In-Depth
Analysis: Intel, Sun Tie the Knot
After years of animosity, Intel and Sun have patched things up. Why? What’s the catch?
Former Sun Microsystems Inc. CEO Scott McNealy might not have coined the term "Itanic," but he certainly did his best to keep it afloat.
That’s why Sun’s accord last week with Intel Corp. came as a shock to so many in the industry: McNealy, after all, was usually first to dismiss Itanic—a derisive nickname used to describe Intel’s first-generation Itanium microprocessor and, by association, Intel’s enterprise aspirations in general. Nor was Sun any more congenial to Intel’s more successful 32-bit chip offerings—the Pentium IV or Xeon processors. When the Unix giant designed its own line of x86 server offerings, it partnered with Intel rival Advanced Micro Devices (AMD) Inc.
Last week, Sun and Intel announced a broad strategic alliance that spans Sun’s Solaris, Java, and NetBeans software as well as Intel’s Xeon microprocessors. Under the terms of the alliance, Intel will endorse the Solaris operating environment while Sun commits to deliver servers and workstations based on Xeon. Elsewhere, the Sun/Intel alliance has joint-engineering, joint-design, and joint-marketing components, too.
Sun CEO Jonathan Schwartz was positively buoyant when he hit the stage at last week’s launch event. "As someone earlier remarked, it was interesting to see those two logos side by side up there with no spontaneous creation of energy around them," he quipped. Opening badinage aside, Schwartz was deadly serious. "This is a market-changing event. This totally changes the perspective that a customer has on how they can do business with Sun and similarly how they can do business with Intel," he later said.
That sounds like a head-over-heels rapprochement by two companies that just four years ago were at loggerheads with one another. What gives?
Times change, industry watchers say, and so do product strategies. Back in 2003, there was a kind of logic in Sun’s spurning of Intel in favor of rival AMD. "Sun was just getting into the x86 game, a space about which it had never truly been serious before. It thus didn’t have extant x86 supplier relationships to consider. This made going with AMD a fairly straightforward decision for Sun. Indeed, it was a more natural choice than choosing Intel would have been," writes Gordon Haff, a senior analyst with consultancy Illuminata.
Charles King, a principal with consultancy Pund-IT, says the Sun/Intel pact has a kind of that-was-then, this-is-now logic to it.
"Sun’s 2004 embrace of AMD’s Opteron processors finally gave the company an x86 platform it could believe in and sell effectively," King argues.
In retrospect, Opteron was a slam dunk—and Sun correctly intuited as much.
These days, Schwartz says, most of Sun’s software runs on non-Sun hardware. He’s also counting Sun’s Java software—which runs on just about any conceivable platform—and its NetBeans IDE, but even Solaris, Sun’s crown-jewel, is most frequently deployed on non-SPARC platforms, Schwartz claims. More to the point, he says, Sun hardware must be flexible to power non-Sun software platforms. "Our servers at this point … are not just about running Solaris, they’re about running Windows, they’re about running Linux, they’re about running Red Hat," Schwartz indicated. Approximately 70 percent of Solaris downloads—that’s seven out of 10 downloads—are of Solaris x86, he said. "The software we ship dominantly runs off of Sun hardware. The majority of software that sun builds is running on Nokia hardware, or on Intel hardware, or on certainly non-sun hardware."
Why Intel? Why Now?
In 2003, Sun found AMD to be a better alternative to Intel, because—at the time, anyway—it simply had a better product: Its 64-bit Opteron was superior to Intel’s Xeon microprocessor in almost every respect, claims Illuminata’s Haff.
"AMD was on a roll, and Intel was in a bit of a funk. Simply put, AMD had the superior product—excellent performance, graceful 64-bit extensions, fast memory access, easy multiprocessing, etc.," he observes.
AMD and Opteron also made it possible for Sun to mature its fledgling x86 server enterprise. As a result, Haff claims, Sun now has a credible x86 server business. "Sun has stuck with its x86 efforts and sold no small number of systems. Many run Linux, but quite a few run the x86 version of Solaris, the open sourcing of which has also been a far more successful initiative than many expected," Haff says. "The bottom line is that Sun now has a mainstream x86 product line that looks to be an important, even essential, component of the company’s future. This, along with changes at Intel, makes for a different game."
Why Intel? More precisely—why Intel now? Much of it has to do with momentum, says Pund-IT’s King: if AMD was the no-brainer choice back in 2003, Intel is the no-brainer choice today. "Two-plus years after initially ignoring the 64-bit x86 boat launched by AMD—and losing a quarter or more of the x86 server market as a result—Intel spent 2006 refreshing its entire server and PC processor lines with solid new products," King indicates. "[I]f Sun and Intel were going to get together, it would be difficult to find a better time."
Illuminata’s Haff, for his part, says the deal also goes some way to establishing Sun as a "serious" x86 server provider. "Sun’s made its x86 beachhead" with its Opteron-powered Galaxy servers, he indicates. "Now it needs to plan for long-term growth in x86. For a serious Tier 1 provider, that means offering both AMD- and Intel-based products. Maybe not top-to-bottom or in every product category, but substantial product lines based on each supplier."
The upshot, Haff says, is that Sun, like Dell Computer Corp., IBM Corp., and Hewlett-Packard Co. before it, has merely faced up to the arithmetic. "[O]ne need only look to Houston, Round Rock, and Raleigh, where the other three major system makers have all concluded, however belatedly and reluctantly in some cases, that limiting an x86 lineup to one supplier only limits your market and your supplier leverage—substantially," he points out.