In response to a question about Microsoft's monopoly, Mills defined the "Windows franchise" as consisting of Windows and Microsoft Office. This franchise, he said, wasn't going to repeat itself. The tone in his voice when he said that was not just one of pragmatism. I had the distinct feeling that it was of also one of commitment and personal conviction.
Besides the Windows franchise, there are a lot of things that IBM would rather not have repeat themselves. Not buying MS-DOS when it had the chance. Allowing the cloning of PCs. The 32-bit OS debacle (when then-OS/2 partner Bill Gates basically gave IBM its walking papers). Once Microsoft's NT dispatched OS/2 as a viable option, IBM was cast adrift in the fast-growing software segment of Intel-based operating systems. Adding insult to injury, the "Wintel" platform became so heavily commoditized that it was now too easy for enterprise hardware competitors like Dell and the now possibly merging Hewlett-Packard and Compaq to carve out their own slices.
Having lost three golden opportunities (one hardware and two software) to capture a "tax" on every Intel system sold, IBM found itself in a marginalized position quite uncharacteristic of the world's largest computer company, and beholden to the same company that once jilted it--Microsoft.
When Mills looked me straight in the eye and said that it wouldn't happen again, he may have been referring to more than just the Windows franchise. IBM is a company that prefers to control its own destiny. It knows too well the risks of having its fate tied to the whims of Microsoft and Intel. In order to regain control of its destiny and prevent history from repeating itself, the company will have to do some marginalizing of its own--particularly to Microsoft and at the expense of Sun. To accomplish that, IBM must do something it failed to do with OS/2. It must provide its customers with a clear path from Microsoft's platforms to extremely compelling alternatives based on its own intellectual property, or, at the very least, to some open-source IP over which it exerts tremendous influence.
Enter Java and Web services
As a tool for marginalizing Microsoft, Java is everything OS/2 was not. Java is the heart and soul of IBM's WebSphere (the middleware for the rest of IBM's software portfolio). Java is the primary software vanguard that keeps Microsoft from penetrating the datacenter. Even better, Java represents the first credible threat to Redmond on the client side. By all counts, some version of Java now exists on many more client devices than does Windows (including desktop systems, phones, gaming consoles, set top boxes and even credit cards). Java has the all-important buzz too. 3G networks may be the all-the-rage at this year's CeBIT, but more important are the applications they will enable. Judging by the number of Java-enabled phones (e.g., Sony Ericsson's Z700 and P800), Java is far ahead of the Windows alternative (a variant of Windows CE). Finally, Java is winning the hearts and minds of developers. According to the Hurwitz Group, 35 percent of developers surveyed plan to deploy Web services on Java vs. 15 percent for .Net, and 31 percent plan to use both.
Java is the golden egg and Sun is the goose that laid it. To truly keep history from repeating itself (especially on the server-side, where IBM is somewhat beholden to Sun), IBM has few choices. It can take ownership of Java and, therefore, Sun as well. IBM could also try to dislodge Java from Sun's control and move it into the open-source community where IBM might not own it, but could pretty much have its way with it. When asked about the desire to own Java, IBM's Director for eBusiness Standards Strategy Bob Sutor said "I don't know about owning it, but we'd sure like to see it open sourced."
If IBM did manage to bring Java into its intellectual property portfolio, some of IBM and Steve Mill's biggest competitors--BEA, Oracle, Computer Associates--would be beholden to IBM in the same way that IBM is now beholden to Microsoft. In addition to new revenue streams from the licensing fees generated by Java, this is exactly the position in which IBM needs to be.
However, getting there (if IBM chooses to do so) would be a different story. The words "hostile takeover" would be an understated way to describe any such move by IBM. The only thing golf lover and Sun CEO Scott McNealy would probably like to see on a tee more than golf balls and the heads of Microsoft executives is the head of any IBM executive. Especially Mills.
While Mills and his software division have probably done more for Java than any other company besides Sun itself, IBM's Java-related activities have grown increasingly self-serving. Most notable is Big Blue's leadership role behind Eclipse. Bearing a name that appears to foreshadow the coup IBM may have in mind, Eclipse is a specification for adapting heterogeneous application development tools into an integrated development environment.
Prior to Eclipse, NetBeans was the specification for this sort of tool integration. NetBeans continues to get support from most members of the Java constituency, including Sun. Now, Sun's VP for Java Richard Green is claiming that Eclipse violates the tenets of the Java Community Process (JCP) as well as the integrity of the write-once, run anywhere promise of Java. IBM's top cheerleader for Eclipse, Scott Hebner, claims that Eclipse does nothing of the sort. Experts I've spoken with understand the need for both specifications, but do not disagree with Green's overall assessment.
And then there's the name: Eclipse. Choosing the word "eclipse" for something that was bound to cause friction in the Java world isn't exactly an olive branch to a company named Sun. IBM's Sutor claims the name was derived from the code names of Eclipse's predecessors--Lunar Eclipse and Solar Eclipse. "The name is purely coincidental," Sutor says. "I have no idea why those names were picked." Even so, the fact that IBM is the driving force behind Eclipse now makes the name appear more than coincidental. It could very well be the first of several visible moves to wrest control of Java away from Sun. Meanwhile, Sun claims that the JCP, and not Sun, controls the fate of Java. But IBM disagrees.
Taking control of Java isn't the only thing IBM must do to marginalize Microsoft. IBM must make it easier for Microsoft's customers to migrate from Windows to Java. If Eclipse opens that door, then IBM's collaboration with Microsoft to form the Web Services Interoperability Organization (WS-I) and the interoperability standards it supports knocks that door down. As I point out in my investigative analysis, the circumstances under which Sun was issued a last-minute invitation to join the WS-I as a contributor (as opposed to board member) reek of hardball politics. Should Sun join without a board seat, the company's plans for Java could be dictated in part by the companies that license it: IBM, BEA, and Oracle. All are WS-I board members. Advantage: IBM.
Microsoft is probably happy to participate in anything that sticks it to Sun the way the WS-I could. But I wonder if Redmond sees IBM's interest in the WS-I the way I see it--as an interoperability standard that will allow IBM to marginalize Microsoft by easily migrating Microsoft's customers to IBM's Java-based solutions. Should control of Java, through purchase or donation to the open-source community, ever be wrested from Sun, IBM would surely be in the catbird seat. Talk about a deal with the devil.
While IBM does what it can to keep Java's star shining brightly on its own terms, the company is also doing everything it can to dismantle the rest of Sun. The two companies are engaged in a bloodbath on the hardware front that has resulted in an increased share of the server market for IBM, at the expense of Sun, HP, and Compaq. (Sun vociferously debates the accuracy of IDC's figures.) The now merging HP and Compaq have each announced plans to discontinue their RISC technologies (PA-RISC and Alpha, respectively). IBM, however, is going full steam ahead with the PowerPC microprocessors that power the newest AIX-based enemy to Sun's Solaris and SPARC processors: Regatta.
Overall, the results for Sun, which is also embattled by Compaq, HP, and now Dell, are reflected in its stock price. As of this writing, Sun, which once traded in the 60s, was dangerously close to a three-year low.
Should Sun's $28 billion market capitalization decline any further, there's only one company that could absorb Sun with pocket change: IBM. In addition to putting IBM in control of its own destiny and giving it the IP and market presence needed to combat Microsoft, an IBM takeover of Sun would also turn the microprocessor battle against Intel into a two-horse race. If IBM makes its move, when will that be?
Should IBM buy Sun? Can IBM pull this off? Write to me at david.berlind@cnet.com, or share your thoughts in TalkBack.
