Top Liquid Bar





The Future of Bandwidth Trading

Everybody wants to trade bandwidth. But what’s the underlying?

By Barclay T. Leib Last December Enron Bandwidth announced that it had inked the world’s first forward bandwidth contract. Wall Street liked the idea – to put it mildly. Within six weeks, parent Enron’s stock price jumped from $38 a share to $70 a share, adding a cool $23 billion to its market valuation. That’s a lot of bang for the PR buck. The lesson wasn’t lost on competitor Williams Communications. After months arguing that bandwidth wasn’t a truly tradeable commodity, it announced it was rolling out its own bandwidth trading initiative in mid-February. By now almost everyone has heard some new-economy pundit forecasting that bandwidth is poised to become the biggest commodity market of all time. After all, while traditional voice telecommunications traffic continues to chug along at an annual growth rate of roughly 7 percent, demand to send data and voice calls that are broken down into data bits over clear-channel fiber-optic networks is said to be doubling every 72 days.

Add to this the nascent (but potentially huge) market for video-on-demand services, and the appetite for “long-haul” bandwidth capacity could easily burgeon into a $700 billion global industry. Right now, both established and start-up telecom companies are madly laying new fiber cables across the world to meet the continued step-up in demand. And old-line investment banks. including Goldman Sachs and Morgan Stanley, are ramping up dedicated broadband units to make sure they are not left behind.

But where the hyperbole stops and the reality begins is another question. A well-organized and standardized bandwidth spot market is months or years away. And without a spot market, a derivatives market will remain little but a (fiber-optic cable-filled) pipe dream.

While everyone talks expansively about this potential new market, few are completely sure what market they are really talking about. Are they talking about trading the actual fiber strands in the ground? Or simply the right of passage over those strands for a period of time-something often referred to as “telecom applications”? Some argue that the biggest commercial opportunities won’t be in the long-haul transmission of broadband at all, but rather in the hooking up of those strands to other people’s networks. That would make the physical space at various telecommunication hubs (typically referred to as co-location sites) the crucial commodity in short supply. At present, this type of real estate trades at five to six times the cost of normal office space.

Then there is the question of price. The value of at least one low-end bandwidth application-voice minutes-has done nothing but spiral lower for the past several years. And in spite of the burgeoning Internet demand for bandwidth, the cost of data passage over fiber networks may also be headed sharply lower. Not surprisingly, the forward pricing of bandwidth-if one could find a definitive curve-is sharply backwardated. Anyone care to price a long-dated broadband option in such an environment?


All this suggests that would-be bandwidth intermediaries might do well to question whether old business models used for oil and gas trading or even electricity trading necessarily apply here. “The bandwidth market is not yet a trading market. It is today a procurement market,” explains consultant Lynn Franks of Andersen Consulting in Houston. Unlike traditional commodities as we know them, bandwidth, Franks points out, is not used up once transited, nor are all the various technical interfaces particularly fungible. Capacity doesn’t go away once used, and different interface speeds and routing possibilities leave literally thousands of possible traded products. (Do you need an OC-3 line into Lima, Peru, or would it be better to have an OC-41 fat pipe, or might an STM-1 circuit suffice?) Some even think bandwidth lends itself less to a traditional exchange-traded product and more to an electronic auction platform such as an E-Bay.

In such an environment, it is thus not surprising to see one early market entrant, Band-X (, putting together a number of different kinds of transactions rather than focusing on any one narrow aspect of broadband.

“There are a lot of people out there claiming that they are doing bandwidth trading, but it’s a little bit like the emperor’s new suit-a great deal of hype but very little substance.”
-Richard Elliott, Band-X

The London-based electronic broking firm was conceived at a private London tennis club on a sultry summer day in 1996. Richard Elliott, an investment banker covering utilities for Kleinwort Benson, was matching his form against Marcus de Ferranti, a former Harrier pilot working for the British government on guided missile systems. The post-match conversation somehow slipped to the Internet and this thing called bandwidth. Surely, they thought, this must be a commodity of some sort being traded behind the scenes, but was it being traded efficiently? Was there room for a market-neutral middleman to facilitate the process?

