Investors in semiconductors can be forgiven for feeling testy these days. Though the chip industry has formally begun a recovery, stock prices have swooned in the past few months. The benchmark Philadelphia Stock Exchange Semiconductor Index has lost more than half its value since peaking in March.

While unit sales of chips are on the upswing, stocks have stayed under pressure as leading companies like Intel ( INTC) and AMD ( AMD) duke it out for market share. Meanwhile, major equipment suppliers like Novellus ( NVLS) and Lam Research ( LRCX) just forecast a potential drop-off in orders for the quarter under way.

But Manoj Nadkarni, engineer and founder of online investment newsletter, cautions investors against getting too tweaked about the parade of nervous earnings revisions issuing from Wall Street. He offers his own take on how to assess chip and equipment stocks, and explains why he thinks tech giant Intel could see some revenue gains in the current quarter.

1. We're technically supposed to be in the beginning of a semiconductor-industry recovery, even if it's slowed down lately. Do recoveries always look this choppy?

If we look back at what's happened in the last few months, part of it is purely bad news from some semiconductor and/or semiconductor equipment companies that's obviously hurt the stocks. But there has been positive news as well on a micro level. Many individual companies have been able to meet guidance, and one or two have exceeded it.

Mutual Fund Monday
Dear Dagen: How Full Is Your Glass? Optimistic or pessimistic? Pros talk about the economy and value stocks.
Corporate Cold Calling: Companies Try to Lure Investors With Bonds: For individuals, the drawbacks outweigh the advantages. Bond funds remain the way to go.

If you look at the equipment industry, bookings have been going up the past few months. Many equipment companies came out with stellar results for the June quarter. I think we will see some pause in the order rates, but in our estimation it's not going to be as bad as the previous downcycle -- though with the recent announcement that TSMC ( TSM) and UMC ( UMC) and STMicro ( STM) will cut capital expenditures, there's been some concern and a lot more speculation.

On top of that, superimpose all this negativity about CEOs telling the truth or not, and more and more well-known companies getting under SEC investigation. None of that helps investor confidence.

A big part of this volatility is based on expectations, especially on the part of Street analysts, and also to a large degree investors, because they follow what analysts put out.

2) What do you make of analysts' revised expectations lately?

We have a different approach to looking at the semiconductor equipment industry. Fundamentally, we think that the way the Street analysts analyze companies -- like predicting revenues and earnings a year in advance -- works well for companies like Procter & Gamble ( PG), stable companies with stable earnings.

But the equipment industry is very cyclical. Applied Materials ( AMAT) has been through three or four cycles over the last 10 years, so the cycles are -- in our mind -- normal. The CEOs themselves cannot predict reliably what their revenues and earnings will be say six months or further out. So if they cannot do that, how can somebody else do it? We don't get into predicting beyond three months or six months.

3) Do you think the Street approach contributes to the volatility in share prices?

With AMAT or KLAC ( KLAC), you have 15 to 20 analysts covering it, coming up with estimates for fiscal year 2003. And invariably they will change up and down. Then what happens is as analysts change estimates, they upgrade or downgrade the companies and then these stocks see volatility. Often we see -- not with every analyst but with some analysts -- that they will be bullish when the numbers are really good, and bearish when they're bad. But this business is going to come back.

We are going to see more and more chips produced in this world, with finer geometries. And if manufacturers don't use equipment from AMAT, they don't have a lot of choices. You can pick any big name, AMAT, Varian ( VSEA ), and there will be cycles, but through these, companies continue to grow.

So you might say, if you have these cycles, why would you want to even invest in these companies? The reason is that their business is a classic example of oligopolies. For deposition, there's Applied Materials and Novellus ( NVLS). For automation, there's Asyst ( ASYT) and Brooks-PRI Automation ( BRKS). So because of that, because demand for products is fairly inelastic, they grow rapidly and make good money during upturns. There may be 10, 15 different chipmakers, and they all have to have implanters from Varian or Axcelis ( ACLS).

Manoj M. Nadkarni

Founder and Editor,

4) Would you say, then, that the market overreacted to semiconductor manufacturers saying they'll cut their capex budgets?

So just because TSMC cuts their annual capex projection from $2.6 to $2 billion, AMAT stock comes down by 40%. But in reality the companies are not worth 40% less. That is normal for AMAT, and it will grow through these cycles.

5) You like semi equipment because it's an oligopoly, but that doesn't apply to chipmakers. Given that they're so volatile, and in some cases sell commoditized products, what's the justification for investing in these companies?

You have to think about what kind of margins they have. On the one extreme, you have Micron ( MU) and Samsung, with very elastic demand for their DRAM memory chips. It's a very highly competitive business with very cost-oriented manufacturing and to survive in that business is difficult. Only those with the best economies of scale will do okay, and even then, making money is not as easy.

