Some new names here, but I don't want the ones that have to break new ground, except for those who hold the promise of mass production of things made by hand now. So I have been reading a little about InP, and bought a few shares of one of the InP players;companies like GLW and JDSU are just fine with me. If these little guys have a breakthrough the big vacuum cleaners can get them.
The Optical Train Rolls On
By Tiernan Ray
November 24, 2000
Related Stories
In three months, the domain of sophisticated computer-networking equipment built of glass fiber and tiny mirrors has gone from a fund manager's sure thing to a disaster. One day in August, Nortel Networks (NT) is telling the market it will invest several billion dollars in new optical-manufacturing capacity to meet unbridled demand; two months later, it's coming up with the-dog-ate-my-homework reasons why it can't sell enough fiber optics to meet expectations. The wheels come off at Lucent Technologies (LU) for a second and third time and WorldCom (WCOM) stuns us by saying it can't make money running phone lines.
It all confirms for many that pushing an optical-communications switch — the bread-and-butter at Sycamore Networks (SCMR) and Corvis (CORV) — just has to be "tech's last great bubble," as James Surowiecki casually dubbed the sector in a recent issue of The New Yorker. There's little at present to support the argument that fiber optics is now a "value play." Year-to-date, the fall of Sycamore (44%), Corvis (67%) and others has scared all but the most rabid investors for the foreseeable future.
I think the sell-off is all very shortsighted, though. I don't much care that fund managers who were lapping up optical networking in August won't go near it these days. They'll be back, mainly because the history of fiber optics is a grand one, and it continues to shape the very destiny of public-communications networks. However, the only way to understand the investment premise is to understand what's actually happening to those networks. So what is this optics stuff?
I've spent the past few weeks making the optical-networking "circuit," and seen much. The circuit has gotten bigger because, far from being a local mirage rising from the sands of Silicon Valley, optical-networking technology is spreading fast from one lab bench to another around the world. From the exclusive confines of San Jose, Calif., and the suburbs of Boston, the innovations have spread to small towns like Newark, N.J., and Dublin, Calif.; to university centers such as Research Triangle Park in North Carolina; to Monmouth County in New Jersey and around Princeton University, and down to Glen Burnie and Columbia in Maryland. Optical has spread the wealth to unlikely tech corners, such as Melbourne, Fla., where Nanovation Technologies, a maker of cutting-edge optical circuits, is headquartered; and centers of optical excellence in Scotland, Belgium, Quebec, Ottawa, Sweden and many other points.
We're entering what many are calling the third generation, or 3G, of fiber optics, a development that may actually offer new revenue opportunities to the telco's. First was the primordial dawn. In the 1970s, researchers like Jim Hsieh made it practical to transmit light in glass fibers over long distances. Paul Lazay developed the very first fibers for undersea transmission. (Today, Hsieh's got a new firm, Nova Crystals, in San Jose, that sells cutting-edge laser chips; and Lazay runs IronBridge Networks, a privately held firm developing Internet routers that transmit data at trillions of bits per second.)
Seeing the Light
Here are some companies to watch as optical networking begins to spread across the Internet.
Company Location Business
Nanovation Technologies Melbourne, Fla. optical-switching components
Nova Crystals San Jose, Calif. VCSEL lasers for telecom equipment
Corvis Rockville, Md. optical amplifiers, switches
Sycamore Networks Chelmsford, Mass. DWDM switching and transmission
ONI Systems San Jose, Calif. metropolitan optical equipment
IronBridge Networks Lexington, Mass. terrabit-speed routers
Iris Labs Plano, Texas optical-enabling technologies
Metera Networks Richardson, Texas optical-networking equipment
Latus Lightworks Richardson, Texas long-haul optical-transport gear
Iolon San Jose, Calif. tunable-wavelength lasers
Then Ciena burst on the scene in 1997 with the first so-called dense wavelength division multiplexing, or DWDM, equipment. DWDM multiplies bandwidth by sending data simultaneously over many colors of light, using multiple lasers and combining the beams with a prism. Sycamore and ONI Systems (ONIS) have followed with second-generation products that cram lots of phone lines and data traffic onto each wavelength. Both first and second generation help to cut costs: DWDM saves on the amount of new fiber being built, and boxes like Sycamore's allow carriers to save equipment costs to transmit data over those DWDM channels. As one analyst remarked to me not so long ago, technologies like DWDM create essentially "free bandwidth."
