Three Restless Riders: OS Poker in the Club Car

by John Martellaro

March 26, 1999

"If you resist change, you die. If you adapt to it, you survive. And if you cause it, you lead. Of course, if you cause change, you can die too, but it's a more exciting death."

- Ray Noorda

I think it's possible to analyze why.

Penny a Point, Ain't No One Keepin' Score

One of the things that corporate managers do for any new proposal is assess the risk, evaluate the resources, and try to quantify the upside and downside. Let's do that for Apple Computer.

Assumption #1. Apple is a hardware company.

Assumption #2. About 95% of Apple's revenue comes from hardware sales.

Assumption #3. Mac OS Sales comprise well less than 5% and Mac OS X Server comprises about 0% of Apple's revenue.

Assumption #4. Mac OS X Server for Intel would generate negligible revenue and generate Intel-compatible PC sales.

Conclusion: Apple gains next to nothing in revenue but would showcase the hardware capabilities of Intel-compatible PCs - damaging their own marketing for the PowerPC.

Discussion. It is well known that Linux runs rings around Windows NT on Intel-compatible hardware. Clearly it cuts both ways. Mac OS X Server on Intel would look great -- but so does the hardware. Is it worth the minuscule sales to assist in showcasing how fast a 500 MHz Pentium III can really be?

Riding the Magic Carpet...

It would appear that the obvious strategy is to cancel Mac OS X Server on Intel. But there is another solution. Apple could sell Intel-based computers and preload Mac OS X Server for Intel.

Assumption #1. Apple and Microsoft are in some semblance of détente.

Assumption #2. Mac OS X Server on Intel would embarrass Windows NT.

Assumption #3. Microsoft would be displeased about this.

Assumption #4. Knowing Microsoft's business practices, they would probably put some kind of uncomfortable pressure on Intel if they struck a deal with Apple to supply, say, 100,000 CPUs per year loaded with Mac OS X Server.

Conclusion: Microsoft is still too powerful for Apple to strike a deal with Intel.

Discussion. Apple probably has to wait for a weakened Microsoft. Events that could weaken Microsoft are: 1) Further blunders with Java, 2) An unfavorable court ruling, or settlement, in the US DOJ antitrust trial, 3) Problems with the rollout of Windows 2000, 4) Y2K lawsuits, 5) Major defections by corporations to Linux, 6) an NT system security disaster at some U.S. Government agency of epic proportions.

If one or more of these events weaken Microsoft to the point where Intel feels emboldened to make a foray into new OSes, then Apple could profit by selling Intel PCs with Mac OS X Server for Intel.

Half Way Home, We'll Be There By Mornin'

So the three restless riders continue into 1999, dealing card games and keeping their poker faces.

Are there any other considerations? A few perhaps.

Apple, in the long run, has no specific reason to avoid Intel hardware. Whatever sells puts money in the bank for Apple. Prior to Mac OS X Server, Apple worked on a project code-named Star Trek which was, apparently, a port of the legacy Mac OS to PC hardware. The port was successful, but the work involved in creating drivers was considered too horrendous. That's somewhat of a consideration as well for a server system. Even though a server is supposed to sit on a desk in a locked room, there are always a few cybergeeks who want to attach some gizmo, sit at the console, and steal CPU cycles away from Apache.

I suppose that with the Mac OS X Server being Open Source, there might be enough people out there who'd write drivers just for fun and fame, as is the case with Linux.

• Is the Open Source decision a lesson learned from the Star Trek project??
• Will Mac OS X Desktop also be Open Source?

But right now, it's a chicken and egg situation. It seems sensible to let Mac OS X Server gain a solid reputation on PowerPC before launching any efforts on Intel-compatible PCs. Gotta walk before running.

Finally, while PowerPCs and Pentium II/IIIs are similar in performance today, ignoring the ByteMark and looking at SPEC Int95 and SPEC fp95, a metric which has more credibility for servers, that may not be so true in the future.

Another Apple strategy is to technically embarrass the PC world with a spectrum of technical innovations. (Perhaps that's why Intel hired some CPU design engineers away from Motorola -- triggering a lawsuit by Motorola.) A combination of better RISC design, AltiVec, Firewire, MaxBus and other improvements are designed to put the PowerPC well ahead of the Intel architecture.

In light of the above thoughts about selling PCs, I don't see this as a problem. Companies often follow dual-strategies, moving towards the one that appears to have the highest payoff. In this case, if the PowerPC development can achieve a large enough advantage over Intel architectures to entice IT managers, that is, say, twice the performance for the same price, then Apple may go that route. Adoption of Intel CPUs would be deferred. On the other hand, if that initiative fails, then Apple might elect to sell high quality PCs and just take the money to the bank.

Good night America...

Clearly, it is premature for Apple to engage in Mac OS X Server for Intel. However, it would seem that the events of 1999 might make it entirely possible for Apple to change tracks and go that route. I think it's very smart of Apple to let Microsoft slowly get into more and more trouble while they test the waters with Mac OS X Server on PowerPC.

Apple is reentering a server market that remains skeptical -- due to past misadventures with Apple and Unix. It's not a bad idea for Apple to reestablish their own reputation as well as the product's reputation before carving out a small piece of a much larger PC marketplace.

If and only if the end of 1999 finds Microsoft a damaged and weaker company than today, I would expect to hear new rumors about Mac OS X Server & Desktop for Intel hardware. After all, flexibility is the key. Without it, any computer company disappears into technology blues.