Into the darkness
In the first of two articles on the history and future of the fibre market, E&T looks back to the first fibre-optic network boom and bust and asks what lessons can be drawn.
In the heady days of the late 1990s, many believed that the Internet would revolutionise the world in ways that could scarcely be imagined. It would make communications easy and ubiquitous, and render old ways of communicating redundant.
The Internet revolution went well beyond engineering: people changed their business models and lifestyles. There was a sense of barriers breaking down. Nothing seemed impossible. For some, the Internet became almost a faith. Companies based on the Internet were much more highly valued than 'old economy' rivals, even if they had never turned a profit. Venture capital funds (VCs) threw money at Internet projects like it was going out of fashion. As one entrepreneur says of those days, "All you had to do was say the word 'Internet', and the VC wrote out the cheque."
At the outermost stretch of this bubble, billions of dollars were spent installing fibre-optic networks that, it was imagined, would be needed to carry all that Internet traffic - and billions were lost.
"The investment in submarine fibre-optic networks peaked in 2001 at around $16bn," says Julian Rawle, managing partner, Pioneer Consulting. This compares with an average annual investment in submarine fibre since 1986 of just $2.4bn. Investment in terrestrial fibre-optic networks was similarly out of control.
"In 2001 there was so much money sloshing around looking for a home, people were not behaving rationally," says Rawle. "All that money was waved in front of their noses, and it made them go crazy."
The collapse of KPNqwest
One of the biggest failures was KPNQwest, the Netherlands-based joint venture between Dutch national operator KPN and Denver, US-based telco Qwest. KPNQwest's strategy was to combine state-of-the-art fibre-optic networks with the customer base of EUnet International, whose Internet pedigree dated back to the first European Unix networks, and which had been bought by Qwest in April 1998.
KPNQwest borrowed billions and built a state-of-the-art, high-capacity European fibre-optic network that ran for 25,000km through 18 countries. The company swelled to 2,500 employees. KPNQwest had about 100,000 corporate clients, including Microsoft, Cable & Wireless, and many Internet service providers. Some estimates say that the KPNQwest network carried between a third and more than half of all European Internet traffic by 2000. Two years later the company was bankrupt.
KPNQwest went bust because it overestimated demand, borrowed too much, built too much, and was unable to pay back its banks. It filed for bankruptcy in May 2002. Before the coup de grâce, executives tried frantically to find the money needed to keep the network running. Rumours circulated that Russian oil giant Yukos was about to buy it. Other interested parties were said to include Dublin-based telco eTel, and US investment bank Lehman Brothers. All negotiations failed.
The network remained in operation for some weeks after the bankruptcy, but with no money to pay suppliers, in July 2002 it was officially shut down. For a while, interested viewers could log on to a webcam at the KPNQwest Network Operation Centre and gaze at the sad sight of an empty room with dozens of terminals, but no people.
The Dutch, German, and Belgian parts of the KPNQwest network were eventually purchased by Dutch parent company KPN. Other parts were bought by the UK-based Interoute, which now runs the largest fibre-optic network in Europe. The Austrian and Swiss parts of the KPNQwest network went to local operators in each country. Operators refocused on their local markets, leaving continental and trans-continental fibre links to long-established players such as AT&T and BT.
Qwest, unlike its joint venture KPNQwest, never filed for bankruptcy but was left with massive debt, as was Level 3, another large fibre owner. AT&T, Verizon and Sprint managed, through foresight or luck, to sidestep the fibre bubble. AT&T was never a big builder of fibre, but both AT&T and Verizon bought up fibre assets when the bubble burst, as did Reliance and Tata. Sprint does not own much long-haul fibre, even though it built one of the first US fibre-optic networks, preferring to lease capacity.
Another victim of the bubble was FLAG. In 1996, a consortium began laying a 28,000km Fibre-optic Link Around the Globe. FLAG became the world's longest cable network. It stretched from Cornwall in the UK to Spain, Italy, Egypt, India, Thailand, Hong Kong, Korea and Japan. The idea was to provide capacity for connections to the Far East. Side connections were built to the United Arab Emirates, Malaysia and China.
On 2 April 2006, Flag Telecom Holdings filed for bankruptcy protection. It had been unable to cope with insufficient demand, over-capacity, and $2.6bn of debt. FLAG was later bought by a subsidiary of Reliance Infocomm, the Indian mobile network operator, for $207m. The largest shareholder in FLAG prior to the takeover was Verizon.
"Verizon (formerly Bell Atlantic) divested most of its interest in FLAG, but FLAG is still healthy and provides excellent trans-oceanic connectivity to numerous countries," says Stu Elby, vice president of network architecture for Verizon's technology group.
Other giants who came crashing down as the Internet bubble burst were WorldCom and Global Crossing. WorldCom's bankruptcy in 2002 was the largest in US history -until the collapse of Lehman Brothers. WorldCom mis-stated $3.8bn of expenses as capital costs. It broke accountancy rules to keep profits looking healthy. Its former CEO, Bernard Ebbers, was sentenced to 25 years in jail. Like FLAG, WorldCom was eventually rescued by Verizon, which bought it for $7.6bn in July 2006.