At the other end of the spectrum, you have Linear Tech ( LLTC) and Maxim ( MXIM) and Xilinx ( XLNX), whose products are fairly proprietary. Especially companies like Linear Tech -- they have 35%, 40% operating margins. So that's the best of the breed in terms of how a semiconductor business can generate real income. For a long time, Intel was in the same class as Linear in terms of operating margins, but in the last two or three years, competing with AMD on price, we've been seeing their margins reduced.

Also, some of the midsize companies with their own plants still have very good margins. They have proprietary products but don't need as advanced factories, and they still make very good money, like Maxim or Microchip ( MCHP).

6) What are some of the criteria you use to determine whether to invest in a company?

You have to look at which subset they are in and what is their market share, how proprietary their products are and whether demand is elastic or inelastic.

7) You've said before that within chips, you like the PLD programmable logic devices space. Why is that?

In the PLD space, barriers to entry are fairly high. You have to have a big infrastructure, and somebody else cannot get in without building that infrastructure. That holds selling prices up. You don't see the kinds of price wars in that space that you see for memory or even graphics, where you know you have so many alternatives in the market.

Also, if you design a regular chip, it takes three or four months to see if it works. But with PLDs, you can program an experimental chip, and if it doesn't work you figure it out quickly. That helps system makers bring products to market quickly, and is a big advantage when time to market is critical.

During an upcycle, you will see these PLD companies get very good operating margins. That translates to earnings.

The analog segment is another area that's fairly robust, part of the reason being that the chips are more proprietary. Maxim and Linear are doing well -- but then valuation-wise, they're not that great, though better than before. So you trade off good profitability for valuation there.

Texas Instruments ( TXN) had sequential gains in their June quarter and I think they're guiding for around 5% in the September quarter. There's a good likelihood they'll meet it. One of the drivers is the growth in 2.5G and 3G handset applications, and the infrastructure required for implementing those standards. If you think 3G is here to stay and prosper, they they're certainly a big player in that; they make the chips that go into those handsets.

8) You mentioned valuation -- are you starting to see some good values across the industry? And what metrics do you use for figuring out valuation?

Certainly price to sales is a good starting point, and price to book value. Then when you look at price to book, what does it comprise? Does a company have a lot of cash on hand? That's the best kind of book value.

You also look at all the critical ratios -- the quick ratio, current ratio, what kinds of margins they're running at now, what they had during the up cycle. So there's no magic formula, but generally those types of comparisons are good for at least relatively judging A versus B.

Valuation-wise, practically everything is looking good. But then you have to make a judgment as to who has good long-term growth prospects, which are the market leaders and why, what are the barriers to entry.

9) Speaking of market leaders, what do you think of Intel (INTC)right now? It's been under a lot of ASP average selling price pressure lately. Do you expect that to continue, or will things turn around later this year?

The PC business is a seasonal business. Typically what happens is the December quarter is the strongest because of the holiday. In terms of unit shipments, the March quarter has less units than December, and June has less than March. June is typically the lowest point of the four quarters.

So we were expecting lower unit shipments anyway for the June quarter . Then on top of that, AMD ( AMD), Intel's primary competitor, started ramping up its .13 micron process in its Dresden plant. That output started in the April-May time frame. It certainly had more volume coming to the market during the seasonally worst quarters, so there was more pricing pressure that way.

Then we think Intel panicked if you look at their pricing strategies, especially on mobile processors. They cut prices on all their processors in May, thinking that would stimulate demand. They dropped the 1.8 Gigahertz Pentium 4 mobile processor from $637 to something like $348, and just wiped out their price spectrum.

But they've corrected that with the Pentium 4 mobile 2 Gigahertz. In the last week of June they came out with a new pricing strategy, laddering up the price. We feel fairly certain that in the third quarter, mobile ASPs will go up. Intel should be able to maintain prices at the high end, and if they're smart, ladder them up a little because AMD doesn't have anything of comparable performance until they come out with the Hammer products.

There's a good chance that Intel's ASPs will go up in the third quarter compared to the second quarter. It might be only a few percent, but we don't see any negative movement likely. So if ASPs overall go up and then the third quarter sees typical seasonal improvement with back-to-school sales, then unit-wise things will be better. Then, Intel's opening up a wider gap with AMD and should be able to hold market share, if not gain on that. The only negative is a potentially more serious economic downturn.

We think Intel will probably meet guidance and post numbers better than the second quarter.

10) So if your take on Intel is relatively positive, or at least not as bearish as some, what does that mean for the broader tech market?

Intel is like a Mack truck -- at least, that's what's happened in the past. When they do badly, there's a major shadow cast over the market. When they do well, appropriately the tech sector will also rebound. They pull the entire market. So if you take into account these factors, things aren't, in our opinion, as gloomy as some are saying.