Since the birth of the industry, the world’s greatest IT companies have regularly faced a pivotal decision: when new and potentially disruptive technologies emerge, should they adopt and try to take ownership of them or, alternatively, do all in their power to suppress their evolution and acceptance? As Harvard professor and business author Clayton Christensen spotlighted in The Innovator’s Dilemma, many one-time high-tech giants failed to take the first route and were subsequently overtaken by new technology upstarts.
Perhaps executives at Sun Microsystems, the IT systems giant, are too keen students of history. Today Sun faces that same basic dilemma, but with a twist. It has developed many innovative and potentially disruptive technologies itself, including the network file system (NFS), the Unix-based RISC workstation, the Java software environment, the first 64-bit enterprise servers – technologies that have provided the backbone to corporate computing. Yet while those helped fuel a meteoric rise in the 1980s and 1990s, Sun’s fortunes have turned and over the past two years its sales and profits have tumbled. To its critics, that downfall has been at least partly self-inflicted.
Certainly, Michael Dell, chairman and CEO of one of Sun’s most bitter rivals, Dell, thinks so. Delivering a keynote speech in September, just hours before Sun founder and CEO Scott McNealy was due to take to the stage, he launched a thinly veiled attack on Sun’s credo that innovation is what makes companies winners.
“R&D is over-rated,” Dell said. “People say we have our weaknesses, but I look at our results and they seem to be pretty good. Something must be working right here.”
Although an easy target right now, the statement clearly infuriated McNealy. His company has consistently spent a larger proportion of its revenue on research and development than almost any other mature IT company, and even as sales have tumbled, that proportion has grown from about 10% to over 16%. Thousands of employees have been laid off elsewhere since the dot-com bubble stopped fuelling Sun’s runaway growth but the company’s R&D operation has escaped virtually unscathed.
By comparison, Dell spends a tiny fraction of its sales on R&D and yet is the most profitable computer maker (McNealy might say ‘assembler’) in the world, thanks to its build-to-order, direct sales model and finely tuned supply chain.
McNealy wasted little time in responding. “Michael says R&D is over-rated. That is good that he believes that because we believe the opposite. Imagine if we were to launch a computer that we launched 10 years ago with the innovation level that we had 10 years ago. Imagine if we were to launch the UltraSparc computer today, instead of the UltraSparc III. Imagine if we were to re-launch Solaris 2.1. Get the barf bags out.”
He added: “Innovation matters, it does. We have invested like crazy through the bubble and the post-bubble, and even through some pretty challenging economic times we have maintained our $1.8 billion-plus R&D investment.”
Much of that investment is spent on the twin engines of Sun’s systems: Solaris, its highly rated implementation of the Unix operating system, and UltraSparc, its 64-bit reduced instruction set computing (RISC) processor. Sun also now sells Solaris and Linux-based servers running on Intel chips, but its support for Linux and other open source technologies has struck some critics as lukewarm at best. McNealy says such views are misinformed. “I croak when the press repeats our competitors’ dribble that we are closed and proprietary. There is no company on the planet that has more of a legacy, a tradition and a dedication to open interfaces, to leveraging the open source community [than Sun],” he says.
At any rate, it is still the combination of Solaris and UltraSparc that has consistently differentiated Sun from the rest – whether that is Hewlett-Packard (HP) with its combination of PA-RISC and HP-UX or even the mighty IBM with AIX and the PowerPC chip. At a time when IT managers are seeking to cut costs, Sun’s problem has been a perception that there are cheaper alternatives to the Solaris/Sparc platform available that offer good enough functionality and scalability for most computing tasks.
Many organisations have begun to move an increasing slice of their IT infrastructure to what they perceive to be cheaper Windows and Linux-based platforms. For example, Merrill Lynch, E*Trade and Oracle have ripped out banks of Sun servers and replaced them with cheap Linux-on-Intel (Lintel) boxes. The case of E*Trade is particularly symbolic. The online share-trading site – just the kind of Internet business that fuelled Sun’s rapid growth during the boom years – says it has cut costs significantly after replacing 60 Solaris/Sparc machines with 80 Linux-running Intel servers.
All this is worrying for Sun because its business model depends on its ability to leverage contracts for its highly innovative systems to pull through add-on storage, software and services sales. “While Sun has multiple revenue sources, they are all anchored by the health of Sun’s systems business and [the company] cannot prosper without a healthy systems business,” says Richard Fichera, an analyst with Giga Information Group (recently acquired by Forrester). Indeed, as McNealy himself is wont to say: “The name of our company is Sun Micro…systems.”
But Sun faces an uphill battle in the processor wars. “At some point, Intel CPU sales will become an attractive alternative for much of Sun’s customer base,” says Fichera. “[Sun is fighting] a long-term losing battle against Intel and IBM.”
Yet, for all the company’s problems, giving up on Solaris/Sparc systems remains a taboo subject.
“We have our own operating system, it’s called Solaris. I know this statement bothers some in the community, but there it is,” says Jonathan Schwartz, Sun’s executive vice president of software. “Back in 1993, the world told us to abandon Solaris for Windows. We disappointed Wall Street and some customers by staying our course with Unix, with Solaris. Thankfully, we didn’t blink. You can’t even name the companies that did – they no longer exist.”