From 1999 to 2000 Global Crossing grew from a small company of 250 staff to a giant of 14,000 employees with an international high-speed fibre-optic network. It filed for bankruptcy in January 2002. Its former vice president of finance, Roy Olofson, said the company had become involved in swap transactions including a $100m exchange of capacity with Qwest. The company denied any wrongdoing. Global Crossing was broken up after its bankruptcy. Part of it became Asia Netcom, today known as Pacnet. Another part became Pacific Crossing. Today Global Crossing is owned by Singapore Technologies Telemedia and employs about 5,000.
Some suppliers also caught Internet fever. Tyco Submarine Systems was spun off from AT&T as a supplier of submarine cable systems, and decided to build its own fibre-optic submarine network, the Tyco Global Network (TGN), in competition with its customers.
"It seemed a crazy idea on the face of it," says Rawle. "And it was." Tyco nearly collapsed because of the TGN venture, which was later acquired by Tata, the Indian giant. Dennis Kozlowski, the former CEO of Tyco, was convicted of fraud and misappropriating more than $400m. He was sentenced to jail for at least eight years.
The downturn hit other suppliers. KDD Submarine Cable Systems, which built FLAG and many other systems, exited the market in 2001. Pirelli's submarine cable arm was bought by Alcatel, which now dominates the market alongside Tyco.
One symptom of the Internet bubble was an overemphasis on bandwidth trading. A paper by Chris Kenyon and Giorgos Cheliotis of IBM summed up the new philosophy: "As bandwidth becomes a commodity … the dynamics of network markets are set to be the basis for spot and forward prices."
Fibre capacity was recast as a tradeable commodity. Telecoms executives hoped to marry their world to the apparently more glamorous world of broking and banking. Many who pursued bandwidth trading did so in the belief that if telecoms sales became more like the financial markets, they would become more efficient. They dreamed of swift, dealer-type trading replacing the lengthy contract negotiations typical of the telecoms world. But their plans should have been questioned.
A network connection from A to B is a very specific offering. Engineers know this, and know how to work around the issues. Financial markets, by contrast, depend on the interchangeability of the items traded: one barrel of Brent crude oil has to be the same as another in order for a futures market to work. But bandwidth doesn't work this way.
Bubbles often attract people with little experience in the expanding sector. The largest player in the bandwidth trading market of the late 1990s was Enron, an energy company. In 2000, Texas-based Enron reported $101bn in revenues and employed 22,000 people. It had been named America's Most Innovative Company by Fortune magazine for six years running. In 2001, it sought bankruptcy protection.
Bandwidth and fibre trading was not responsible for Enron's collapse. Enron's problems were due to the fact that much of the 'profit' it claimed to have made was generated by trading with itself. But bandwidth trades provided one way for fast-growing companies to inflate their earnings. Enron and Qwest were close trading partners and in 2001, the two companies transacted a $500m swap that included dark fibre between Salt Lake City and New Orleans, despite the fact that both companies had overcapacity on that route.
During the bubble, people thought that the Internet would achieve far more, far more quickly, than it did. They built too many fibre-optic networks, flooding the market with unwanted capacity. After the fall of Enron, prices of bandwidth, including over fibre, collapsed. Operators watched as prices for bandwidth tumbled to a fraction of what they had been when they built the networks. Fibre capacity was being sold and resold for much less than the cable had cost to build, sometimes because the seller desperately needed cash.
Some cities were connected by thousands of fibre pairs, most of which were left dark. The prices for dark fibre became so low that some customers, including banks and international corporations with large data transfer needs, began buying up dark fibre and lighting it themselves.
According to Stephan Beckert, a researcher at consultancy TeleGeography, prices fell up to 70 per cent a year on major routes between 2000 and 2003. Between 2003 and 2005, carriers hardly added any capacity to networks. Though price cuts slowed after that, TeleGeography saw prices for monthly leases on 10Gbit/s links between Miami and New York City fall from around $75,000 in 2005 to below $30,000 at the end of 2007. Prices for 10Gbit/s connections between New York and London fell by 80 per cent from 2002 to 2007, according to TeleGeography.
During the downturn, no-one dug up their fibre infrastructure - they just left it unused. And most of the world's submarine fibre-optic network capacity remains unused. According to Rawle, only 18 per cent of world submarine fibre is lit, up from 13 per cent in 2001. This low use is mitigated by the need to include redundancy. A number of submarine cables have been damaged recently, demonstrating the need for back-ups.
Engineers often ask management for more capacity when they have filled about 50 per cent of current capacity. Today, telecoms companies are laying fibre in some regions, but prices remain low.
What lessons can be learned from the fibre boom and bust? Massive infrastructure projects can be highly risky. Getting them wrong can have long-term consequences. When large sums of money are at stake, driven company executives can be tempted to bend the rules to maintain the appearance of success. And that new technologies often take longer than expected to deliver their promise.
The Internet promised to change the world and is doing so, but not quite as fast as had been imagined.