BroadVision Sees the Java Light

The stakes are huge for the latest version of its software, which moves squarely into the open-architecture arena

By Margaret Popper

Despite dot-coms' spectacular tumble, Internet infrastructure isn't dead. In fact, among the companies that write Web-applications software, an interesting drama is playing out. Over the last year, a standard has emerged in applications-server software, the platform on which all other Web applications run. And that's forcing a group of companies to retrench. Hardest hit is BroadVision (BVSN ), a Standard & Poor's 500 company whose CEO, Pehong Chen, is considered the grandfather of applications-server software.

At its current price of around $13 a share, 86% off its 52-week high last March, BroadVision may look pretty attractive to investors. But even at this apparently bargain-basement price, analysts are advising caution. On Mar. 31, BroadVision will introduce Version 6.0 of its applications-server software. The company's fate pretty much hangs on the market's reception of that product. Until analysts get a sense of whether 6.0 will be well received, they say this isn't necessarily a buying opportunity in BroadVision.

The irony is that BroadVision's new offering is responding to developments in the very applications-server market the company helped create with its innovative products. Unfortunately for it, where this niche's hardware is concerned, the market has moved away from proprietary architecture -- a unique hardware setup that will operate only with instructions from a specific programming language -- to open architecture that can take instructions in Java, an open programming language.


  BEA's WebLogic and IBM's WebSphere have become the de facto standard for applications servers. BEA is the leader in this market, with IBM strongly entrenched in second place. Both of these servers use open, Java-programmed architecture that can be easily accessed for updates or fixes. That's not the case with proprietary architecture.

By embracing open architecture and therefore Java, the world has bypassed BroadVision's formerly revolutionary software based on a programming language called C++. "Pehong pioneered this space," says Heather Bellini, an analyst with Salomon Smith Barney. "In a way he's the victim of having been so successful."

In 2000, BroadVision took the first step toward retrenchment with the introduction of its 5.0 applications server. This product could interface with Java programming to the extent that it could run new applications in Java. But it had limitations and particularly affected was a key group of clients, including financial institutions, utilities, and telecommunication companies that need more control over their hardware's functions.


  BroadVision's 6.0 applications-server software is much more flexible, allowing clients to do many more functions in Java. In addition, at the end of 2000, BroadVision entered into a partnership with BEA to write applications that run on the latter's servers. If nothing else, the partnership is a testament to BroadVision's commitment to getting with the new open-architecture standard.

Still, some analysts argue that 6.0 hasn't gone far enough because it isn't written entirely in Java. But Simon King, Broadvision's vice-president for advanced technology, claims that as a programming language, C++ has certain speed advantages over Java that make it worth keeping. "C++ makes sense for certain performance-sensitive tasks. The trade-off is the easier maintenance and memory-space advantages of Java," King says.

Of course, the new standard isn't Broadvision's only worry. Other competitive threats lurk. If BroadVision can't satisfy clients' need for open systems, it's possible that up-and-comer Art Technology Group (ARTG ) will. Art achieved profitability only this year, with a 5.4% operating profit margin, vs. BroadVision's 10.8%, for 2000. The challenger is in a slightly different business than BroadVision, selling the tools to build applications-server software and only a few applications. But Art's toolkit is Java-based -- a key advantage.


  Even with the competitive threat of the new standard, it probably doesn't make sense for BroadVision to trash its entire product line and build one with Java programming from scratch. The bulk of its sales comes from selling packages that include software for nine different applications -- everything from finance and commerce to travel, billing, and publishing. The sophistication and breadth that C++ products offer probably can't easily be replicated using Java. "Seven hundred man-years have gone into developing these applications," says King. "It's a robust group of applications with 1,200 customers running on it."

The strength of BroadVision's software was (and is) its personalization, which is built into the framework on which the applications run. Personalization is the ability to show relevant content to a specific user, and it's the basis for the customer-relationship management software that lets companies target buyers for cross-selling opportunities. "Our customers tell us we are the personalization engine of choice, without equal," says King. The hope is that with 6.0, they'll get flexibility without losing any of the functionality of BroadVision's traditional offerings.

Evidence that Broadvision is struggling appeared most recently when it reported earnings. The company spent so much money beefing up its staff, hiring salespeople, service reps, and engineers that it missed its fourth-quarter 2000 earnings target. "When I see a company making that kind of investment, I wonder if it's because they've realized they're farther behind than they should be, or their core market is slowing down, and they've identified something else they should be doing," says Melissa Eisenstat, an analyst at CIBC World Markets.


  The analyst community is taking some wide-ranging bets on this stock, a clear reflection of the uncertainty surrounding the company's future. The 12-month target price estimates of the 26 analysts who follow the stock swing from $18 to $45. Consensus estimates for the first quarter of 2001 are 2 cents earnings per share and $137 million in revenue. That's flat growth and 1% sales gain over the fourth quarter of 2000. The prognosis for the year is better, with per-share earnings climbing 31%, to 17 cents in 2001, and revenues rising 50%, to hit $622 million.

Which way the stock goes will depend on the performance of 6.0. "I don't think it'll be a failure," says First Union Securities analyst Jason Maynard. "They've got a lot of work ahead of them, and it's not something they can do overnight, but they have the potential for some real success down the road." For the short term, however, too much may be uncertain for investors to get too excited about the stock.

Popper covers financial markets in New York for Business Week Online

Edited by Beth Belton