Sun Selling Windows? Stranger Things Could Yet Happen

by Timothy Prickett Morgan

When the executives from Sun Microsystems and Microsoft took the stage in San Francisco this morning to announce their stunning partnership and the settlement of hostilities from these two warring tribes in the IT industry, it was pretty obvious why Sun and Microsoft didn't make their announcement a day earlier, on April 1. Everyone would have thought it was a joke.

But it was April 2, and we were not in an alternate universe, but rather in the real one where the gravity of industry giant Microsoft can warp everything in proximity, even a staunch competitor such as Sun. When Scott McNealy, Sun's chairman and chief executive officer, announced the Microsoft alliance, he said it was yet another example of Sun's "distruptive innovation" and that it was "a huge day for customers," who had been telling him that they needed peace between the two companies, who have been at each other's throat, in the courtroom and in the marketplace, for 10 years. Since both, in fact, became upstarts in the data center.

While Sun would never cop to this interpretation of events, both Sun and Microsoft only got into the data center in any measurable way after a decade of getting onto the corporate desktops--Sun in the engineering offices of manufacturers and research institutions and the trading rooms of financial institutions, Microsoft on more plebian desktops with Windows-based X86 machines. The two companies did not clash so much on the desktop, but they certainly have knocked heads in the server market. And now, to the surprise of Sun's partners and competitors, the RISC/Unix workstation and server company is getting into position to sell X86 Windows and Linux gear. This is not the belligerent Sun Microsystems we knew from the heady days of the dot-com boom, where it was growing revenues at 50 percent a year and looked like it might have become the dominant server maker in the world, bypassing IBM and far outstripping Hewlett-Packard and Compaq either before or after their merger. This Sun would, if trends had persisted (meaning the economy didn't tank, if Sun was not so late with getting its UltraSparc-III and UltraSparc-IV products out the door, and if Linux hadn't take root in the market), be roughly the same size as Microsoft is today, about $30 billion in revenues. And it might have even maintained an equal market capitalization even if there is no way that it could have built up the $53 billion cash pile that Microsoft has.

This Sun is a practical one, the one that will take $1.95 billion today from Microsoft instead of awaiting an uncertain fate in courtrooms in the United States and Europe, where two antitrust cases were moving along at their normally glacial pace. Sun didn't just take the settlement money today in the hopes of using it against Microsoft tomorrow, although that is clearly part of Sun's plan. Sun is one-third Microsoft's size in terms of sales, and prior to the settlement, its cash hoard was just over $5 billion--a tenth the size of Microsoft's. Sun does not have a huge margin of error. It's about two quarters worth of sales, in fact. Microsoft has something closer to five quarters of slack. The amount of slack (by which we mean how many weeks of cash does it have to run the company if it didn't make any more sales) that IBM and HP have can be measured in weeks, but IBM at least has a large backlog of recurring software and services money and HP is working on building its own recurring revenue streams. So are Microsoft and Sun, by the way, as is every other player in the IT industry. Everyone is sick to death of the product sales cycle, which just keeps getting longer and more difficult. IT shops have gorged themselves on technology, and they are as sick of buying what they do not need as IT vendors are as sick of selling it.

It is a practical, humbled Sun that opted last year to sell Xeon-based Linux servers and brought Solaris for X86 back from the dead. It is with humility as much as excitement that Sun has embraced Advanced Micro Devices' Opteron processors for a new entry and midrange server line, running both Linux and Solaris. And it is a deadly serious Sun that will sell X86 machines running Windows 2003 and its entire solution stack because in many cases this is what Sun's own Unix customers want. Sun wants to have customers, any customer it can get, more than it wants Sparc-Solaris customers. Sun would rather be the preferred platform vendor in a shop and avoid the religious wars about operating system platforms. In this regard, Sun has become no different from IBM and HP, and it is still, in many ways, better off than Dell because Sun still invents stuff. Sun has smart engineers, and they do not just repackage what Intel and Microsoft tell them to. Whatever twist Sun can come up with to make Windows unique on its platforms, you can bet it will deliver it.

Neither Sun nor Microsoft would elaborate on the technology alliance between the two companies last week, saying that they were still hammering it all out. But wouldn't it be interesting if Sun decided to port Windows to the Sparc platform? There is no technical reason why this could not happen, and for all we know, the Sparc architecture might run certain Windows applications better than Xeon or Itanium alternatives. Such a move would give Sun a counter argument to the HP Integrity line, which supports Linux, HP-UX, and Windows on the same Itanium-based platform. IBM could similarly port Windows back to the Power platform (Windows 3.51 ran on the Power processors, and then IBM pulled the plug on development). In any event, Sun could sell giant Sparc boxes that ran Solaris, Linux, and Windows in partitions, much as HP is doing with Integrity machines. IBM will soon be pushing the fact that its Power5-based "Squadron" machines can run AIX, Linux, and OS/400 concurrently and equally. Sun could make the same boast and position its Sun Fire Sparc-based machines much the same way. Alternatively, Sun could mix-and-match Sparc, Xeon, and Opteron uniboards inside Sun Fire chassis and accomplish the same thing, much as Unisys does with its Clearpath mainframes, which support mainframe (MCP or OS2200), Windows, Linux, and Unix platforms on proprietary mainframe and Intel engines.

Even stranger still, now that Jonathan Schwartz, a software guy, is now president and chief operating officer, Sun could finally get around to forging a processor, server, and Solaris development partnership with Sparc clone rival Fujitsu-Siemens and stop funding its own Sparc line entirely. If Sun can bury the hatchet with Microsoft, it can put away the katana for Fujitsu-Siemens, which is more of a partner than rival anyway. If Sun wanted to get to profitability quickly, it might outsource manufacturing of its entry and midrange servers to a third party, co-develop high-end machines with Fujitsu-Siemens, and keep the future mightily multithreaded "Niagara" and "Rock" processors in house as a future profit engine.

The resulting Sun would not be the platform company that we once knew and usually admired. But it might actually make some money and win some customers, and that is what business is really about.