Confusing Activity and Results

It may seem like the markets have been making great progress despite all the geopolitical dislocations and the weak macroeconomic backdrop, but when you get right down to the numbers, stocks in the semiconductor and semiconductor equipment sector have been treading water.

What should an investor do?

I think it is very important during these volatile times not to confuse activity with results. It seems to me as though we have plenty of time to apply our capital to the markets. I have written that the long-awaited recovery in the semiconductor business will not materialize overnight. A number of positive events need to take place before a really positive move begins.

As we move into the earnings reporting season, I suspect we are going to hear about flat to weaker business conditions. On the good side of the ledger, it would not be surprising to hear some companies talk about higher quote activity. Fabs are planning for the next ramp — they just need that event combination to take root before shelling out money. Again, this is not going to happen in a heartbeat.

There are several reasons why I believe this will be a gradual recovery. First, the pace of device shrinks and the move to 300 mm appears to be slowing. Of course, there are exceptions, but in general I'm sure you'll agree with the many articles in this publication and others as to the difficulties manufacturers have been having with sub-0.13 µm device production. Don't get me wrong — innovation and new advances are taking place. It's just taking more time.

In my opinion, the No. 1 hurdle facing those who wish to work at the leading edge is cost. The cost burden is not only faced by large IDMs and foundries — fabless chipmakers must grapple with high price tags. I've been saying for a while that higher costs would eventually separate the men from the boys. That appears to be happening today.

One segment that I am particularly worried about is the fabless design house. Many believe fabless is the way to go, but I sense this thinking will change over the next few years. A number of hurdles are cropping up. The cost to get a device into the market is at the top of the list.

An estimate of the cost to bring a new design to market was provided by one of the guest speakers at last month's Semico Summit. Basically, a fabless semiconductor house must fund $30M in R&D to "prepare" a leading-edge device for production (some estimates show this as high as $50M). Remember, this is only R&D. There are many more costs to be absorbed before the device hits the street. If we build a business model where R&D runs close to 20% of total revenue, we find that $150M in revenue from this single design will be required to justify this cost. For the sake of argument, let's say our device is really good and we grab 10% of a market. To generate $150M means we are addressing a market that is $1.5B. How many addressable markets are more than $1.5B in size?

I've glossed over some items here, but I think the message is rather clear. The stakes to play the game are going higher, and eventually they will force a reduction in the number of companies in the business.