Tellium Triumphs

It looks as if optical switch vendor Tellium Inc. is finally getting some respect, courtesy of two first-class contracts and Friday's IPO filing (see Tellium Bids for $250 Million IPO).

Where does that leave the competition? In a position to celebrate, not mope, I would submit. Tellium's step up in stature puts a major stamp of approval on this entire market space.

Tellium hasn't had an easy time of it. Early on, it was switching lambdas as part of MONET (Multiwavelength Optical Networking Technology), the government-sponsored testbed. Other participants included AT&T Corp. (NYSE: T), Bell Atlantic (now Verizon Communications [NYSE: VZ]), BellSouth Corp. (NYSE: BLS), Lucent Technologies Inc. (NYSE: LU), Pacific Telesis, and SBC/TRI (NYSE: SBC).

Actually, Tellium, per se, didn't exist during this project, since it was part of Telcordia Technologies Inc. (Bellcore, at the time). Still, it was able to develop the technologies for line-rate optical circuit switching and showcase the first optical switch running in a live network. It followed with a small contract with the Department of Defense, the industry's first optical switch contract.

This should have placed Tellium well ahead of their emerging competition, but startups tell better stories, and Tellium often betrayed its roots in the decidedly undynamic Bellcore. In 1998 Lightera caught everyone's attention, particularly that of Ciena Corp. (Nasdaq: CIEN), who bought it in early 1999 for $463 million. Monterey followed, and was quickly brought into the Cisco fold for $500 million in stock. Sycamore Networks Inc. (Nasdaq: SCMR) started its own optical switch skunk works in 1999, and I began wondering what would become of Tellium. Speculation most often centered on a Lucent buyout, but an OEM agreement was announced, nothing more.

That helps explain why Tellium's contact with Extant (now part of Dynegy Inc. [NYSE: DYN]) wasn't considered big time, though $250 million is an order of magnitude larger than most startup contracts these days. Everyone was waiting to see what Lightera and Sycamore had to offer.

When Tellium recently inked a contract with Cable and Wireless (NYSE: CWP), though, it was big news indeed. Working with a top-tier ISP gives vendors substantial (instant) credibility. And Tellium's deliverables should definitely boost its reputation: An optical system scalable to 512 ports that synchronously switches OC48 and OC192 circuits in a mesh architecture has been trialed and approved. Sometime in early 2001 it will be deployed on one of the biggest IP networks in the world.

A week later Tellium signed with Qwest Communications International Corp. (NYSE:Q): another deal worth hundreds of millions with another first-class customer. And another reputation boost: Optical networking vendors covet Qwest contracts because the company is seen as both an emerging carrier and a tested stalwart with an outstanding engineering staff.

To wrap up the hat trick, Tellium filed for an IPO on Friday, September 22, proving that it's as ready to take its story public as any other optical networking vendor in the past year. (Note that like all the other optical startups that have already had IPOs, Tellium has issued its service provider customers with shares).

As I said, Tellium's triumph helps validate the entire optical switching industry. Big contracts with big ISPs and carriers will be immensely important this year, setting the stage for significant growth in 2001.

How much growth? My first forecasts for Pioneer Consulting, published in February 2000, were very high. I predicted the combined market opportunity for North America and Europe was worth $31 billion over a five-year period, $15 billion in 2004 alone.

At the time, it was important to demonstrate the impact optical switches could have on core optical networks, effectively displacing well-established markets for Sonet OC192 ADMs, large broadband digital crossconnects, and even terabit routers. The market opportunity was thus much greater than the expected uptake of optical switches in the 64x64-port to 256x256-port range alone.

In June I revised my optical switch forecasts sharply downward. I didn't feel any differently about the viability of the market, but six months of listening to carriers convinced me reliability would be an overwhelming factor in purchase decisions, even though they remained overwhelmingly favorable toward optical switches. Combine that with the fact that vendors are taking a bit more time than anticipated to bring products to market, and forecasts shift noticeably to the right.

My new forecasts, shown in the figure below, also accounted for carrier purchases of large matrix switches with a smaller number of port cards. Since much of the cost of these switches (typically $25,000 per bidirectional OC48 port) is in the cards, this led to a much lower cost per switch. A box with 64 OC48 ports would run roughly $1.9 million (64 times $25,000 plus $300,000 for the switching system).

The outcome is a lowered near-term forecast but a more dramatic curve upward after 2001. By 2004, the market is forecast to reach a robust $7.5 billion, creating ample room for numerous successful vendors. This curve is in line with many emerging optical networking markets, which tend to be smaller in the near term than many hoped but larger over the long term, as the inexorable migration continues.

In the end, the optical switch market remains very strong, with extremely aggressive growth predicted for 2002 and 2003, when shipments will reach into the billions. Right now, the OEO (optical-electrical-optical) switch vendors are doing quite well and should continue to expand their market presence. Some will offer STS1-level grooming (Brightlink Networks Inc., Ciena, Sycamore); others will stick to the high road, switching only OC48s and above (Cisco, Tellium).

What remains to be seen is how well the all-optical switch vendors fare in the coming five years. Alcatel SA (NYSE: ALA), Calient Networks, Cinta, Lucent, Nortel Networks Corp.’s Xros, Siemens AG (Frankfurt: SIE), and others yet to come all face a challenge from the now-established OEO players. They're going to have to prove both the economics and the reliability of their photonic systems to the same small customer group. As carriers like Qwest, C&W, Dynegy, Williams Communications Group (NYSE: WCG), and Enron Corp. (NYSE: ENR) deploy OEO over the coming three years, it's critical to watch how they scale their core nodes. Will they supplement their OEO switches with new cores or add an all-optical "superlayer" on top, relying on new signaling schemes like MPLambdaS to manage lightpaths with the same dexterity as virtual circuits?

In recent conversations, silicon vendors have assured me that electrical switching fabrics are far from topped out. That's convinced me that today's OEO switch is not an interim measure on the way to all-optical networking. It's a solution with more lasting power than Sonet ADM or even the telephone switch. Who knows? It may even outlast the terabit router.

Scott Clavenna is president of PointEast Research, and director of research at Light Reading http://www.lightreading.com