Their first effort at a web site was a simple bulletin board with two buttons labeled, “Buyers Enter Here” and “Sellers Enter Here.” Four years and several site redesigns later, Band-X is an active trader of, among other things, switched voice minutes between geographic locations on its facilities-based clearing exchange. This places Band-X in the middle of phone calls, rerouting them across different networks through its physical telecommunications center. The switched voice market typically trades in one-year contracts of leased capacity, quoted on a monthly cost basis, although shorter contracts for seven, 14 or 28 days are starting to emerge, as switching capacities improve.

In other corners of the market, Band-X will help arrange anything from co-location space to a long-term dedicated OC-3 line into India. This business arguably is a more interesting one for Band-X than its facilities-based trading. The firm charges 1 percent of the contract price to close such a deal, with the seller of the services bearing this cost. Much of this business is done on a wholesale reverse auction basis, with auctions taking place on the Band-X web site every two weeks or so.

Although many would argue that the synergy between minutes trading and bandwidth trading is low (certainly the technologies are very different), others see the former as a natural jumping-off point as they await improved bandwidth trading standards. One such company is New York-based Arbinet-theXchange, Band-X’s biggest minutes-trading competitor. Acting on a software model rather than a commission model, Arbinet charges an access fee for major corporate and telecom companies to hook up dymanically to each other in real time-exchanging actual phone calls, not just the capacity for a given number of minutes. Are you a long-distance carrier that wants to capture more traffic into Australia? Just lower your price offered for that route on Arbinet and the next series of calls will be flowing your way.

Arbinet is not yet trading in bandwidth, however-for two reasons. First, the company wants to secure its presence in the minutes market before leaping into new forays. And second, according to chairman and CEO Anthony Craig, “There’s no standardized way to count data flow yet. Oil has barrels, gold has ounces, but no one is sure yet how data-flow quality and quantity are going to be measured, and by whom. Once these standards are there, so likely will we.”

Band-X claims to have patented technology to trade specialized packet routing on standard terms, but whether it will gain broad market acceptance is another question. “We’ve tried not to be too missionary and say ‘Here’s a system, one day all communications companies will use it,’” says Band-X’s Elliott. “There are a number of operators in this space taking that line today. To the contrary, Band-X has always tried to offer what is understandable to and useful for the wholesale players at any given time. That’s why we have the revenues we do.” Band-X earned $6.5 million in gross revenues through the fourth quarter of 1999, helping facilitate all sorts of global telecom capacity usage.


And capacity usage is the name of the game. Traditionally, telecom companies have spent considerable time negotiating among themselves on shared capacity. By the time a deal is struck and the facilities hooked up, three to nine months may have transpired. In some cases, the purchased capacity isn’t even needed anymore when all the paper work and physical connections are finally complete. After much hard work, one side occasionally tries to back out of a deal.

All of this costs a considerable amount of money, of course. While sales and general administration expenses are estimated in most American industries to equal just 16 percent of revenues, the figure runs closer to 25 percent in the telecommunications industry. There is tremendous waste and duplication of effort, as marketing agents traverse the world looking for customers and partners to use their networks. The question then becomes, Why should a telecommunications company with a cost-efficient network spend considerable time and resources selling capacity directly when increasingly it can find an outlet for this capacity in a brokered market or electronic exchange?

This makes Band-X’s biweekly auctions particularly popular and useful. A client will approach Band-X with a specific buying interest, whether it’s a need for leased space over a one-year term or a dedicated line for a 25-year term (something referred to in the industry as an IRU, Irrefutable Right of Usage). Alternatively, the need might be for a certain right of way to lay cable or co-location space. Band-X will organize a list of potential sellers of this capacity-typically prescreened for creditworthiness and capability to perform on the contract-that will be known and acceptable to the buyer in advance. Then, via the Internet, Band-X will hold a reverse auction where the sellers will compete to show the lowest cost to the buyer. These auctions are never visible to the public, so how much of a discounted price the buyer is able to receive is never known outside the immediate circle of participants.

“There are a lot of people out there claiming that they are doing bandwidth trading,” explains Band-X’s Elliott, “but it’s a little bit like the emperor’s new suit-a great deal of hype but very little substance. I’d estimate that in reality only 20 or so big deals are getting done per month, but certainly some of these are done through our auction process.”

