NT: JDSU's partner:3rd time lucky? The all photon network.
canadianbusiness.com
Once a leading maker of telephone switching systems, Nortel transformed itself into an
Internet titan. Now, boss John Roth must win the race to develop the all-photon network
By Ian Austen | Issue date: Feb. 19, 2001
For its "What Do You Want The Internet To Be?" television ad campaign, some agency
smoothie persuaded Nortel Networks Corp. that featuring a bowler-hatted twerp in a field
with a cow was a good idea. But the communications equipment giant at least had the good
sense to eliminate most of the loony lyrics from the 1969 Beatles song "Come Together,"
leaving its swoopy bass line to fill the soundtrack. After all, what would viewers be
expected to think of Nortel president and CEO John Roth after hearing gibberish like "He
bag production/He got walrus gumboot"? Not to mention shareholders in the company
(TSE: NT), who are still trying to make sense of the blue-chip's precipitous slide last
year.
A little more than three years ago, Roth—after a near-religious conversion that involved
buying a used glove box for one of his vintage British sports cars through the
Internet—ordered up a now famous right-angle turn at Nortel, shifting its focus away from
voice network equipment to the Net. The result was a doubling of sales, Nortel's
transformation into arguably the world's leading network equipment maker and, for Roth,
more CEO of the Year awards than his credenza can likely contain. Now, the easy stuff is
over.
Nortel, headquartered in Brampton, Ont., is the world's biggest producer of optical
networks, which are the guts of the Internet. It likes to boast that it helps the Net move at
the speed of light. But that's only true up to a point. Since Roth became CEO in late 1997,
the company—aided and abetted by component makers such as JDS Uniphase Corp.—has
continually boosted the speed and the amount of information that streaks through fibre
cables to the point where several digital copies of the entire Encyclopedia Britannica can
now zap almost instantly around the world. What fibre fans don't like to talk about,
however, is that those networks to a large extent have all the elegance and efficiency of a
diesel-powered typewriter. Boosting those pulses of light—and all but the crudest traffic
direction of the data carried on fibre—creates bottlenecks, as photons (the essence of
light) are changed into electrons, run through computers and then are transformed back into
photons to complete their journey. (It's a bit like making a transcontinental journey with an
airline that occasionally lands the plane and then forces you onto a bicycle before
allowing you back to your economy-class seat for another liftoff.) The solution, and
perhaps the key to Nortel's future, is taming the photon. And there's no time to waste.
Shortly before Roth took over, Nortel opened a massive expansion of its R&D centre on
Ottawa's western fringe. Down in the sunlit lobby, Nortel conspicuously highlighted what
was then its crowning achievement: a giant DMS-100 digital telephone switch. With its
brown-and-avocado equipment racks sitting behind a long row of picture windows, the
sprawling DMS switch looked like an army of overgrown appliances that had escaped
from a '70s kitchen. If nothing else, it was impressive for its sheer size.
It was also impressive for being the product that created the modern Nortel. Before DMS,
telephone companies switched calls in their central offices with clackety
electromechanical machines that required careful and nearly constant cleaning and
adjustment. (For further information, or to at least see one in action, visit your local video
store and rent Alfred Hitchcock's Dial M for Murder.) But if kept cleaned and oiled, these
machines provided highly reliable service.
When Northern Telecom Ltd. (as Nortel was then known) unveiled the DMS in 1976, it
took a huge gamble. While its former corporate sibling Bell Canada was a likely customer,
the idea of digital switching wasn't warmly received by the famously conservative phone
industry. Even worse, the doubters included the then all-powerful AT&T Corp., which
dominated local and long-distance phone service in the US and was going slowly with
digital switch development at its own equipment division (today's Lucent).
Like most successful gamblers, Nortel got lucky. In 1984, AT&T was busted up by an
antitrust action. Suddenly, the US market was full of new phone companies that, in many
cases, were eager to do business with anyone but AT&T.
Getting the DMS switches into central offices, however, was only the beginning. Once the
telcos were hooked, Nortel was kept busy creating and selling them high-margin software
upgrades. Shut out of the long-distance market by the antitrust ruling, US-based telcos
were able to grow revenue by selling services dependent on Nortel's software. Among
them: allowing customers to spy on incoming calls, interrupt callers to take other calls,
and have their phone answered by the phone company. Less than a year after moving up to
Nortel's top spot, Roth made a subtle but telling gesture. At the R&D centre, blinds came
down on the picture windows displaying the DMS test-bed installation. Narrowband
switching—Internet engineers' idea of a slur—had become the crazy old aunt in the attic no
one talks about anymore.
