Semiconductors: Up From the Bottom

With plenty of signs the industry is taking baby steps toward a recovery, S&P sees some good buying opportunities

By Thomas Smith, CFA

The semiconductor sector has nearly finished tracing out its market bottom after the deepest cyclical downturn in its history. A slow, stuttering rebound in revenues has begun in 2002.

With the first stirrings of an industry recovery, which semiconductor subsectors might be most attractive for the bigger expansion we see in 2003-05? And which are the best-positioned stocks in those groups? More about those later. First, let's look at the indicators pointing to a recovery for the chip industry:

Chip Sales. The pace of sales, as tracked by the Semiconductor Industry Assn. (SIA), shows stabilization at low levels after a sharp decline in the first half of 2001. Worldwide semiconductor sales totaled $43.2 billion in the first quarter of 2001, $34.6 billion in the second, $30.6 billion in the third, and $30.5 billion in the fourth. More narrowly speaking, "integrated circuit" sales actually ticked up in the fourth quarter, while a smaller category of simpler circuits slipped, including discrete, optoelectronic, and sensor chips.

For 2002, my expectation is that first-quarter sales may also be low, near $30 billion, but with chip inventories clearing out and the economy picking up, sales should creep higher later in the year. Overall, sales for the year could grow 5% to 10% from the depressed level of 2001. I expect a stronger rebound of 20% in both 2003 and 2004.

Chip Equipment Orders. The book-to-bill ratio for North American-based makers of semiconductor capital equipment bottomed back in April, 2001, at 0.44. The preliminary book-to-bill ratio for February, 2002, is 0.87. Ratios below 1.0 indicate a contracting industry where bookings (orders) are less than billings (sales). So sales of chipmaking gear are still soft, but they have seen an uptick from a low base.

The low this cycle for orders appears to have occurred in November, 2001, while the low for sales may have been hit in January, 2002. However, this interpretation is more tentative. When orders for chip equipment accelerate, investor interest in the whole electronics ecosystem -- from companies that make chip design software to the producers of end-products like computer hardware and consumer electronics -- can perk up in a hurry.

Chip Inventories. A hallmark of the 2001 downturn was large inventory overhang in all chip categories. Of course, large inventories kill chip sales because customers draw from stockpiles rather than buying newly minted chips. By now, some chip categories -- such as digital signal processors (DSP) chips, microcontrollers, and high-end analog chips -- have already seen normalization of inventory levels. Programmable logic devices (PLD) chips should be clearing up this spring.

Other categories, such as processors and memories for personal computers, are touch-and-go, with a rush at Christmas and a subsequent seasonal relapse in the first quarter. Heavy inventories of chips for wireline communications still exist and seem likely to remain a bearish fact of life for most of 2002. Overall, though, chip inventories are a fading problem the deeper we get into 2002.

Plant Utilization. The rate of usage of chipmaking plants, known as wafer fabrication plants or "fabs," is on the way up from very low levels. Fab utilization as measured by SIA's statistical group showed cycle peak capacity utilization at 96.4% in the third quarter of 2000.

The apparent low for the 2001 downturn was set in the third quarter, at 64.2%. Utilization rose to 65.9% in the fourth quarter. The uptick was more pronounced at the newer plants making advanced chips (with circuit linewidths smaller than 0.2 micron), where utilization jumped from 79% in the third quarter to 83.5% in the fourth.

Industry Consolidation. In a downturn, there usually is plenty of merger and acquisition action as healthy companies gain market share by buying out stragglers at attractive prices. This is evident in Micron Technology's (MU ; S&P ranking 3 STARS, or hold) December, 2001, acquisition of Toshiba's DRAM operations and its present talks to acquire Hynix Semiconductor's DRAM operations (a $4 billion deal is still under discussion at this writing).

There also are many examples outside of DRAM. Hitachi and Mitsubishi Electric announced on Mar. 18 that they plan to merge their system LSI (large-scale integration) operations, which include microcontroller, logic, analog, and discrete categories of semiconductors. In November, 2001, Vishay Intertechnology (VSH ; 4 STARS, or accumulate) acquired General Semiconductor, allowing it to claim the No. 1 position worldwide in diodes and rectifiers and No. 2 in discretes.

