Blindly following IBM’s lead led Madge Networks into a cul de sac. And competing in the mainstream markets is a doomed struggle against the networking giants. So where does the company go from here?
Get ready, said the slogan on Madge Networks’ web site at the end of 1997. Data, video and voice are rapidly converging to dramatically change the way you do business. Few people in the networking business would argue with the truth of this slogan. But Robert Madge, founder and chief executive officer of the British, but Nasdaq-listed supplier of networking equipment, might sense a certain irony as he reads it. As he looks back on not one but two deeply troubled years for Madge Networks, he will be aware that his own company’s business has indeed changed dramatically, but not in the positive way he intended. And the reason his plans went awry? Data, video and voice did not only fail to converge as rapidly as expected. In most businesses, they did not converge at all. We tried to do voice, video and data over a local area network and we made a lot of progress, says Madge, who embarked on an ambitious plan to drive the market in high speed, multimedia networks. This involved a big investment in the emerging asynchronous transfer mode technology. The backing of ATM in all areas of the network was strategic initiative made by IBM Corp. IBM failed to push ATM in the local area network, and now the industry has moved on, instead choosing high speed Gigabit Ethernet concentrating on adding network quality of service and multimedia support into internet protocol.
Timing was bad
Madge understandably partnered with IBM, having historically been a major supplier of Token Ring equipment. Token Ring started out as a proprietary IBM networking protocol, and Madge grew up with a culture of following IBM’s lead. So it blindly followed IBM with ATM to the desktop, which was seen as the natural way out of the networking cul-de-sac of token ring. But only a small percentage of customers are currently demanding that kind of infrastructure. Our timing was bad. It was too ambitious. In retrospect, Madge admits that the decision to focus on high-end, high-speed networking was hopeful. But at the time, it seemed to make sense. The company was awash with cash. It employed many outstanding technologists, and the executives were aware that sales of the token ring local area network cards, on which Madge Networks built its business, must soon slow down. So the company expanded into Ethernet, the rival and more widely accepted local area networking technology, and threw itself into asynchronous transfer mode, the transport technology which, most analysts were forecasting, would eventually supplant all others in both wide and local area networking. As it has turned out, ATM is important for wide area network backbones, but has not been adopted for the desktop. Nor has its ability to give guaranteed quality of service yet led to its domination of wide area networks, although most analysts still think this will happen. Instead, existing technologies, such as Ethernet, have been given a new lease of life through faster implementations. And even now, voice, video and data have not converged in corporations. The result is that from 1996 onwards, Madge found itself with products that were ahead of the market, but which few people wanted. And rather than enjoying the fruits of its investment in high-end technology, it found itself competing head on with giants such as 3Com and Cisco in the mainstream market. And frequently, admits Madge, customers preferred the big, end-to-end suppliers. Sometimes technology matters less than account control. The figures tell the story. Madge has reported losses in every quarter but one since mid 1995, while revenue growth has gone into reverse. In Madge’s third quarter, revenues were down 25% to $83.2m, while losses were $14m. But Madge also took a one-time $48m charge to restructure the company. This involved closing its Silicon Valley headquarters just six months after it moved to San Jose (although it will keep a development center there), killing off its MadgeOne end-to-end networking strategy, and laying off 700 people. An ATM backbone switch, announced in April, will not be developed further. Madge isn’t the only company to make these mistakes. Attempts by other second-tier data networking companies to expand into the big time have also failed. IBM has had its original equipment market destroyed by Cisco Systems Inc, Digital Equipment Corp has sold its Datacoms business to Cabletron Systems Inc, which is itself having to take some strong medicine to straighten itself out (CI No 3,313). All the companies have at one time or other tried to play the end to end networking game, and have been squeezed out by the big three players, Cisco, Bay Networks Inc and 3Com Corp. Madge has gone back to its roots, focusing on token ring network and ATM cards (60% of revenues), the Ethernet software products that came with its 1995 acquisition of Lannet (30%), and providing some specialist high-speed video networking, mainly using ISDN (10%). Madge has chosen to extend itself into video networking, having launched its first products last year(CI No 3,034) with ISDN and LAN based video networking equipment as it reckons that its video revenue will grow by 100% a year, and as a pioneer it will be able to take a sizable slice of a fast growing market. But where does Madge go next?
Immature market
Our goal is to only be in markets where we can be a leader. We had got ourselves into a market where we couldn’t lead. Madge’s aim is to lead in video networks and collaborative conferencing – as yet an immature market. In the meantime, with Ethernet software revenues flat, Madge may get a lifeline from an unlikely source. The slow adoption of asynchronous transfer mode means that many big customers are looking once again at re-investing in token ring. Madge, along with IBM Corp, Olicom A/S, Cisco Systems Inc and others, is working to get new standards adopted which will result in fast token ring products being quickly developed. 100 megabits per second token ring products should be available late this year with one gigabit products a year or two later.