When I met Roth that spring at Nortel's lab, he shared his thoughts about the company's
right turn (and the difficulty of buying parts for old cars with notoriously bad electrical
systems). He also offered the usual bromides about wanting the Net to seep into every one
of Nortel's corporate pores. With the exception of the stubborn software engineers who are
still dreaming up new ways for the DMS switches to make telephoning even more
unpleasant, that's probably true today. But the thing Roth really did after that meeting was
place another huge bet.
This time it was an engineer's kind of a wager. On the assumption that the Web was about
to become something more than a way for middle-age men to find car parts, Roth pushed
an optical networking technology that goes by the less-than-catchy name OC-192. In its
first version, OC-192 was at least four times faster than anything underway at Lucent. Fast
enough that if everyone in the US decided to make a phone call at the same time, an
OC-192 system could handle the load on a single strand of fibre.
But the market yawned. One of the first sales, to Denver-based Qwest Communications
International Inc., then a small but growing long-distance carrier, closed only because
Nortel offered unusually easy terms. Even at that, recalls Qwest's president of worldwide
operations Afshin Mohebbi, "People were laughing at both of us. They thought these
systems were too big for [the market].''
Of course, the fact that Mohebbi offered that recollection through a Nortel Webcast
suggests that everything worked out all right in the end. But again, luck was with Nortel.
While Roth had clearly predicted that there would be a huge rise in demand for
communications, no one knew just how big it would be. Similarly, the long-running stock
market boom that made it possible for new network operators to quickly raise huge
amounts of capital to buy goodies from Nortel also defied any credible prediction.
Now, of course, the spending spree has come under scrutiny. British Telecommunications
PLC, a Nortel customer in the UK, is so cash-short that it has put most of its real estate, a
portfolio valued at about £2 billion, on the market. In late January, AT&T reported a
US$1.7-billion loss for its fourth quarter. WorldCom Inc., the debt-heavy empire cobbled
together by Bernie Ebbers is, as it's politely put, retooling. Ebbers' favorite-son status in
his hometown of Edmonton is becoming as shaky as Peter Pocklington's. For John Roth,
business has never been more stressful.
When Nortel made its much anticipated fourth-quarter announcement in January, one
comment by Clarence Chandran, chief operating officer (and a leading contender to
replace Roth), passed unnoticed in press reports the next morning. The portion of the Net
where Nortel remains particularly weak is the router market. Routers are those specialized
computers that read the labels on the packets that data are broken into as they travel the
Net, and make sure they're sent along a path that leads to their intended destination. One
objective of Roth's US$9.1-billion purchase of California's Bay Networks Inc. in 1998
was to become a router player. But today that sector is, if anything, even more dominated
by Cisco Systems Inc. So, not unreasonably, an analyst at the Q4 announcement asked
Chandran what had become of the router strategy. Chandran didn't put it in so many words.
But he more or less confirmed the widespread belief that Nortel had ceded the router
market to Cisco. He did make clear, however, that the future was one thing: the
all-photonic network.
Not that Nortel can deliver that today. But even if the current downturn becomes more
severe and sustained, there's little doubt that a network that carries and directs traffic
purely as light is about as sure a thing as Nortel could hope for. Even better, it could also
give the company the same dominant position in smaller networks—metro networks, as
they're known in the trade—that it now enjoys in the national and global network market.
JDS Uniphase pointed the way. Its workforce has grown rapidly over the past four
years, in large part because it found one all-optical solution to a big problem for networks.
Unless it's boosted, a photon signal fades away after about 80 kilometres. JDS, however,
developed fibre optic components, such as lasers, that can do the amplifying job without
forcing the speedy photons to first be converted into electrons.
Still, there's a hitch as those photons approach their general destination. All but simple
bulk switching demands the awkward photon-to-electron conversion. Not only does that
involve a lot of expensive equipment, it slows the network down. Indeed, in some
scenarios, the fibre is so good at delivering data that it overloads the switches. To avoid
losing data, switches generally store the overflow in a digital inbox until they can deal
with it. The delays, in human terms, aren't all that long. But they're one of the many things
that conspire to make all video Webcasts look like jerky Max Headroom reruns.
