PROFESSOR MANJUNATH HEGDE IS BACK TO teaching at the Washington University in St. Louis. Only a few weeks ago, he was the Chief Technology Officer at the $160 million Celox Networks, which was into developing carrier-class IP services delivery platform for the network service providers. On December 18, 2002, Celox shut its doors because of the protracted slump in telecom spending. The lone remaining page of the company's website says,“It had become increasingly clear that carrier CapEx spending would not return in the foreseeable future. As a result, we could not reasonably expect to execute on our business plan.”

Hegde, one of the five founders of Celox, says, “When the company was founded in 1999, carrier spending was still high. Right from day one we worked towards the system requirements, design and architecture of our dream switch.”

For Hegde and his team, raising the first round of funding wasn’t any easy. “The dot-coms were fairly well established. The service providers were becoming aggressive and competing with operating companies. We approached 10 to 15 VC firms and managed to get three VCs commit to us. We raised about $7 million in the first round,” recalls Hegde.

The startup grew aggressively thereafter. Headcount scaled to about 300 employees. It did everything that a successful firm would do—expanding to foriegn bases, like China, and recruiting key engineering talent. About 2000 Chinese engineers took up Celox’s admission tests and 50 of them were hired. The company lobbied and arranged H1B visa for these engineers and was successful in getting them to its headquarters in Southboro, MA. “We first went to Bangalore in 1999 and found that the Indian market had been 'farmed' by many U.S. and European companies. It wasn't easy to hire in India then,” laughs the Mangalore-native Hegde.

The company even made strategic decisions and outsourced some of its development work to Wipro, HCL and Microland. At one point there were about 180 engineers from these three companies working on the Celox project. State-of-the-art network processors that contained eight million gates and 20 boards with about 2000-4000 components were being developed at Wipro in Bangalore, HCL in Chennai and Celox's offices in St. Louis, MI. The initial success in developing a chip across three different locations indicated that it was possible to develop this kind of a chip in an offshore manner. “It does require a lot of discipline and a lot of time and energy in terms of managing the team,” says Hegde.

Celox adopted the same model for software development of network protocols. Hegde is quick to admit that he was not happy with the software development. “It was hard to get a well-knit team. Software is more complex than hardware. The discipline required as a software developer is less than that of a hardware developer. The tools for developing software are not as rigid as the tools for developing hardware. This gives enough flexibility for the software developer; a flexibility that comes with the danger of difficult management.”

Celox generated widespread excitement within the telecom industry as it developed a super-high-capacity switch—the Celox SCx 192—that handles six million users (simultaneous sessions) and delivers up to 80 gigabits per second of traffic. That weighs in at 60 times the capacity of the nearest competitors Lucent Technologies and Cosine Communications. The router was supposed to ease the deployment of IP functions in a service provider network. The switch was targeted at carriers, Internet service providers, application server providers and others.

Like any other product company, capital was needed to sustain the product development. In its second round, the company raised $68 million from three VC firms. Since it was towards the end of the technology boom (March 2000), Celox managed to get extremely high valuations. In July 2001, Celox raised $80 million in the third round of financing. “The beta version of our system was ready and we had got customer traction. Our product was successfully tested for 5-6 months at one of AT&T’s divisions in Chicago,” says Hegde. The trial with AT&T had validated Celox's technology. From a startup's perspective, it was indeed a milestone to get in early to a big service provider like AT&T and understand what their needs are. “People were seeing the telecom downturn as a temporary gap and were expecting the market to come up in 6-12 months,” recalls the founder.

However, the VCs were willing to commit more funds only if there were signs that the money would last for a long time, in light of the telecom market slow-down. “There was no way of reducing the burn rate other than reducing certain fixed costs. We decided to put a hold on some of the additional features for the system and have a smaller engineering team. We had to lay off 65 employees in October 2001,” says Hegde.

The third round of money was spent on testing the systems and for additional digital features requested by the customers. Testing a high-end router like the SCx 192, which could handle a maximum of 6 million IP interfaces, was no ordinary task. “When you have a scale like that, reliability is difficult. Testing for reliability is both difficult and expensive,” quips the CTO.

Celox had spoken to almost all the carriers in the U.S., including some of the biggest spenders in the space. Despite a successful trial and good feedback, AT&T was unable to ink any deal. And AT&T's budget cuts in August 2002 came in as a big jolt to Celox's plans.

Events took a turn for the worse when decisions by potential clients such as Verizon Group, BellSouth and Phoenix were delayed. One government agency that had shown initial interest was slow in taking any decision. Interest was shown by some Japanese carriers as well. In fact, one of the carriers had several rounds of talk, and when Celox hoped the deal would close, the Japanese firm sold its broadband business unit. “We also spoke to many small carriers, many of which had plans in the early days of 2002 for deploying our solutions. But they reversed their plans during the second half of 2002,” comments Hegde.

Probably Celox wouldn't have shut if at least one of the deals had come through. “Essentially we still do not know when the carrier spending would come back even to the level which it was at last year. But certainly for the next year it is yet another round of budget cuts. Taking these into consideration and the fact that we had limited resources, we decided to close our business,” says Hegde, who as CTO, would go on sales calls where scientists from service providers would ask critical questions related to architecture and performance of the SCx 192 system.

“The single biggest issue we did not succeed in resolving was to estimate market timing. Alcatel, Cisco and Nortel had evaluated our product. Even they, with all their sales and knowledge of the market, could not estimate the market’s need for SCx192. Till December 2002, Alcatel was keen on buying our technology. But, like all others, Alcatel too was worried as to when the market would need a system on the scale that we had built. The market has been receding by the month,” recalls Hegde. “When we started in 1999 we expected demand for our boxes in 2002. That has not happened. It is hard to judge the market and believe circumstances. But, certainly there is always a possibility of huge upheaval, like what occurred when we launched two years back. It is very difficult to predict the market,” sighs the founder.

Lessons Learned
Celox had raised $160 million in a total of four rounds of funding. There were in excess of 30 VC firms who had invested in the company. Decision to shut operations was taken knowing that raising money this year, or 2004 was not going to be easy. Hegde agrees that the founders did lose control of the company because of the amount of capital raised. “For the kind of product we were building, we required huge investments.” He further adds, “Whenever one raises money one should have a clear vision of how the company is going to be self sufficient. When we raised money, our attitude was “you build it and they will come.” There was so much apparent demand for it. There were expectations that a big company will come along and integrate our product and company into their structure. That may not happen always. One should examine the business aspect of his product. He has to take the saner view that if he does not get any money, how will his product work? How can he create a business out of his product?” Though the founders lost control of the company, all six of them were with Celox till the last day, an indication of the passion and commitment for the technology they were building. Hegde shares his hard-learned lesson on changing market dynamics. “A company like IBM can afford to make mistakes. They have the flexibility to reorient and resegment into other product line. A startup cannot have the same luxury.”

Hegde also stresses the importance of feedback from customers early on. “A person who gives feedback or advice is a person who shows commitment. In fact, he also has a certain responsibility for the advice given. It is true in real life and more so in business. We didn’t do enough of that in Celox early on. We didn't go to AT&T in 1999 and ask them to tell us what they wanted,” says Hegde, sadly.

Celox Networks is the latest in the list of venture-backed technology companies to pull the plug last year. The company shut doors even before it could sell anything. Celox, which means “swift” in Latin, lived up to its name in the speed to build, and its speed in closing its doors.