With 3G optical networking, the emphasis shifts from cost cutting to creating new revenue. It helps to think about how phone companies struggle to sell bandwidth in order to understand what's going on. In the late 1960s, phone companies invented what's called the "T" system, the main representative of which is the well-known "T-1 line." If you're lucky enough to have one of these, you've got a connection to the Internet of 1.5 million bits per second. When fiber was first widely deployed, in the mid-1980s, carriers used T-1 as the main product for large businesses that wanted high-bandwidth networking. T-1, in other words, became a unit of commerce.
The phone companies now buying all the fiber-optic gear need a new unit of commerce they can sell for a new age, if they are to monetize all that DWDM and switching equipment. That new unit of commerce will likely be the individual wavelengths of light created inside DWDM systems.
I recently made a pilgrimage down to Dallas, to a research conference called Opticomm — in the suburb of Plano, not far from where the Maharishi would break ground — with the intent of understanding this third generation of optics. Plano, famous mainly for J.C. Penney (JCP) and Frito Lay, is also being transformed by fiber optics: The population has nearly doubled in the past decade as telecom firms such as Lucent, Nortel and divisions of Fujitsu and Alcatel (ALA) have set up operations. Now optical firms are flocking. H. Michael Zadikian, the entrepreneur who sold Texas firm Monterey Networks to Cisco Systems (CSCO) in 1999 for a cool $7.4 billion in stock, has launched not one but three new optical firms in the area, Iris Labs, Metera Networks and Latus Lightworks.
There was plenty of talk at Opticomm about the continuing efforts to create more bandwidth. One development is the tunable laser, which lets a single optical transceiver perform at any wavelength that's needed in a DWDM system. That makes it cheaper for system vendors like Sycamore to build DWDM systems. It's an odd state of affairs, almost as if the microprocessor were just being invented even as the first personal computers were rolling off the lines. One company to watch is privately backed Iolon, a tunable-laser firm.
Back to the Drawing Board
Conrad Leifur, a Wall Street analyst who covers JDS Uniphase (JDSU) and Avanex (AVNX) for Piper Jaffray, predicted that next year everyone will be talking about networking equipment that can transmit data at 40 billion bits per second, instead of 10 billion bits per second, which is the high end for most networks today. However, "40 gigabits has a marketing angle that's bigger than its bandwidth importance," said Leifur. That's because what's happening in 3G to change the Internet is far more important than the speed offered by fast, tunable lasers. It's no exaggeration to say that the shift to 3G may well change the entire way the Internet works. The way data move around the Internet today is rather like the "traveling salesman" problem from high school or college math courses: The weary peddler must pass through each town on his route only once in order to complete the sales trip most efficiently. Similarly, data on the Internet wends its way to your desktop through a statistical process that seeks to optimize the number of stops. That statistical process is computed second-by-second on the Internet inside large "routers" sold by Juniper Networks (JNPR) and Cisco, among others.
Retreat of the Light Brigade
3G is all about a new kind of software that doesn't merely fling data across the Web, but rather pays attention to the conditions of the roads at each point. Called MPLS, the software wouldn't only compute the statistics of each stop (the "salesman" approach), but would also have information about each and every wavelength of light in a WDM system. MPLS can route each color belonging to a customer to its destination based on which fiber cables are most full at the moment, or based on which cities are reporting broken fiber conduits thanks to backhoes. By allowing phone companies to manage each color of light, MPLS transforms WDM from a speed technology into a system of commerce. Individual colors of light could be priced and sold to each customer, creating a new revenue engine to replace the aging T-1 while giving the Internet a whole new structure.