Despite its faith in Sparc and Solaris, Sun is not unmoved by the economic realities of the Lintel model. Sun prices are coming down and customers are now being urged to capitalise on the company’s problems. “Buyers have an opportunity to get some great Sun technology without breaking the bank,” wrote Charles Rutstein, a Forrester Research analyst, in a recent report. Several organisations have told him that Sun offered them “terrific discounts” on maintenance and support fees when faced with the threat of desertion to Linux. He also urges companies using ageing Sparc machines to use “Linux pricing pressure” to get deep discounts on replacement Sparc equipment. “There’s time to migrate that core app to Linux on Intel in 2007,” he says.
Characteristically, Sun hopes to innovate its way out of trouble. Rather than close down a large section of R&D, it is hard at work on a broad strategy designed to cut the costs and reduce the complexity inherent in running large IT systems.
The centrepiece of the strategy is N1, a blueprint for utility or organic computing. Sun’s biggest rivals for this nascent approach, IBM and HP, may have made more marketing noise than Sun, but Sun’s moves have been just as shrewd, say analysts.
For example, while IBM and HP arguably have more experience in systems and storage management software, Sun’s acquisitions of Pirus and Terraspring – bought in the last 12 months for $167 million and $30 million respectively – have helped it narrow the gap. The deal to acquire Terraspring was particularly galling to HP, which had made the start-up a partner in its utility computing strategy.
To Pirus’s storage provisioning software and Terraspring’s server provisioning software, Sun will shortly be able to add the recently acquired CenterRun’s application provisioning technology. Include the blade servers that Sun launched in spring 2003, and analysts say that a potentially strong – albeit still embryonic – utility computing strategy is taking shape. “Smart acquisitions like Pirus and Terraspring prove that Sun is serious about executing on its organic IT vision,” says Rutstein.
Fichera agrees. “Virtualisation has immense value to customers, particularly customers with large, complex environments – exactly the kind of users which have committed to Solaris in the past and which are the backbone of Sun’s customer base,” he says. “Along with new systems, productising N1 must be the highest priority for Sun during the next 36 months.”
Another strand of Sun’s new strategy is its attempt to tackle contractual complexity with its customers through a new and simplified way of selling middleware licences and desktop systems. The new stack – known as the Java Enterprise System (formerly codenamed Project Orion) and including Solaris, the SunONE application server and Sun’s directory, portal and provisioning software products – is being offered at the per-employee price of $100. (Customers can still buy the stack under a more conventional licensing model if they prefer.)
Since the price of the software is not tied to the numbers of processors or users, analysts say that it will suit organisations with big computing needs on account of having many online customers, such as financial services institutions, government agencies, media companies and telecommunications operators.
Sun hopes the new middleware stack will raise revenue by as much as 40% among its biggest customers. But the supplier stresses that this does not mean clients will be paying more for IT. Although the cost of the software may rise, the closely integrated stack will enable customers to cut down on integration costs, says Sun. The $100 fee includes some consulting, training and support ‘credits’, depending on the size of the company. And the new stack will be updated every quarter and comes with a considerably shorter licensing agreement, helping to simplify the task of portfolio management.
Sun hopes to take further cost out of IT through its new Linux-based desktop model. The Java Desktop System (formerly codenamed Mad Hatter), also launched in September 2003, includes the StarOffice personal productivity suite, the GNOME graphical user interface and Mozilla, the open source web browser. Sun charges $100 for the desktop system, which it claims is up to 80% cheaper than buying a desktop package from Microsoft.
But Sun’s Schwartz is under no illusions about the task in hand. “I was with the CIO of a large investment bank in New York about six months ago and I was 20 minutes into a one-hour meeting. The CIO looked at me and said: ‘Son, Microsoft could triple its prices and we would still buy its products.’ I looked at him, got up, shook his hand and said: ‘Thank you very much for your time. You are not my target demographic.'”
Issue of trust
Meanwhile, Sun’s hundreds of engineers continue to work on Solaris. In the face of criticism from Wall Street, and despite growing evidence of customers migrating applications to Linux, Sun believes that the increasing importance of IT security will give renewed vigour to Solaris’s value. The next version of Solaris, simply codenamed Solaris ‘Next’, will essentially contain all the elements of Sun’s security-hardened ‘Trusted Solaris’ operating system. This version, specifically developed in the early- to mid-1990s for government and intelligence agencies, does not allow ‘super user’ or ‘root access’ capabilities common in other operating systems. With super user privileges, a single user can obtain access to every element of an operating system, without first seeking permission from an administrator. Trusted Solaris also offers a tamper-proof audit trail that logs all activity.
But even Solaris is not completely secure. The announcement of Solaris ‘Next’, on 17 September 2003, was overshadowed by the revelation that a major vulnerability had been discovered by IT security experts in Solaris and Trusted Solaris. Although there was no evidence that hackers had exploited the flaw, the news – and particularly its timing – was embarrassing for Sun and prompted John Pescatore, a Gartner analyst, to state: “This confirms that security weaknesses are by no means limited to Windows.”
In any case, by the time Solaris ‘Next’ is ready to be shipped, probably some time in the second half of 2004, McNealy expects his company’s utility computing, middleware and desktop strategies to be gaining traction. By then, either his R&D-intensive strategy will be looking vindicated – or the calls for him to step aside will grow louder.
But he says he is not afraid of a challenge. “I have never believed in doing something the way it is already being done. That does not improve anything, that does not solve anything. Not having a slightly controversial strategy means you have no differentiation, no value to add, no way to make a difference.”
If there has ever been a point in Sun’s illustrious history when McNealy needs to pull a major differentiator out of the R&D bag, it is now.