“Right now this market is missing two crucial elements. The provisioning time to deliver bandwidth is too slow and too variable, and the quality of service between carriers is ill-defined.”
-Julian Harding, Tradition Financial Services

Although many telecom executives deign to discuss Band-X or other trading platforms, at least one New York-based account manager states that he “was pleasantly surprised by the number of European telecom companies looking for prices on Band-X. I sold some capacity via the exchange, and the deal went very smoothly. It’s certainly not my only outlet to sell our network, but if all these other companies are going to be there, it’s far easier for me to sell to them there than to market each of them individually.”

The problem, of course, is that this individual’s actual job is to market his firm’s capacity. If doing so is going to be as easy as hitting a bid on an exchange, many marketing managers may feel their very job description is threatened. Gone will be the expensive business trips through Scandinavia and the Continent. Now buyers can be found at the click of a mouse. Provisioning for service may come in days or perhaps even in minutes instead of the previous multi-month lag.


This leads to another fundamental problem: While Band-X and others are trying to commoditize telecommunications trading, they first must deal with the virtual absence of a trading trading mentality and culture among telecom executives. With the exception of Arbinet’s real-time platform, there is still little that is “firm” in telecom minutes trading when two counterparties agree to a transaction.

“Although in minutes trading, settlement has gravitated now to four-day delivery,” explains Richard Kates of, a New York-based competitor of Band-X and Arbinet, “either side can walk away from a deal at any time, and say, ‘Sorry, I don’t need your minutes now,’ or ‘Sorry, my network went down and I can’t deliver the minutes.’” In order to improve this situation, Kates’s company has worked with the Communications Clearing House Association (CCHA) to develop both a “soft” contract and one with more standardized trading teams-inclusive of severability terms for non-performance. Band-X, for its part, has a publicly posted “blacklist” of counterparties that back out of trades without sufficient cause.

Julian Harding, co-CEO of broker Tradition Financial Services, a Stamford-based global commodity and financial broker, has seen it all before. “In the early days of electricity trading, it was exactly the same,” he says. “I remember when we were working on a deal between one experienced power marketer and one neophyte utility. For days, we had been bringing the two sides of the transaction closer together. Then when the experienced marketer finally dealt on this utility’s bid, the response was, ‘I gotta check with a few guys on this end. I’ll call you back in a bit.’ Our broker started to panic as the marketer grew concerned as to whether the deal was done or not. When we finally heard back from the power company, the individual says, ‘Well, no, we ain’t gonna do that right now.’ When our broker protested, the individual told us, ‘Now listen here, we’re just good ol’ boys down here. We don’t know much about this trading stuff.’”

In that instance, Tradition had to stand up and help shape the market. “We didn’t deal with that utility company again, and we cut the power marketer a substantial check to let them know how serious we were about our obligations as broker,” he states. For its part, Tradition is in a non-disclosure phase of some sort of joint venture in bandwidth broking. Don’t be surprised to see it emerge as a late entrant with a significant presence.

Not surprisingly, one well-known power-marketer, Enron (, is also at the forefront of trying to shape standards for bandwidth trading. The company is marketing its private network of over 14,000 miles of fiber cable for specialty streaming video use by major corporations. Most recently, it entered a movie-on-demand provisioning agreement with Blockbuster Video, and since the spring of 1999, has had a set of standardized terms and conditions for bandwidth trading out for comment. From this initiative, the Bandwidth Trading Organization, a 14-member industry panel working under trade group Comptel’s auspices, is now trying to refine a final ISDA-like document acceptable to all.

Harding of Tradition thinks that the industry is close to establishing an acceptable master agreement, and none too soon. “Right now this market is missing two crucial elements,” he states. “The provisioning time to deliver bandwidth is too slow and too variable, and the quality of service between carriers is ill-defined.” Harding says that when a bandwidth trade is initiated it is impossible to definitively say when the start date is-there are simply too many “physical switch problems, local loop connectivity issues, union sensitivities as to who is physically allowed to make a switch connection, and other technological differences.” He believes that delivery time and quality are the two principal terms that need to be addressed by the industry before true commoditization can occur. The mission statement on the web site offers a similar view.

