Grand Junction Networks. Primary Access. Centillion Networks. The names ring out like war dead. If industry giants Cisco Systems, 3Com, Cabletron, and Bay Networks were warlords, Genghis Khan would be shivering in his boots.

For venture capitalists interested in backing initial public offerings, this kind of carnage poses problems. In order to take a networking company public today, a venture capitalist has to look for a company that is "absolutely world-class," according to Jim McLean of Highland Capital Partners. "There are opportunities out there, but it isn't nearly as easy as it was."

IPO, therefore I amAndrew Verhalen of Matrix Partners experienced this dilemma last November when Cisco bought Grand Junction Networks, a company that Mr. Verhalen co-founded. Although the $350 million in stock that Cisco paid is now worth $500 million, Mr. Verhalen always dreamed of taking the company public.

But what are you going to do? As Mr. Verhalen points out, "The channels of distribution are becoming increasingly important now that the major players are really focusing a lot of their energy on switching. It would be hard for Grand Junction to achieve as much as it could with Cisco's sales force behind it." All in all, Mr. Verhalen claims that he is happy with the change and believes that it has led to a dramatic upsurge in Grand Junction's sales.

In Mr. Verhalen's opinion, venture capitalists need not be afraid of acquisition. Instead, they should regard it as just one more path to liquidity. He thinks that the strong state of the IPO market will continue to ensure respectable multiples for the foreseeable future. If Bay Networks' acquisition of Centillion Networks is any guide, Mr. Verhalen may be right. As Amar Senan of Volpe, Welty notes, Bay paid $140 million even though Centillion was only generating $1.5 million in revenues at the time of the acquisition.

Douglas Leone of Sequoia Capital arguesthat new networking companies must accept these marketplace realities. "In the back of your mind," he observes, "you always have to think about building a designer company -- a company whose likelihood of being acquired is pretty high." In this kind of environment, equity effieciency is key. "If the investment requires $20 million in paid-in capital, and the company is acquired for $140 million, there's not a lot of money there," he says.

Of course, there are exceptions. IPO candidate Xylan, which makes switches for Ethernet, ATM, and Token Ring networks, is "flying straight in the face of Bay and Cisco," according to Seth Neiman of Crosspoint Venture Partners. Backed by Crosspoint, Brentwood Associates, U.S. Venture Partners, Norwest Ventures, and Technology Crossover Ventures, the company is using indirect channels so successfully that one source predicts that it will achieve a $1 billion market cap when it goes public.

Still, most venture capitalists would agree that companies like Xylan are the exception and not the rule. One networking executive is fatalistic when the subject comes to the Big Four. "Hey, it's their world; we just live in it," he observes.

Take a byte out of timeOne of the most promising areas according to Mr. Leone is improving the speed and performance of wide area networks. Mr. Verhalen and Mr. McLean agree: they both state that the thirst for wide area network (WAN) bandwidth will be especially acute. One VC we talked with even went so far as to predict that there will be a "brownout" on the Internet sometime this year. Kevin Fong of Mayfield Fund does not believe that the situation will reach this point, but he does think that there will be another large-size company opportunity in the WAN space in the coming years.

Mr. Fong's special focus these days is gigabit Ethernet switches. In addition to chairing a panel discussion on the subject at Networld + Interop on April 2nd, he is bullish on Granite Systems, a secretive startup founded and funded by Andy Bechtolsheim. As one of Sun's first employees, Mr. Bechtolsheim definitely knows hardware, but he has not opened the company to venture capital investment. One VC questions whether Mr. Bechtolsheim has the networking expertise to grow Granite to its full potential.

Be that as it may, the rise of gigabit Ethernet represents a clear setback for ATM (Asynchronous Transfer Mode) technology. Although the IEEE has not approved the move to a 1,000 megabit standard, a decision seems likely in the next year. What's more, ATM is still mired in standards battles of its own. With only a fraction of the foothold that it has established in Japan and Europe, U.S.-based ATM technology does not appear to have enough momentum to resist its brash challenger.

Mr. Leone is blunt when the subject turns to ATM. Even though Sequoia has several ATM investments, he does not believe the "hoopla" surrounding the long awaited standard. "Right now, ATM is a niche opportunity in the LAN," he says. "ATM is going to happen in the telco world first, and it's going to move slowly down to the desktop. It's a TCP/IP world out there."

At first glance, gigabit Ethernet's appeal is puzzling. As those Silicon Valley veterans who are old enough to remember know, Ethernet was only designed to handle 10 megabits per second; ATM, on the other hand, is much more scalable. In addition, Ethernet has long been known as the "party line protocol" because users could not communicate when someone else was using the line. (The advent of LAN switching has changed this to a substantial degree.) Finally, when one also considers the fact that prices for ATM equipment have been dropping at a rate of 40% per year, it must seem strange to some that so many venture capitalists are so excited about gigabit Ethernet.

