To all - WSJ article about JDSU (sorry if already posted).
January 20, 2000
A Fiber-Optics Powerhouse
Is Poised to Speed March
By SCOTT THURM
Staff Reporter of THE WALL STREET JOURNAL
Fiber optics is leaving Moore's Law in the dust.
Decades ago, Intel Corp. co-founder Gordon Moore correctly predicted that
the number of transistors that could be squeezed onto a piece of silicon would
double every 18 months. Generations of faster and cheaper computer chips
have driven technological progress ever since.
Now, the amount of information that can be transmitted over a strand of glass
is doubling every nine to 12 months. Advances in fiber-optics stand to replace
advances in chip speed as the new pace-setter in computing. And if there is a
budding Intel of the fiber-optics business, it may be JDS Uniphase Corp. of
San Jose, Calif.
JDS Uniphase is among a group of companies that make powerful and precise
lasers, amplifiers and filters that underlie the rapidly accelerating advances in
fiber-optic technology. The pace of their innovation suggests an era of
ever-cheaper pipelines for data, video programming and almost any other
digital-information product.
"They and companies like them are really what's enabling this tremendous
increase in bandwidth," says Scott Grout, chief executive of Chorum
Technologies Inc., a start-up maker of fiber-optic gear and a former researcher
at Lucent Technologies Inc. Just as microprocessors have dominated the last
two decades, Mr. Grout says, "I believe the next two decades are going to be
about connectivity and bandwidth."
Greater bandwidth allows telecom companies to keep up with the explosive
growth of the Internet. And lightning-fast connections on fiber-optic networks
are reshaping computing by letting computer users store information in remote,
central servers and tap that information over the Web from relatively simple
devices such as cell phones and personal organizers.
Investors have certainly seen the light in recent months. Start-up makers of
fiber-optic equipment with meager sales are going public, or are being gobbled
up for billions before completing a single product.
Shares of JDS Uniphase have risen more than
tenfold in the past year, giving the company a
value of about $60 billion even before its planned
$16.5 billion acquisition of fiber-optic component
maker E-Tek Dynamics Inc., also of San Jose.
That deal was announced on Monday.
The story of JDS Uniphase's rise stands out among the tales of other, much
younger, fiber-optics high-fliers. JDS Fitel Inc. and Uniphase Corp., which
merged last summer, each had a tortured 20-year history before striking gold.
They were led by two immigrant chief executives, who are now co-managing
the merged company from 3,000 miles apart.
JDS Fitel was formed in 1981 by four engineers from what is now Nortel
Networks Corp., the Canadian telephone-equipment giant. In 1990, the
co-founders sold half the company to Japan's Furukawa Electric Co. for
access to Furukawa's fiber-optic technology and sales force. JDS Fitel, based
in Nepean, Ontario, specialized in building so-called passive components,
which manipulate beams of light generated by lasers.
Fiber optics definitely wasn't a hot industry in the early 1990s and that was fine
with JDS Fitel. The company deliberately kept a low profile. Its executives
were even elated one year when JDS Fitel was omitted from lists of Canada's
biggest technology companies. Chief executive Jozef Straus, now president of
the merged company, is a Czech emigre and Orthodox Jew well known around
Ottawa for his ever-present black beret.
Uniphase, in San Jose, struggled making lasers for chip-making machines and
barcode scanners until Australian-born Kevin Kalkhoven arrived in 1991. Mr.
Kalkhoven, a scuba diver and jet pilot, says bankers had to "twist arms" to find
investors for Uniphase's 1993 initial public offering, which raised all of $11
million.
Fiber optics, in the early 1990s, didn't seem promising. The technology was
plagued by an expensive barrier to widespread use: Signals weakened as they
traveled over many miles of glass and had to be boosted at regular intervals.
But in 1994, Mr. Kalkhoven found what he thought was the solution and bet
his company on it. He was reading a magazine article how about MCI
Communications Corp. had used a new type of optical amplifier to send a laser
beam from Chicago to Sacramento, Calif., without using expensive electrical
equipment to regenerate the signal. With no experience in telecommunications,
Mr. Kalkhoven decided to gamble the company by acquiring and developing the
technology.
"If it hadn't worked," Mr. Kalkhoven says, "we would have been broke."
That bet paid off, partly because companies such as JDS Fitel and Uniphase, as
well as traditional powers such as Nortel and Lucent, began a fantastic game of
technological one-upmanship.
Fiber-optic networks work by turning the 1s and 0s of computers into pulses
of light. In the mid-1990s, engineers were racing to speed up those pulses.
Other engineers devised new technology to split a beam of light into multiple
"colors," each of which can carry a separate steam of information. As both the
number of pulses and the number of colors grew, the capacity of the network
grew enormously.
Ciena Corp. introduced the first commercial system to break light into sixteen
colors in 1996. Each color could carry 2.5 million bits of information. Soon
fiber-optics companies were splitting light into even more data-carrying colors.
Today, suppliers make equipment to break light into 80 colors, each carrying
10 million bits of information, a 20-fold increase in data-carrying capacity in
just four years.
The fiber-optic equivalent of Moore's Law calls for innovation to continue at
that white-hot pace as new colors are added. Lucent recently said it had
developed experimental equipment to break light into 1,022 colors.
In 1998, four years after Mr. Kalkhoven took his big bet, the Internet was
creating huge demand for fiber-optics capacity and he was ready to market his
technology more aggressively. First he needed the components of fiber-optics
networks that he lacked and he found common ground with Mr. Straus.
Their two companies' strengths complemented each other perfectly: One made
lasers that create the beam, the other built the devices that moved the beam
around. Mr. Kalkhoven and Mr. Straus took off on a three-day hike in the
Canadian Rockies in the summer of 1998 to see if they could get along. "We
drank a lot of Scotch," Mr. Kalkhoven recalls.
Since completing their merger last summer, Mr. Straus has remained based in
Canada and Mr. Kalkhoven in San Jose. But the combined JDS Uniphase has
been running hard. Acquisitions have been a big part of its strategy: It has
completed or announced seven, including the E-Tek deal. The resulting
company now has the broadest array of building blocks for fiber-optic
networks, and such a stranglehold over some parts of the market that the
E-Tek acquisition likely will face tough antitrust scrutiny.
JDS Uniphase also has eroded the traditional tight vertical integration in the
industry, where companies such as Nortel and Lucent made their own parts
for the fiber-optic gear they sell to telecom companies.
Still, there are hurdles. In particular, JDS Uniphase will have to get a lot better
at manufacturing. Building optical components is still heavily labor-intensive,
with workers typically feeding the glass fibers by hand into tiny housings.
Filters, meanwhile, are made by applying coatings, individually, to a piece of
glass. Each coating can deflect a specific wavelength of light, in much the
same way that sunglasses block ultraviolet rays. By applying dozens of
coatings, a single beam of light can be broken into dozens of "colors."
But the requirements are so precise that defect rates are high: Only about one
of every 20 filters is commercially usable, says Hugh Martin, chief executive of
Optical Networks Inc., a start-up maker of fiber-optic switching gear.
And Mr. Kalkhoven is uneasy about comparisons to Intel, whose engineers
have put Moore's Law into action, by squeezing transistors closer and closer
together on a piece of silicon. "We've mastered the science," he says. But "Intel
became great because they learned how to manufacture the microprocessor in
large volumes at an economical price."
Copyright ¸ 2000 Dow Jones & Company, Inc. All Rights Reserved. |