Yet Harding does not believe that its fast start to date and its attempt to carve out a lead role in establishing standard terms and conditions-will come to dominate the bandwidth-trading world. “The company has taken a proactive leadership role to date, which we applaud,” he says, “but no one group is likely to fully dominate a marketplace.”

Enron’s model places a strong emphasis on paired-city trading and neutral pooling points to connect and measure the quality of the broadband flow across different networks. It is clearly a market-maker’s trading perspective, steeped in the firm’s past trading success in oil, gas, and electricity. Two other companies, San Francisco-based RateXchange and Washington, D.C.-based LighTrade, are forming similar point-to-point models in the rush to develop this market (see “Building the Bandwidth Infrastructure,” Page 24).

But key members of the bandwidth team at Williams think the automatic jump to outside vendors and exchanges to bring the industry together is premature. “To a large extent, we have already been ‘trading’ bandwidth for multiple years,” says Chris Lemmer, head of bandwidth trading at Williams, “but generally we have been doing so with a traditional telecom perspective of risk mitigation and risk management in mind. I’m not sure we need outside vendors to do what we can do anyway over our pre-existing infrastructure.” Lemmer only hedges himself by conceding, “If the exchange efforts prove useful, we’ll take a look at them eventually, but we are not doing so right now.”


No matter how the spot market for bandwidth sorts itself out, most industry participants believe it will do so over the next year or two. As fiber is laid, connections at pooling points are established, and broker and industry standards are promulgated, most would concur that a viable two-way market for global transit on the network is likely to emerge. And that will pave the way for derivatives.

“The spot market has to come first,” says Band-X’s Elliott. “Then a forward market will probably emerge in clear-channel circuits capable to move IP packets. But in the long run, it’s the packets themselves that are the real commodity.”

“We have already been ‘trading’ bandwidth for multiple years. I’m not sure we need outside vendors to do what we can do anyway over our pre-existing infrastructure.”
-Chris Lemmer, Williams

When that day comes, options on bandwidth transit may be a natural. Already, many telecom companies offer their clients special clauses for “bursty capacity.” This means that for peak-usage periods or special events, they can burst above their normally allotted bits per second usage. To be sure, if clients use extra capacity, they must pay for it, but there is a certain amount of optionality built into this structure. The telecom companies are natural sellers of such optionality and corporates are natural purchasers, but right now the option is just treated as part of a larger contract. “No one’s paying for this option now. It’s just there as a courtesy,” explains Nick Cioll, vice-president of “But all the concepts of options as we know them in other markets are the same here. They’ve just changed some of the terminology.”

The only question may be: At what volatility will bandwidth be priced? Because it has fewer variable costs and more fixed costs than electricity and energy products, some think bandwidth’s price will be less volatile as well. But demand can also erupt instantaneously. “When Victoria’s Secret announced in the middle of the 1999 Super Bowl that it was putting its models up on-line, so many people went to take a look, the Internet slowed to a crawl,” says Franks of Andersen Consulting. “I think the already-extreme volatility we see today in electricity trading could pale in comparison to the demand-driven spikes we’ll see in bandwidth.”


Amidst a plethora of other details that need to be resolved in bandwidth trading is what access, if any, corporate clients should have to the wholesale market.

One competitor to Band-X is, a McLean, Virginia-based company staffed by a group of professionals more telecom-oriented than either Band-X’s Richard Elliott or Marcus de Ferranti. Although Band-X does not aggressively market directly to corporate clients, it does accept their participation. Sony, for example, could come in as a buyer of network capacity and ask the telecom companies to bid for its business. The principals of consider this strategy problematic.

“There is no doubt that someday telecommunications companies will be on-line offering corporate clients point to point or point-to-broadcast capabilities for any application that a customer wants to pay money to move,” explains Andrew Romans, chief information officer of “But this is their business, not ours. We don’t want to get into the retail business and compete directly with our customers. Instead, we want to be in the wholesale application business. We see no need to upset the telecom companies by going after their bread and butter and killing their margins.” platform just went live in late June. Starting with the minutes market and a key strategic alliance with Bank of America, theGTX hopes to provide not only automatic connectivity, but also automatic provisioning of capacity and payments. In other words, hopes to be the “inside” market for data and voice applications, and allow the telecom companies to deliver the application over the last few miles of connectivity to the customer premises. The firm considers itself analogous to a foreign-exchange Reuters Dealer terminal between banks, while the banks in turn, sell more customized transactions in foreign currency to outside clients.