There were a lot of hoary academic presentations at Opticomm on how Internet routing will change under MPLS, but the technology is far from tweedy: A week later I attended the Next Generation Networks conference in Washington, D.C. Less a research forum than a trade conference, most NGNers wear golf shirts with company logos and laugh too loud at industry jokes about fiber optics. I put the question of MPLS to Dr. John McQuillan, the sage-like impresario whose mild, contemplative manner and brushy beard lend him the air of a country doctor. "MPLS is very real and I'm astounded by the progress it's making," says McQuillan, noting that over 200 Internet switches were running the MPLS software live on at least 20 networks around the world.
Sounds Like Redmond
With a kind of groundswell for MPLS, the question now may be who will ultimately control the stuff. While it's tempting to think optical upstarts like Corvis will be running the pipes, the Net's existing stalwarts are not giving up. Scott Kriens, chairman, president and CEO of Juniper, gave a talk at NGN alluding to "natural boundaries." It was a lively refresh of the lecture Microsoft (MSFT) has given the software industry for years, the message being, "stay out of our way, or else."
However, if the MPLS saga plays out, Juniper's routers will have competition from a host of other, fascinating devices. Researchers are figuring out how to build the optical equivalent of Juniper's routers, something they call "photonic packet routers."
A recent Boston conference, Photonics East, hosted a whole day of scholarly presentations on building these devices. IP networks, like the Internet, send data in small chunks, called packets. Routers, like the kind made by Juniper, know all about packets, which helps them decipher MPLS. But routers are based on electronics, not photonics: they must first convert the wavelengths inside DWDM into electrons before they can switch the channels of communication. That slows down the transmission.
Today's optical switches can't read packets of data. I've mentioned recently (see "The Amazing Technicolor Light Show") that Nortel and others are building switches based on tiny mirrors that rotate to deliver the light. Tiny mirrors may become as emblematic of the Naughts as disco balls a generation ago. For they allow DWDM gear to direct light irrespective of its speed, which is very helpful for making telco equipment simpler to use.
Mirror, Mirror...
But mirrors don't know anything about the data inside the stream of light. And even if they did, they move too slowly to switch those packets. Mirrors are built using a mechanical technology known as MEMS, which can only rotate the beam of light once every thousandth of a second or so. At speeds of 10 billion bits per second and higher, which is now the standard for big telecom networks, packets of data arrive at a router every 40 billionths of a second — far too fast for MEMS, in other words.
One of the most intriguing academic presentations at Opticomm was given by Dr. Ben Yoo, from the University of California at Davis, who is working on photonic packet routers using what are called "wavelength converters." These devices can switch each packet of data by changing its color to a new wavelength every billionth of a second, or nanosecond. "Basically, we're building an all-optical router," he remarked in his talk. Yoo says the resulting optical router can scale to about 1,000 times the largest electronic router and the speed could be 1,000 times faster. That would result in 1,000 trillion bits, or one "petabit" per second switching, something unthinkable for electronic routers relying on repeated signal conversion between optics and electronics.
The secret is Yoo's design for wavelength converters fabricated in a standard semiconductor material that has no moving parts like MEMS. Firms such as SmartMoney pick Vitesse Semiconductor (VTSS) already know how to fashion such high-speed semiconductor materials in high volume at commercial prices. Yoo's all-optical router research is in two phases, the first utilizing off-the-shelf commercial components, and the second utilizing advanced wavelength converters.
Yoo's research may be a few years from commercial viability. And while he receives calls daily about his work, he has no plans as of yet to found a start-up. Skeptics will argue that the present sorry state of the telecom market doesn't bode well for mirrors or any other expensive 3G technologies. Participants at Opticomm made a valiant pitch for the sustainability of the market, however. Rob Coneybeer, a venture capitalist with New Enterprise Associates, a West Coast firm, poo-pooed lingering concerns about young phone companies suddenly cash poor. "CLECs [competitive local exchange carriers] represent less than 15% of annual U.S. capital expenditure," he stated. "The available capital expenditure at large, young phone companies is tremendous given the cash on their balance sheets."
Yes, the Qwests (Q) of the world may indeed rush to the rescue. But I think the only way to understand this market for the long haul is to appreciate that technologies such as DWDM and MPLS, far from being an extravagant expense, may be a way out of the darkness for phone companies craving new revenue. For optical investors, now is a time to just believe. Take my word for it.
Tech Market Archive |