An all-optical switch is the answer. And it wouldn't just be for the likes of a long-distance
carrier such as Qwest. Metro network operators—such as competitive local phone
companies—are also interested, although for somewhat different reasons. The more
switching they can do with photons, the greater the amount of fibre cable they can lay in
place of copper wires. Copper is much more expensive than fibre. And in crowded urban
centres, finding underground space to accommodate much larger (relative to their
information-carrying ability) copper cables is becoming increasingly difficult.
Deregulation has meant that a range of competitors is now fighting for position beneath city
streets.
The fact that no one has pulled this off yet indicates that building the all-photon network
won't be easy. For big switching—separating the packets destined for Manitoba from the
ones en route to BC—Nortel and others currently use a trick to boost the capacity of fibre
and make things manageable. With so-called Dense Wave Division Multiplexing, every
strand of wire simultaneously carries several different wavelengths—effectively different
colors—of light. (This works because unlike radio waves, say, light waves ignore each
other.) So you assign the Manitoba packets a specific wavelength and, as they approach
Winnipeg, use some kind of optical filter to divert them on their way.
But that's a long way from the ultimate solution: a system that can read and divert packets
when they're photons rather than electrons. A system, in short, that could make Cisco's
electronic routers as obsolete as mechanical adding machines. With microphotonics, as
this dream is sometimes called, things get very murky. Much of the related lab research
now underway involves quantum mechanics, the branch of physics that deals with atomic
and subatomic particles and where nothing seems real. (A particle can be in two places at
the same time? Yeah, sure, tell us another one.)
One thing's for certain: making optical switching a reality is going to take lots of money.
Not only will developing an optical switch eat up internal R&D budgets, it most certainly
will involve further acquisitions. Fortunately Nortel still has the touch when it comes to
attracting money. Its late-January decision to raise $1 billion in debt was warmly greeted
by the stock market. Similarly, if Nortel goes ahead with long-promised plans to spin off
part of its optical components business (like JDS, it makes the widgets that are cobbled
together with fibre and Nortel software to make a network), expectations are that the
company could pull in between $2.5 billion and $4 billion.
Of course, having a big bankroll and a good track record are no guarantee of future
success. The same kind of praise that's been coming Nortel's way over the past year was
going to Lucent back when Roth started his right turn. Since then, Lucent's been mired in a
series of mishaps (see "The lessons of Lucent" ) and fallen far behind in the optical
market. That's not to suggest, however, that you can now count it out of the photon
switching race. If nothing else, Lucent has unmatched depth when it comes to science. And
as Nortel did after Paul Stern (the former CEO who has since been all but airbrushed from
the corporate photos) severely chopped R&D, Lucent may learn from its mistakes.
It's also unlikely to be the only challenger. While Nortel seems disproportionately big
within Canada, several of its international competitors are of comparable scale. On top of
that, there are newcomers such as Nokia Corp. and Cisco who might want to expand their
product lines. A merger between, say, France's Alcatel and Finland's Nokia could provide
a powerful competitor. Depressed stock prices make such a move even easier to digest.
Meanwhile, there's always the potential for Nortel to lose its focus. As head of the world's
No. 2 wireless systems supplier, Roth is, not surprisingly, pushing the concept of
high-speed wireless Web access, the "wings of light" (don't ask). Part of the attraction, of
course, is the US$16.5 billion spent in a recent US government auction for new wireless
radio spectrum. (A similar auction raised $1.48 billion in Canada.) But in Europe, where
cell phone use is rampant compared with North America and wireless networks are far
more advanced, only 0.5% of French subscribers sign up to receive the Web on their
phones. Although British and German spectrum auctions generated unexpectedly high
returns for those governments, the current auction in France has been mostly notable for
bidders withdrawing.
But assuming Nortel plays it smart, and is lucky for a third time, the potential for
all-photon network products is enormous. Among other things, cracking the photon
switching problem would be a major step toward developing a whole new branch of
computing based on light. The idea of a communications gear maker leading the way is not
without precedent. (It was Bell Laboratories, now part of Lucent, that invented the
transistor.) But even if Nortel doesn't actually reshape the face of computing, a successful
move into the all- photonic network should allow it, if nothing else, to buy some decent
new ads.
-------------------------------------------------------------------------------- |