The company is a bit late to the party relative to Band-X or Arbinet, and lacks certain of the Band-X patents to route IP packets, but one wonders if might receive additional telecom support given its less threatening posture to the traditional telecom carrier business model.



If a forward market on bandwidth transit does develop, it would allow telecommunications companies to hedge some of the risk of investing hundreds of millions or billions of dollars in building and operating their networks. What is today common practice in many project-finance deals could migrate to the telecom arena. A carrier wanting to borrow money to build out its network may be required by a bank lender or bond investor to sell a certain amount of bandwidth forward, thereby locking in future revenue well before the fiber is even laid. If there’s a way to hedge residual risk away, it’s a safe bet Wall Street will demand that it be used.

And down the road, the arbitrage opportunities could be endless. Investment bankers potentially could buy up big pipes of fiber, hedging their purchases with forwards or futures, then cut up these big pipes into smaller more marketable pieces-something referred to in the industry as “channellizing.” Peak versus non-peak trading also will surely exist, just as today helps fill up airline seats that would otherwise go unused. Different grades or quality of connectivity will also be appealing to different classes of users, opening up the possibility of spread trades between the two.

Even “time arbitrage” between locked-in longer-dated capacity and immediate market needs should be a business that is of some interest. One need only be careful not to load up too much on too many longer-dated Irrefutable Rights of Usage at the wrong price, and get stuck with them. This is something that a company called Star Telecommunications managed to do a few years ago, only to watch this capacity fall in value to less than a tenth of Star’s locked-in cost. As one might expect, the stock in Star has fallen as well-it’s currently trading at just 7 percent of its 1998 high.

Lastly, with no degradation of fiber optic signals across different potential routes, imagine as well all the possible permutations of arbitraging the most efficient path for getting a signal from here to there. It’s nothing short of a Wall Street quant’s optimization dream.

Harold Kamins and Jules Putterman, two principals at Morgan Stanley, are currently working madly to secure Morgan Stanley’s nascent toehold on this market. The firm previously did a first-rate job of penetrating the electricity business-while many others on Wall Street got left behind-and clearly wants to do it again in this new space.

Meanwhile, at Goldman Sachs, Oliver Frankel, a managing director and commodity strategist in its J. Aron subsidiary, has been the recipient of the bandwidth chalice. “The market is potentially huge-it’s amazing,” he says, “but then again, it’s still early days.” Frankel sees all the various point-to-point efforts by Enron, Lightrade, and RateXchange as encouraging, “but if these separate platforms don’t interconnect, it still doesn’t work.”

So clearly there is lots of hype and some huge expectations, but to date just not quite enough standardization and interconnectivity to keep anyone but the venture capitalists thoroughly satisfied. Indeed, adequate funding to expand trading platforms may be one of the few things that is not currently a problem.

“There’s plenty of money chasing this market’s development,” says Elliott of Band-X.

It’s just far from certain which team of participants will take the checkered winner’s flag.


The logistics of measuring and re-routing minutes traffic appears to have been conquered by numerous firm-Band-X, Arbinet and among them. But different physical equipment is needed to measure and redirect bandwidth traffic, and the infrastructure for this is still largely being put together. Two companies standing in the middle of this effort are Washington, D.C.-based LighTrade Inc. ( and Chicago-based Universal Access (

By September, LighTrade expects to have its first two “neutral pooling points” located in Seattle and Washington, D.C., up and running, with plans to establish an additional 12 hubs in key cities across the United States by year-end. The equipment to be used is Lucent-made, and not surprisingly Lucent holds a position on its board.

LighTrade’s is clearly a “build it and they will come”-type strategy, pegged to hopes that the telecom industry will use its facilities to measure the quality and amount of data bits sent between any two points. If its hubs prove popular as a neutral platform, it could do wonders to help spur the standardization of trading.

