Hewlett-Packard Co. (HP) squandered a chance to become a leader in software used for automating business processes and other services over the Internet, a failure highlighted by its disclosure last week that it plans to stop developing its middleware products, industry analysts said.
HP told financial analysts last Tuesday that it plans to "retire" its middleware line, which it said has suffered heavy losses, as part of an effort to make its software group profitable by year's end. Besides its family of Netaction software, which includes a J2EE (Java 2 Enterprise Edition) application server, a transaction server and developer tools, its middleware also includes e-Speak, a technology that lets devices such as printers share services over networks.
HP has sought to downplay the news, saying speculation about its middleware plans is premature. It will detail its software strategy at a customer event later this month and in the meantime declined to comment for this article. However, industry sources have said HP is trying to sell off all or part of its middleware business, with plans to rely instead on partnerships with vendors such as BEA Systems Inc. and Microsoft Corp. to provide the middleware for IT systems that it sells to customers.
Whatever the strategy, many analysts agreed on one thing: HP blew an opportunity to stake out a leadership role in middleware, an increasingly important area of IT that has become a focus for much of the development work happening around Web-based computing. Some analysts said a partnering strategy is the right course, while others said it raises questions about whether the "new" HP can achieve its goal of being a "complete provider" of enterprise IT solutions.
HP laid out a vision for using software to create Web-based services at least as early as Sun Microsystems Inc., Microsoft Corp. and IBM Corp. In 1999 it proudly opened the doors to HP Labs, its renowned research division, to show how technologies such as e-Speak could be used to link people, computers and even objects in a distributed computing model, a concept it called CoolTown. HP engineers talked of services that might, for example, check Web-based traffic reports and adjust a commuter's alarm clock to give them the right amount of time to get to work.
In October 2000 HP built on its vision when it announced plans to acquire Bluestone Software Inc. for an estimated US$470 million, which became its Netaction product line. But while Sun, Microsoft and IBM worked hard to build out their software platforms for Web services, HP let slip its early advantage, industry analysts said.
"They had many of the key pieces early on and didn't execute or project a vision," said Dana Gardner, an analyst with Boston-based Aberdeen Group Inc. HP's middleware "sat on the shelf for a long time, while their competitors rallied and put together a full software stack" for Internet-based computing, he said.
"HP has a long history in software and has shown some real leadership in terms of ideas, but they really have not executed," agreed Martin LaMonica, a senior analyst at Nucleus Research, in Wellesley, Massachusetts. "The whole Netaction-Bluestone thing is another indicator of this."
HP's missteps can be traced to the time of its Bluestone acquisition, said Shawn Willett, an analyst with Current Analysis Inc., in Sterling, Virginia. While IBM and BEA solidified commanding leads with their application servers, HP's product never gained traction among customers and the company never built on its acquisition. For example, it never took steps to build or acquire an integration server, which would have strengthened its middleware hand.
HP's software organization is "very OpenView-centric," he said, referring to HP's IT management software. "Perhaps the management, or others inside HP, didn't realize the value of what they had bought with Bluestone."
Late last year HP said it would begin offering its application server for free with other product bundles, charging instead for more advanced features and add-on products. As an employee with HP's middleware division said last week: "(HP) said Bluestone would be the first part of a five- or six-step strategy. In the end they became preoccupied with the merger (with Compaq Computer Corp.) and never took those other steps."
HP's middleware ambitions at one time appeared substantial. In November 2000, a few weeks after announcing the Bluestone deal, HP Chairman and Chief Executive Officer Carly Fiorina took the stage at the Comdex trade show to declare that HP would be a leader in software for Internet-based computing. "By adding Bluestone's XML-based Web application server and tools, we're creating the richest development environment for the services-based model that I'm talking about," she told the Comdex crowd. She also highlighted e-Speak and other HP software.
HP wouldn't provide information about any customers e-Speak may have won over, but analysts said it found little traction. In fact, HP admitted quietly to some analysts last year that it had brought e-Speak's development to a halt, said David Smith, a senior analyst with Gartner Inc., in Stamford, Connecticut. HP's Web page for the technology directs users to a different set of middleware products.
HP never had a "golden opportunity" to be a leader in middleware, according to Smith, but he agreed that it missed a chance to build on its early lead in defining the Web services model.
E-Speak was likely ahead of its time, and is a credit to the innovative work that has come out of HP Labs, LaMonica said. But as its time approached, HP apparently wasn't there to support the technology, he said.
"The general idea (of e-Speak) was a services-oriented model of computing where tasks were handled by machines. Later the standards coalesced around SOAP (Simple Object Access Protocol) and WSDL (Web Services Description Language), but for a time HP was way out in front," he said.
Perhaps not coincidentally, on the same day that Fiorina touted HP's software to the Comdex audience, the company also announced it had missed its financial targets for the quarter by a wide margin. Referring to the botched vote count for the 2000 Presidential election, Fiorina quipped: "Don't we get a recount too?"
HP soon began work turning around unprofitable areas of its business, and apparently felt that making substantial investments in its money-losing middleware division was not the way to go, Willett and other analysts said.
Aside from middleware, HP is not without its strengths in software, analysts note. Its HP-UX operating system survived the merger with Compaq and is being tuned for Intel Corp.'s latest Itanium processors. OpenView is among the most popular software for managing IT environments, and HP officials have said its OpenCall software for the telecommunications industry is also profitable. The company is hard at work developing software for the emerging "grid computing" model, and HP Labs continues to experiment with new types of applications.
Nor will a middleware strategy based on partnerships necessarily hurt HP or its customers, some analysts said. For customers who favor Java, HP can offer an application server from BEA, while for Microsoft advocates it can leverage Compaq's close ties to the software giant to offer its .Net products. HP could potentially benefit from the strategy, winning customers by positioning itself as a neutral vendor that favors neither software platform, Willett said.
With its money-losing middleware business out of the picture, HP can also focus more on developing its successful products. Its strategy moving forward will likely be to improve the software used by its installed base of customers, by improving the high-availability features in HP-UX, for example, or enhancing its OpenView products, LaMonica said.
"Companies need to be smart and recognize where they do have a core competency" and where it's best to pursue partnerships, said Gartner's Smith, noting that neither Compaq nor HP was strong in middleware.
However, HP's partnership strategy may also have downsides, analysts said.
"If you're hawking your hardware with someone else's software, you lose an interesting part of the relationship with the customer," Aberdeen's Gardner said. "Customers tend to choose software strategically and hardware based on cost, performance and availability, so if you've lost that relationship and it's your partner's software you're selling, then your partner may have a stronger relationship with the (customer) than you."
The partnering strategy also raises questions about the new HP's ability to live up to its own billing as a complete provider of IT systems for large corporations, he added.
"Part of the rationale (for the merger) was that they needed to combine forces in order to be a full solutions provider to the large enterprise and to compete with Sun, IBM and Microsoft," Gardner said. "If they are in fact throwing their (middleware) on the (acquisition) market ... then I'm curious how they are going to become a full solutions provider."
Middleware will play an increasingly important role as businesses, looking to become more efficient and cut operational costs, start to combine J2EE, .Net and legacy applications into a distributed computing architecture, he said.
"It's the sticky part of the relationship (with the customer), and probably a high-margin business," Gardner said. If HP goes through with its plans to stop developing middleware, "it raises fundamental questions about the nature of HP as a company."