter levels. UMC is also expecting March-quarter revenues to be relatively flat versus the December
quarter, which is better than traditional seasonal trends.
Semi Company Order Books Are Incrementally Better, With Revenue Declines Slowing. Our checks across semiconductor companies suggest that semiconductor device bookings may have already bottomed in the September quarter. Most companies are seeing turns business incrementally improving as well as bookings. This is also reflected in actual chip units’ coming in higher than forecast at the packaging houses.
DRAM Pricing Improvement. In the past few weeks, we have seen dramatic increases in DRAM pricing in the spot market, with mainstream 128Mb spot pricing up 80% and 256Mb pricing up 70%. If current price improvements sustain, we believe this bodes well for capital-spending plans for DRAM manufacturers, as they have seriously curtailed spending plans owing to the extremely low pricing environment for DRAMs. Recently, Hynix even raised contract pricing for the first time in about a year, based on our checks. Our anecdotal checks show that DRAM loadings for new PCs have 256Mb with the introduction of Windows XP, almost double the memory content of a few months ago, whereas bit growth from technology transition such as 0.18-micron to 0.15-micron leads to about 40% bit growth (assuming no capacity increases).
New Product And Device Introductions. From a timing perspective, we are currently in the “sweet spot” of new product transition, such as Windows XP, game consoles (e.g.,Microsoft’s Xbox and Nintendo’s Gamecube), new GPRS handsets, Pentium 4-based PCs, and others. Also, these new chips have significantly higher transistor counts, such as the Pentium 4 (42M transistors), nVidia GeForce 3 (57M transistors), GPRS baseband processor (25M-plus transistors), and Xilinx Virtex II (200M-plus transistors).
We believe: 1) chip inventory levels have been worked down to the point where system companies are again ordering semiconductors; 2) we are seeing traditional seasonal builds; and
3) new feature size transitions are coming in, particularly with TSMC’s 0.18-micron and below capacity utilization up significantly. While there is some risk regarding sell-through of these builds, we believe there are indications that this is more than just a seasonal build. Some of the incremental strength is coming from products not necessarily associated with the Christmas holiday season, such as enterprise networking and modest commercial PC strength.
All This Appears To Be Leading To Tight Capacity Utilization At The Leading Edge (0.18-Micron And Below). We see this from the high utilization rates reported by TSMC, as well as our checks with Intel and Advanced Micro Devices (AMD). We believe that this could result in tight capacity at 0.18-micron and below, and if utilization rates at the leading edge remain high through first-quarter 2002, we could begin to see orders by the March/April timeframe.
Also, we think that this downturn is resulting in a large number of 6” MOS fab closures. We believe there have been about 10-14 closures year to date in 2001 (2 at AMD, STMicro, and others), with more expected in 2002 as companies’ 8” fabs increasingly surpass 6” as a percentage of the total (6” fabs currently account for about 35%-40% of wafer starts) and we begin to see a ramp in early 12” fabs.
…And Semi Equipment Orders Could Have Bottomed… North American equipment orders, as tracked by Semiconductor Equipment and Materials International (SEMI), have now been relatively flat for the past seven months (between about $650 million and $770 million). We believe that orders could have bottomed at these levels. However, until global semiconductor revenues turn, we believe it is unlikely we will see a sustainable order recovery. With the incrementally positive datapoints from the packaging houses and wafer foundries, we believe that a sustainable order recovery could begin in the March/April timeframe barring any shocks to the macroeconomic outlook.
…But We Need To See A Turn In Global Semiconductor Revenues Before Semiconductor Companies Can Commit To Capital-Spending Plans, In Our View. Our initial view on 2002 capital-spending plans, based upon a bottom-up analysis, is for spending to be down 20% from forecast 2001 levels of about $35 billion. We believe that semiconductor manufacturers are basing 2002 spending plans on their current run-rates given the poor visibility. However, as demand picks up, we believe capital-spending plans could increase over the course of the year
.
From a top-down view, we believe that actual capital spending in 2002 could be closer to down 5% from 2001 levels. This could lead to a strong order recovery rate for second-half 2002. This presumes an improvement in the macroeconomic environment, along with the resolution of current geopolitical issues. We believe that with both the fiscal and monetary actions taken, we could see improvements in 2002.
Looking at North American orders (as published by SEMI), we believe that we could see orders increase by more than 20% next year to approximately $14 billion from about $11 billion estimated for 2001, with the bulk of orders coming in second-half 2002.
We Believe Investors Should Be Biased Toward Owning Semi Equipment Stocks. In our view, investors should be biased toward owning semi equipment stocks. Our semiconductor equipment stocks are currently trading at about 5-7 times our calendar-2002 revenue estimates following the recent rally. While some may argue that the stocks appear expensive, we would like to point out that historically, the semi equipment stocks have traded at 7-11 times peak-year revenues. As such, while there could be some near-term risks to the stocks owing to lack of a near-term fundamental catalyst, we believe that the 12-month appreciation potential is attractive given that we are at a cycle bottom. We would recommend investors use the weakness to buy the stocks. We highlight our Buy-rated stocks: Applied Materials (AMAT— $44.87; high risk), FEI Company (FEIC—$33.50; high risk), KLA-Tencor Corp. (KLAC—$56.31; high risk), Lam Research (LRCX—$24.85; high risk), Novellus Systems (NVLS—$43.81; high risk), Rudolph Technologies (RTEC—$39.24; high risk), and Teradyne Inc.(TER—$31.20; high risk). |