Conference Calls. The tone of the January earnings calls and the midquarter company conference calls has been one of guarded optimism. Most chipmakers outside the wireline telecom markets can talk about the third or fourth quarters of 2001 as being the low point of the revenue cycle. They cautiously give guidance only one or two quarters ahead, typically for flat to slightly rising sequential revenues.

Earnings visibility is limited because the order backlog is typically low, so sales depend on so-called turns business (orders received and filled in the same quarter), which is hard to predict. Judging from midquarter calls, turns orders at most chipmakers are in line with expectations.

Industry leader Intel on Mar. 7 was able to narrow its guidance with a positive spin on gross margin improvement. However, some perhaps overly eager observers appeared disappointed that there would probably be no upside surprise in the first quarter, and Intel shares slid back a bit. (I think Intel will ultimately do well in the coming upcycle, but I expect no surge in PC sales until 2003-04 and so rate the stock 3 STARS.)

Microchip Technology (MCHP ; 5 STARS, or buy) reported that its recovery was on a steady course and suggested commodity EEPROM memories -- a specialty memory often used with microcontrollers to create an embedded control system for a variety of applications -- was firming, which was a positive surprise. Xilinx (XLNX ; 4 STARS) reported good progress on inventory clearing and triggered a flurry of excitement in the PLD area, which includes rival Altera (ALTR ; 4 STARS).

On the downside, warnings out of Nortel, Lucent, and other customers for wireline communications chips dampened hopes for the likes of Applied Micro Circuits (AMCC ; 2 STARS, or avoid) and PMC-Sierra (PMCS ; 2 STARS).

Triquint Semiconductor's (TQNT ; 2 STARS) midquarter update on Mar. 19 illustrated the divergent fortunes of different chip segments. Triquint's book-to-bill rose above 1.0 for wireless chips, implying a budding comeback in that segment. But the market for its optical networking chips remains weak, which will restrain overall revenue expectations. Its first-quarter 2002 sales should be flat vs. last year's fourth quarter and should remain little changed in the second quarter.

Overall, the tone of the calls indicates an industry bottom has already formed, with a slow, hesitant recovery brewing in 2002 and better prospects for 2003-05.

Where to Look Now. So, we've recognized the key signs of the chip recovery. Which groups within the industry are best positioned to capitalize on the rebound?

Standard & Poor's favors chipmakers that are already fairly well along toward recovery and seem less prone to revenue relapse. Accordingly, we view the high-end analog chipmakers as an attractive group. Analog is more of a custom business than a commodity business, and analog chips are subject to less pricing pressure than, say, PC processors or DRAM.

The analog chipmakers we like include Linear Technology (LLTC ; 5 STARS), Maxim Integrated Products (MXIM ; 4 STARS), Analog Devices (ADI ; 4 STARS), and Texas Instruments (TXN ; 4 STARS).

Analog companies serve diverse markets, and the pure plays (including Linear and Maxim) typically have wide profit margins by industry standards, which contributes to relatively stable results compared with other chipmakers. In a year such as 2002, with some uncertainty about the economy, I believe these are among the steadier models in the often-volatile chip sector.

One caveat about the analog group is that the valuations are high, but then these companies are usually valued highly. Although it may be tough to have to pay up for a quality company, the multiyear upturn I see ahead should generate enough revenue growth to make some exposure to the analog area fruitful.

In considering more beaten-up chipmakers -- with perhaps more chance for an upside expansion for their valuations in the next 6 to 12 months -- I think LSI Logic (LSI ; 5 STARS) may be a good selection. It has exposure to relatively stable business in the consumer electronics and data storage areas and a chance for growth in various communications markets as the cycle unfolds.

Analyst Smith follows semiconductor stocks for Standard & Poor's

    Before it's here, it's on the Bloomberg Terminal.