Semiconductor Slump Expected to Rebound in 1999

-- Semiconductor International, 11/1/1998

This year will mark the beginning of the semiconductor industry's recovery from its 'Asian flu'' affliction, although the equipment sector will probably remain wobbly for another year. This is according to two senior managing directors with NationsBanc Montgomery Securities (NMS), (San Francisco, Calif.): Brett Hodess and Jonathan Joseph. The financial analysts shared their views at the 25th Semiconductor Equipment and Materials International (SEMI) annual dinner and award ceremony recently held in San Jose, Calif.

Joseph, the semiconductor expert for the financial organization, believes several essential indicators - microprocessor revenues and improving DRAM prices - point to an upturn in the chip market. Another factor Joseph considers encouraging is the continued heavy demand for electronics products and chips.

His colleague, Brett Hodess, was not as optimistic, however. The equipment market analyst said current overcapacity issues and continued delays in capital expenditures make it unlikely the equipment sector will see much of a recovery until at least the second half of next year.

'Most analysts are saying that next year will be flat, but I think that is unlikely. I expect to see a decline of 10% to 20% in 1999,'' he said.

Hodess indicated this was the direct result of an oversupply of chip capacity, rather than a slip in silicon demand. He pointed out that bit growth has remained high and that the industry is still absorbing excess production capacity. However, there is light at the end of the tunnel. Several companies are expected to begin deploying major technology transitions next year, all of which will probably lead to capital-spending expenditures. Some of the major chip firms, including Intel and Texas Instruments will begin producing chips at smaller line widths, along with foundries Taiwan Semiconductor Manufacturing and United Microelectronics. IBM is also scheduled to roll out it copper-processing technology in volume.

This is expected to set the stage for the beginning of the equipment industry's recovery, which will be sparked by need for new technology rather than for more chip capacity. 'Most of the large orders we'll see next year will be for technology, not capacity,'' he said. 'But we have another year of pain ahead of us.''

Stanley Myers, president of SEMI, agreed, stating there will probably be no upturn in the equipment market for another 12-15 months, and that the overall economy is an unpredictable factor.

Slightly more promising is the chip market, which is starting to pull out of the slump. Although current estimates show the segment shrinking by more than 10% this year, Joseph said the industry is bottoming out and starting to head upward again.

He pointed to recent signs from Intel that suggest the microprocessor unit buying cycle has been jump-started. Intel announced that its third quarter revenues will be higher than originally expected.

'Microprocessors could be the engine that pulls the semiconductor train out of this very dark and gloomy valley,'' Joseph said.

The NMS analyst also reported that both the spot and contract DRAM markets have shown firming prices - although not yet increasing prices - as well as increasing demand for memory bits. And, although PC demand is slower than in the past, he predicted a healthy 13% PC growth rate this year and a slightly higher rate in 1999.

Joseph quoted SIA figures indicating the semiconductor industry has declined about 17% since July, which can be viewed as a downtrend, since in June it declined about 14%.

Fig. 1. Ongoing overcapacity issues and continued capital expenditure delays will put a hold on the semiconductor equipment sector's recovery until the second half of 1999. Some analysts expect a decline of 10% to 20% in 1999. (Source: SEMI, NationalBanc Montgomery Securities)

'The good news is that microprocessors have begun to re-accelerate upward, still declining, but they have done a U-turn,'' Joseph said. 'We think that given the inventory rebuilding that Intel started and the seasonal nature of the microprocessor business, that this engine could begin to lead the semiconductor train out of this deepest and darkest valley that it has ever visited certainly in over a decade.''

Quoting SIA data, Joseph indicated he sees positive signs, with bit growth re-accelerating in the DRAM sector to about 83% in DRAMs in July, a little better from slightly over 72% last May.

According to Joseph, 'The anecdotal information we get from the industry is not yet positive, but not as negative. I talked to Cypress Semiconductor, and their outlook has brightened. Whereas last quarter, entering the September quarter they had only about 70% of the quarter booked, going into this new quarter in December, they should have about 75% booked. Not a dramatic change but, again, a dip toward the positive side.''

Commerce Department, IDC and Dataquest data show PC demand has remained reasonably strong, not flat. In 1998 it has been in the 14% to 15% range, and the same is expected for 1999. Although this is down from the heady growth of 30% plus of the mid-1990s, it is not disastrous. Continuing growth and a cut in capital spending are viewed as a prescription for recovery.

An added factor in the potential recovery of the U.S.-based industry is the dramatic drop in production by Japanese and Korean manufacturers. Joseph indicated that with the Asian economic crisis continuing, neither Japanese nor Korean manufacturers will be influential players in the semiconductor market in the coming years.

The analyst indicated that for too long a time the math of the industry did not add up, and a dose of reality was necessary. 'Things have been deteriorating for three years and getting worse,'' Joseph said. 'MEMC has negative gross margins and only $16 million in cash; we forecast the DRAM industry will generate $12 billion in revenue and have $11 billion in losses this year. Things just cannot continue like this. But we are heading toward a healthier environment. For the first time in a long time we are in a capacity-absorption mode, rather than in a capacity-excess mode. It is slow and will take time. However, we think that on the demand side; on the semiconductor side, we beginning to pass that bottom.''

'Orders in the equipment industry that had been weak all year are getting weaker.'' Hodess added that most of the companies his organization has been talking to indicated their orders are weaker than anticipated thus far in the quarter, and that it is likely that another round of estimate reductions for the December quarter will be necessary, although they are not expected to be as drastic as some of the cuts made earlier in the year.

Hodess believes most of the market research firms that predict 1999 being a 'flat year,'' up perhaps 2% to 4% from 1998, need to take another look. 'Most companies think they're going to get a sustainable run rate basis by about year's end, based on what their customers are telling them. But if the run rate that they are going to hit is about where they are right now, the second half of next year would have to be up 80-100% from the second half of this year, in order for them to have an up year in 1999. It is pretty likely that 1999 is going to be down between 10 and 20%, unless demand rises extremely strongly next year.''

The good side of this, according to Hodess, is that he foresees a pick up during the second half of 1999, with sequential improvements starting to take place in the industry.

'We see some clear indications,'' the analyst indicated. 'Intel is talking about several new products; Cascade and others are saying they will come out on 0.18 µm next year. TSMC and UMC are going to be prototyping 0.18 µm for boundary customers next year, and IBM is talking about moving its copper process to 0.18 µm in 1999. TI has announced not only 0.18 but 0.10 µm for 2000 for some of their mixed-signal products.''

Hodess stated this indicates that the top half-dozen logic-related companies will begin increasing capital spending about mid-1999.

Even if Japan and Korea stay weak, that will still trigger the beginning of the recovery the second half of next year, gaining momentum in 2000.

Hodess tempered this prediction, by pointing out that even this expected recovery will still leave the industry with 10% to 20% decline in total capital spending for this year and all of 1999. 'We have never had two double-digit decline years in capital spending. In the mid-1980s and early 1990s we had one or two years with declines, but it was a few percent. We're seeing a real capital spending correction that will result in a tremendous amount of capacity absorption during 1999, setting us up for a decent recovery in 2000.''

Hodess indicated he believed that the next 12-18 months would see an unprecedented amount of retrofitting. 'People will take fabs built in the last couple of years and change the equipment set to more advanced processes. Novellus is presently talking to about half a dozen companies about retrofitting copper to existing fabs.''