The reasons for the shift boil down to price and performance. Although ATM has come down in price, the standard requires much more of an investment in infrastructure than Ethernet, which is already installed in the vast majority of U.S. networks. True, nobody knows who will supply the gigabit Ethernet adapter cards of tomorrow, but since their ATM equivalents still cost over $500, few doubt that Ethernet will be less expensive than ATM.

As for performance, Mr. Fong and his peers are not advocating an enhanced party line protocol but a high-speed Ethernet standard which uses ATM switching technology. Mr. Fong believes that the term ATM has been misused. "Strictly speaking, ATM defines a technology and how to switch," he observes. "But in a larger sense, it includes how packets and cells should be formatted." While the notion of switching TCP/IP packets on ATM hardware may seem like a contradiction, Mr. Fong points out that routers and switches are increasingly converging. He sees a strong market for new companies developing ATM hardware for existing Ethernet networks.

One company that is working this niche on the software side is Ipsilon Networks in Palo Alto. Arguing that inflexible 32-bit IPv4 addresses are restricting the growth of global Internet routing, Ipsilon is trying to introduce a next-generation IP protocol that will work just as well with ATM as it will with Ethernet. (The standard also meshes well with low-bandwidth technologies like wireless.) Venture firms betting on Ipsilon include Mohr Davidow Ventures, Merrill Pickard, Anderson, & Eyre, and Sutter Hill Ventures.

In Mr. Neiman's view, however, gigabit Ethernet's main challenge will be addressing customers' needs. He points out that there are still doubts in the technical community about whether collision detection mechanisms will work properly at 1000baseT, and he warns that customers will steer clear of the technology if they feel that it is too complicated or untested. Although he doesn't question the need for gigabit rates, he believes that proponents of this new standard must be careful to avoid the same traps that ensnared ATM.

The network is the computerAnother area that venture capitalists are exploring is LAN network management services and software. As Mr. Leone puts it, "As the market matures and technology continues to change rapidly, the need for network management...is only going to increase."

On the network management services side, International Network Services has had great success managing LANs for Fortune 1000 companies. Founded in 1991, the Mountain View company counts Sequoia, Sigma Partners, Canaan Partners, and Cisco among its investors. With a $70 million run rate and strong investment bank interest, the company is reportedly contemplating an IPO at a $500 million plus valuation.

As for network management software, it is still a long way from being "plug and play," but Roger Evans of Greylock Management calls it "the sleeper of the networking industry." He is particularly excited about Frontier Software, which he describes as a leader in the emerging field of network traffic monitoring. "Frontier is one of the few network management software companies with the profitability and revenue momentum to be a near-term IPO candidate," Mr. Evans states. Another one of Greylock's investments is Optimal Networks. Optimal offers a product line of Windows-based applications that automatically build models of the customer's network and provide "an integrated view of (a network's) topology and traffic." Although the company is not profitable, it has raised first-round financing of $2.8 million from Greylock and U. S. Venture Partners at a $7.5 million post-money valuation.

Because this kind of software requires tools that model flow through a variety of different environments, Mr Evans states that it is "not accidental that it hasn't shown up until now." Even so, the number of network management software companies seems to be growing exponentially.

One venture-backed company that competes directly with Optimal is NETSYS, which makes software for router-based Cisco networks. The profitable Palo Alto startup is operating at a $10­20 million run rate and closed its first round of $3 million a year ago at a post-money valuation of $14 million; investors include Cisco, Bessemer Ventures, and Sequoia.

Two other companies that are plying different areas of the network software space are Shomiti Systems and ISICAD (soon to be known as Cambio Networks). Shomiti, which received all of its first round financing from Sequoia, manufactures a family of network analyzers and is moving into the ATM market. Chairman of the Board Leone is happy about the company's progress and reports that the company is in the process of closing a second round of approximately $4 million. ISSICAD, which makes software that manages hardware and software assets across heterogeneous networks, is also doing well: according to Mr. Evans, the company has completed two venture rounds and is generating approximately $10 million in revenues.

Mr. Fong says that he is interested in network management software but wonders about larger competitors like Network General, Compuware, IBM, and recent IBM acquisition Tivoli Systems. Mr. Evans also acknowledges that network management software is a tricky area for investment. "Frankly, companies delivering these tools are biting off a substantial technical challenge," he says. "It might take several years before their products are fully refined and capable of delivering enough revenues to go public."

Because the development cycle is so complex, Mr. Evans believes that many network management software companies will find acquisition more attractive. His conclusion is not surprising. Given the nature of the networking marketplace, acquisition seems to be the order of the day. This buying cycle has generated multiples that are more than respectable in the view of many of the VCs we talked to.

So who said IPO was the only way to go?