Hi divvie; My specialty is hardware, and like Clint said, a A man's got to know his limitations. Given that, here's why I don't think that the server business is going to rescue a lot of box makers from the collapse in desktop ASPs.
Lets analyze the (hardware) cost of a server... (Mind you, some of this is guess work, but I'm known to be pretty good at this sort of analysis...)
Most of a server costs fall into the following four areas:
(1) Processor
(2) Net interface
(3) DRAM
(4) Disk drives
I expect integration to severely reduce the costs of the first two items. I've posted numerous links to show what is happening in processor design. The network interfaces are undergoing a similar, parallel, integration revolution, but I've not posted links to those articles. I will do so in the future...
DRAM and Disk drive costs seem to be going down, but there is no reason for me to believe that we are about to see a revolutionary decrease in the costs of those components.
So I expect to see the ASPs of servers end up being supported (to use a technical analysis term) by the costs of DRAM and disk drive. This will prevent the systems from diving anywhere near as low as desktops can go, except for low-end servers. Rudedog calls these cheap desktops "thin." I prefer to think of them as sufficient. Sufficient, that is, to get the work done that business needs to get done. There was a time when processing power was kept in a main frame computer, and everybody had low power machines on their desk. I don't see a return to that time. Instead, I see more distributed processors, but what we will call a "thin" processor a couple of years from now, would be a high end processor today.
But the price of processors suitable for server use is going to make a huge drop in 18 to 24 months. I would expect that to open up the server hardware supplier business to companies that currently only put together low end machines. The amount of engineering needed to be done by the box maker is going to reduce again, so I would expect to see higher competition in that area.
My comments on vertical integration, posted in numerous links to this thread, would apply mainly to the desktop market in the near term. Eventually, the server market will also be vertically integrated, as is every other technical commodity industry. But the server market definitely has more legs.
Now, looking at the computer industry as a whole, it is obvious to a lot of us that the desktop market is not going to be a healthy one in the next couple years. I would expect to see a lot of companies move into the high end market as fast as they can. This is normal business behavior, but there will not be room for all those companies in the high end.
As to who ends up ruling the high end, I don't know. In many ways, from the point of view of an investor, it really doesn't matter. All these companies have booked passage on the Titanic. Some of them may have already reserved a seat in a life boat. Great for them. But the fact is that nobody is going to get to New York on time. Right now, DELL is trading at a high multiple relative to what it (and the other box makers) has historically traded at. I am sure that DELL will survive the collision with the iceberg. I bet that DELL will be a great bargain sometime after Wall Street has noticed the ice on the deck. But I don't think the time to buy DELL and hold for the long term has arrived.
My predictions for what is due to happen in the PC industry would be avoided if someone could come up with a reason why the current crop of PCs are going to be severely underpowered soon. I just don't see the next "killer ap." Who knows, maybe it's out there. But if it is, why are we seeing a bunch of sub $1000 processors now? One of my friends wants to update the computer he has used almost daily now for a long, long time. It's an 80486. A lot of what has improved in computer performance since then is useful only for viewing porn sites on the internet, and this is not what business wants to be paying for. The real question is how much machine do you need in order to maximize employee productivity. For most employees, I think we have that machine now, and it is under $1000.
-- Carl
P.S. Did you see the link I posted to where IBM is putting four processors on a chip along with memory and I/O interfaces? It was in the past couple weeks on this thread. Reading that link, it is clear that the processor performance/cost revolution is not going to stop at some arbitrary boundary like Desktop versus Server. |