“We are a single-focus, niche company,” says LighTrade chairman Ted Pierson, and “we hope to play an important role in establishing a centralized routing system. Some of the bigger telecom companies may talk about establishing ‘virtual pooling points’ instead of our physical ones, but I’m still not quite sure what that means or how that would work. Ours will be a necessary cog in this market’s development, and our fees will be acceptably low compared with what telecom companies pay today to get connected to each other.”

Universal Access, meanwhile, is concentrating on the provisioning of “last mile” connections for telecom companies that want to hook up to long-haul circuits, and is apparently doing so in a far more efficient manner that local Bell service providers. The company recently announced a strategic alliance with broker Amerex Bandwidth, a subsidiary of energy and electricity broker Amerex. To help it close complicated deals, Amerex Bandwidth bundles together the long-haul telecom-hub-to-telecom-hub passage with Universal Access’s cost of getting the signal carried onward from the hubs to its ultimate source and destination. Amerex Bandwidth can then quote an all-in price and an estimated delivery time for the connection far more easily than some of its competitors. “The deals are already flowing quite nicely,” says Mike Moore, Amerex Bandwidth’s president, who also serves as a LighTrade board member.

It’s not so important how you get the information from here to there, but simply whether you can get it there at all and at what cost and with what quality.”
-Michael Schwerin, venture capitalist

Yet another company, San Francisco-based, is teaming up with Telcordia to spearhead another point-to-point trading platform. “An open-access-type environment will come with time,” says RateXchange vice president Nick Cioll, “but we’re not waiting for it.”

Even Williams, despite its somewhat skeptical attitude toward the “trading” of bandwidth outside historical norms, has joined forces with Houston-based energy marketer Dynegy Inc. and broker platform eSpeed. Their combined intent is to build an Internet-based trading platform that will include telecommunications products.

But in some ways Lightrade’s and RateXchange’s paired-city models fly in the face of past telecom norms. Traditionally, voice connections have seldom followed a single dedicated route from one city to another. Instead, least-cost routing (LCR) often sees a phone call go from New York to Tampa to Dallas to Los Angeles, or perhaps through Minneapolis and Denver to Los Angeles. There is little degradation of quality based on the path it follows. In addition, the market, to date, seems to be bifurcated between obvious major routes that everyone already owns, and odd routes such as San Francisco to Brazil that few have. It’s not an easy market for a RateXchange to close many deals on, and some observers speculate that RateXchange is lagging relative to Band-X’s more varied trading platform and Amerex’s recent “local-loop included” brokering.

“A paired-city approach is not that important when you can go any which way around the world,” explains Michael Schwerin, a venture capitalist affiliated with Band-X financial-backer Madison Dearborn. “In a network environment, you have a vast interconnected cloud of network capacity. It’s not so important how you get the information from here to there, but simply whether you can get it there at all and at what cost and with what quality.”

In Schwerin’s mind, “Enron recognizes that in the early stages of a market’s development, it’s a huge advantage to have physical facilities, and others seem to be copying them. But the interesting thing about all of this is that Enron probably does not expect to make its money on commoditized paired-city trading, but instead on downstream services from these pooling points. The margin isn’t in paired-city trading and probably never will be.”

While Schwerin thinks that establishing 20 or 40 neutral pooling points will help in the development of a global network, he believes “the real money will be in the 15,000 downstream pops-all the Sun servers that Enron will be putting in all over the place-to service corporate clients that want video on demand, streaming data, added capacity during peak periods, and high bandwidth in the last mile.” Enron is developing a long-haul network to attract those clients initially, but ultimately, Schwerin sees Enron using other people’s long-haul capacity and getting rid of their own. “Just like in oil and gas-where they ultimately sold Enron Oil & Gas-they don’t need to produce this stuff, but they will make money rebundling it.”


voice on thexchange | data on thexchange | mobile on thexchange | Services | Intelligence | About | News | Investors | Contact Us
Disclaimer | Compatibility | Terms of Use | Home

© Copyright 2000-2007 Arbinet-thexchange, Inc. All Rights Reserved.