By now, we've all seen the signature scene in Terminator, when a massive truck explodes, apparently cremating the perfect robot from the future--the movie's namesake--played by Arnold Schwartzeneggar. This is the scene where the Terminator demonstrates his name by walking calmly out of the wreckage. Now reduced to its purposeful metal essence, it is no less determined, but a lot less pretty.

That's roughly what happened to Linux in the year 2000. At the dawn of the new millennium, Linux was personified in the press by two of the most successful stock offerings in history: Red Hat and VA Linux Systems. In August 1999, Red Hat went public with an IPO that went straight to the sky. Cobalt Networks followed with another strong showing in October. Then in December, the VA Linux IPO racked up a record-setting $320 share price on opening day. Andover.net had a strong IPO almost simultaneously, and in January 2000, VA Linux bought Andover for around a billion dollars. That same month at the Javits Center in New York City, LinuxWorld Expo was a gold rush mining town. As with the rest of the dot com boom, promise was everything. With so much well-funded promise, who cared about reality?

Then the dot com market crashed and kept crashing, like an exploding gas tanker rolling down the side of a mountain. A lot of companies didn't come out at the bottom. But the Linux companies did, and so did Linux itself, even though a lot of the attractive stuff was gone. Red Hat and VA Linux's stock prices were down around $10 per share, but their sales were up. Cobalt had sold for $2 billion to Sun and seemed healthier than ever. Linuxcare, Lineo and TurboLinux all pulled out of the IPO market, but who wouldn't when it's hardly there? Corel eased away from the Linux business, but their problems could hardly be blamed on the operating system. VA Linux and Penguin Computing backed farther away from the desktop and laptop business, but with Dell, IBM and other mass market hardware makers starting to put Linux on their branded laptops, that was probably a good idea. The much smaller Tuxtops made the same move, and got out of the laptop business.

But all of them were still in business. While Linux may have looked a lot less pretty to Wall Street, it was no less appealing to customers--or to the big hardware companies that fundamentally aren't out to sell operating systems.

Linux persists because it's real. It was real before the dot com craze, and it's still real after the crash. Demand is huge and growing. Michael Dell claims that 10% of his company's servers run Linux, and 40% of the customers at Dellhost.com choose Linux. IBM says it is spending a billion (yes, that's with a b) dollars on Linux development in 2001 alone. Sun Microsystems now has at least one of its tentacles in the Linux camp, thanks to its acquisition of Cobalt, which invented the server appliance category and whose boxes all run Linux. At this writing the Linux-based search engine Google is a runaway success--as well as one of Red Hat's biggest customers. As of February (when I'm writing this), Google is running on more than 7,000 Linux boxes. IBM reports that its S/390 mainframe (now rebranded as part of the Z series) can serve up many thousands of simultaneous instances of Linux. This all amounts to a bargain of sorts. We've seen claims that running Linux on a $7 million IBM mainframe saves about $40 million over a room full of Suns running Solaris and doing roughly the same work.

At LinuxWorld Expo in January 2001, I talked to one IBM engineer who used to work elsewhere but was attracted to Big Blue because it's where the ``really cool Linux stuff'' is going on. Another engineer told me that rallying around Linux blew away a lot of the ``battling business unit'' politics between different server hardware groups at the company. At the expo, IBM was the vendor with the geek habitat: a floor filled with bean bag chairs on which a herd of geeks lounged, talked and tapped away on laptops.

The expo itself was twice the size of last year's boom town. The added acreage came mostly from three sources: 1) big mainstream players, including IBM, Sun, SGI, HP and Borland; 2) embedded Linux development companies like Lineo, MontaVista, LynuxWorks and Timesys, plus a couple dozen less familiar names; and 3) international hardware development companies deploying Linux in everything from bookshelf servers to consumer electronics and factory gear.

The reality-based message across the whole category now is really quite simple: Linux is infrastructure. It's the way the essentially free and open infrastructure we call the Net manifests as an equally free, open and universal operating system. The importance doesn't derive from displacing other operating systems (Windows is also up in all categories), but rather from its virtues as solid, well-understood and easily deployed infrastructural building material. If you're selling anything other than operating systems, it's probably good for you, in the same way as the Internet is good for any business not directly competing directly with it. Linux is a winner for just about everybody because just about everybody can probably find a good way to put it to use.

The proof is in the pudding you're holding in your hands right now.