PHOTONIC SYNTHESIS
Cultivated with ample capital, Juniper Networks aims to survive among the tall trees of the networking forest.
By Brian E. Taptich
Asked why he left his post as principalscientist at Xerox PARC to found Juniper Networks in 1996, Pradeep Sindhu stares out his window for a disconcertingly long time. Choosing his words carefully, he reflects, "I saw that the development of the communications industry would present an opportunity that comes along once in a lifetime--a chance to do something great."
The development that the philosophically inclined Mr. Sindhu envisioned was the convergence of the disparate communications infrastructures into a single fiber-optic network--a common enough idea now but one that at the time was confined to the musings of computer scientists. The opportunity Juniper has targeted is the development of networking electronics for this new, universal network.
Juniper, to this point, has been tight-lipped about its product development. The company does admit to developing integrated hardware and software equipment designed to route IP traffic at the core of large-scale fiber-optic networks. The family of products ultimately would be used by communications service providers to keep traffic moving along their backbones.
The giving tree
Those who have seen Juniper's efforts thus far clearly have been impressed--the company has garnered outside investments of more than $60 million. While the usual stable of venture capitalists has supplied a portion, the bulk of the capital has come from partnerships with the major communications equipment companies. In Juniper's second round of financing alone, six corporate investors each contributed roughly $7 million.
The notable absence of Cisco from the roster of partners has created the perception of Juniper as a potential "Cisco killer"--an angle that, Mr. Sindhu suspects, was devised to sell magazines. More likely, the company is a victim of its own secrecy. In the end, all communications equipment companies, especially those developing routing equipment for any network, compete with Cisco in some way.
Juniper has chosen a lucrative market to target--Dataquest estimates that today's $700 billion communications services sector will exceed $1 trillion by 2001--and has surrounded itself with impressive investors. But Juniper has yet to claim any of this market: its first product is still in development, and the company has promised its partners it will be finished by spring.
From many, one
Mr. Sindhu and CEO Scott Kriens have helped grow Juniper to more than 85 employees, most of whom are engineers with networking and communications hardware backgrounds. "The talent for building any Internet-scale device is rare," Mr. Sindhu asserts. "And the problems of building a new architecture for an integrated network are especially subtle."
Currently there are five separate networks: cable, data, video broadcast, voice, and the Internet. Because each type of traffic has different characteristics, each network is designed differently. These existing networks are anything but integrated. As a result, most networks can't work together efficiently, and some can't work together at all. "Communications services companies are just like freight carriers or airlines or railroads," Mr. Kriens contends, "and running all of these disparate networks is the equivalent of the railroads' building five separate sets of tracks into every town. It ends up being more than five times as expensive as any one set."
Juniper argues that there is nothing preventing a single fiber-optic, packet-switched network from carrying all kinds of traffic. Simultaneous improvements in fiber-optic technologies (like wave division multiplexing, or WDM) are significantly increasing the capacity of fiber optics, while advances in networking hardware (like very-large-scale-integration, or VLSI, chip design) are increasing the speed with which information can be routed. "WDM is drastically expanding the amount of information that can be carried optically," says Mr. Sindhu. "And advances in optoelectronic technology are moving the information at faster rates."
Juniper claims that its VLSI chip architecture is customized for routing IP traffic across fiber optics at multigigabit speeds. The company aims to integrate WDM technology directly into its products. And IBM has signed on to develop the application-specific integrated circuits for the backbone devices.
A little friendly competition
But Juniper is neither the first nor the largest company to predict this shift: several of the company's corporate investors are developing their own routing architectures. Yet Juniper officials are confident that its optoelectronic IP routing architecture can steal at least a portion of the market. "When there is a change like this, there is an opportunity for a company that can think about only this problem," says Mr. Kriens.
Unlike many of the networking startups that develop a product only to await acquisition, Juniper is intent on going it alone, according to Mr. Kriens. All the company's partnerships, he says, were specifically designed to spur product development and ensure that Juniper remains autonomous.
Juniper's backing may validate its efforts thus far, but Mr. Kriens admits that it is hardly tantamount to success: "Remember, our investors' collective revenues are around $150 billion, so investing in a company valued at $175 million is easy for them."
The company clearly has developed something that the communications incumbents like. And Lucent's recent acquisition of Livingston Enterprises and rumored intent to purchase Ascend demonstrate that the convergence of the communications infrastructure is imminent. But Nokia's December acquisition of Ipsilon Networks also shows that even the most touted private networking companies have settled for acquisition.
Juniper officials recognize that the company's climb toward possible greatness has just begun. "Everybody is our competition--everybody that offers services or products across any network will ultimately be part of this integrated network," says Mr. Kriens. "We're just trying to figure out how we'll fit into the new paradigm."
Juniper Networks at a glance
CEO Scott Kriens
Location Mountain View, California
Phone 650/526-8000
Web www.juniper.net
Ownership Private
Founded 1996
Employees 87
Product High-speed switches and routers
Partners IBM and corporate investors
Competitors 3Com, Avici, Ascend, Bay Networks, Cisco, Ericsson, Lucent, Northern Telecom, Siemens
Revenues FY97 Prerevenue
Financing $62 million
Venture investors New Enterprise Associates, Kleiner Perkins Caufield & Byers, Institutional Venture Partners, Crosspoint Venture Partners, Benchmark Capital, Venture Lending and Leasing
Corporate investors 3Com, AT&T, Ericsson, Northern Telecom, the Siemens/Newbridge alliance, Qwest Communications, UUNet Technologies (WorldCom)
Valuation $175 million |
