The Optical Arms Race: Who Will Build The New Internet? Date:   Sunday , October 01, 2000 So what do visionaries view as the real Internet and communications paradigm for the twenty-first century? They see a rush of photons lighting up fibers and whisking data quickly and efficiently over the globe. They see new generation switches routing huge amounts of information at the command of software that will know in a matter of milliseconds where bandwidth is needed and how to deploy it. They see a new breed of optical network that will replace almost every element of the telecom-era infrastructure that exists today — a network that will finally allow the Internet to shed its cocoon of outdated switches and wires and grow into the technological marvel that it has long promised to become. Who will build it? Today, three standout names come to mind when many people think of the networking industry — Cisco, Nortel and Lucent. That’s only fair, in a sense — the three of them currently have a combined market cap of nearly $850 billion. But the high tech industry moves fast, and apparently un-movable giants can fade away in periods of a few years. Facts are facts, and optical networking is clearly more than just a three-player industry. Today, many would pin the future of the networks on the new powerhouses like Juniper and Sycamore. Some might even bank on a consortium of relatively nascent pre and post-IPO players such as ONI, Amber, Tellium and Avici, among others — all of whom have eschewed acquisition by the big three in hopes of loftier rewards. However you view it, the fresh faces have been instrumental in bringing a new generation of technologies into play. Startups abound, the race is heating up and technologies are evolving at an alarming rate. There will be trillions of dollars awaiting those who succeed, and the big boys are going to have to be nimble on their feet — fighting, innovating, acquiring and scrambling — to preserve their prosperous business lives. The Shock of the New “Let’s step back and get some perspective here,” says Krishna Bala, CTO of Tellium. “We are moving to a point where the bandwidth is increasing by a factor of a hundred to a thousand. What that does, essentially, to today’s infrastructure, is that it makes it disappear. There will be a new network build-out — a paradigm shift … given the shift, there is room for a lot of companies to play in the space.” ONI Systems founder and CTO, Rohit Sharma, adds that “E-commerce transactions this year have already been more than half a trillion dollars. Three or four years from now it will be multiple trillion dollars. The infrastructure needed to support all of this is the one that’s going to get built.” Says Sam Mathan, CEO of Amber Networks, a company that builds network edge aggregation products, “We are not just building a faster, cheaper box, we are addressing a new kind of product — with new technologies that are insurgent in nature — in that they are technologies that will change the way networks are being built.” Nearly every company, new and old, has a statement like this one to advertise the revolutionary nature of its product. Some of this is marketing buzz, but there is certainly astounding innovation taking place. This said, in the complex world of optical networking, how do innovation and business mix? Best of Breed Vs. Big Name What the emerging breed of niche companies — ONI, Amber, Tellium, Avici and others — are banking on is that the carriers will get into the habit of buying “best of breed” networks comprised of products from many different companies — thus permitting a company that builds one best-in-class network element to become a sustainable business rather than just fodder for acquisition machines like mighty Cisco. “Because IP is an open protocol and stack,” explains Sam Mathan, “you can now build the optical Internet out of best-of-class elements.” So far, this best-in-class model seems to be materializing on many fronts. Wall Street has rewarded companies like Corvis, which specializes in long haul solutions (the ability to send data long distances without electronic regeneration), with healthy market caps (Corvis’ cap is currently an enviable 30 billion). But in case the carriers don’t trust Wall Street, the niche players have come together to collaborate on a number of demonstrations. One of these demonstrations, “Operation: Interoptical,” at Supercomm 2000, combined elements from companies like ONI, Tellium, Juniper, Brocade, Corvis and Amber. The demonstration, according to Rohit Sharma, “mimics what [the niche players] go out and do for the carriers today.” According to Surya Panditi, CEO of Avici Systems, “The advances in technology are so rapid that the customers understand they can’t go to one company and expect to get all of the pieces.” The sky seems to be the limit for these hot new players. But John Adler, director of marketing for Cisco’s wavelength router business unit, disagrees. “Single point solutions are going to struggle,” he explains, “Carriers don’t want to deal with nine suppliers in their optical core, they want to deal with very few, and they want to deal with companies they know will be there for the long run.” Adler comes to Cisco from the Monterey Networks acquisition. He explains, “When we were acquired by Cisco, a senior executive at one of our customers said, “I’m really glad you did this, because we love your product — but there’s no way we could bet the core of our network on a startup.’” Says Adler, “In the core optical network, these products are pacemakers. If your doctor says you need a pacemaker, are you going to buy it from a startup that’s innovative, or are you going to buy it from Johnson and Johnson?” He adds, “Our equipment can manage in the neighborhood of 9 or 10 million voice circuits in its first release. If for some reason we had a problem with our technology or we went out of business, you’d have a real problem on your hands, you’d have to cut a system out of the network that’s running ten million phone calls.” Serious repercussions, indeed, but Adler jokes, “If you mess one of these things up, you get a free trip to Washington, all expenses paid by the FCC.” Surya Panditi counter-argues, “The risk is really in not making the best tech decision.” Amar Gupta, founder and CTO of Amber Networks, insists that gone are the days when “you had to be one of the big three to survive....best of breed companies can very easily offer end-to-end solutions that the big three guys had been known to provide previously.” Cisco certainly has a point about the risk involved in choosing point solutions, but they are also aware that the “niche” players are becoming important competition. What is clear is that Cisco can’t simply shrug off these competitors and build their business on the reliability and security of a good name. As the Internet and bandwidth take off, carriers will be looking for more and more advanced solutions. Kathy Szelag, a VP in Lucent’s Optical Networking Group, argues, “There are places where carriers want these brand new technologies, but do carriers want these brand new technologies throughout their network? No — it’s too high risk. So the truth is somewhere in between the two points of view.” Will the Giants Survive? The big three are clearly aware of the thirst for innovation. Eager not to be left behind by startups, Cisco, Lucent and Nortel have aggressively pursued acquisitions — spending something in the range of $50 billion in the last 18 months. “Startups are the R&D arms of these companies,” says Rohit Sharma, though Kathy Szelag is quick to point out that Lucent does a lot of engineering in house. But are acquisitions really enough to keep a company competitive in this space? Krishna Bala has reservations. Says Bala, “If I’m a Corvis, and I’m sitting with a 20 to 25 billion dollar market cap and growing. I’m going to ask myself ‘why would I even bother accepting any offer from Cisco?’ Why? There’s no risk. For example, at Tellium we make the highest capacity switches in the industry. Our engineers are top class. We have a third generation of switch coming out that’s going to be all-optical — it’s pure photonic switching. We know how to do that and Cisco’s not going to be able to build that with any internal group.” But Bala’s theory rests, in part, on Wall Street’s multi billion dollar market caps. Some may question the long-term potential of those numbers. Does Wall Street understand optics any better than it did e-commerce? The world is certainly moving towards optical networks, the market will be enormous, and the need for innovation in products seems established. But it’s only over the course of time that companies can really prove their ability to live up to the multiples. Rohit Sharma insists, “When people look at companies and say ‘so and so’s market cap is so high and it’s not supported by where their business is,’ they need to took at it from the other side and say ‘where does this market need to be five years from now?’ Then you’ll probably see that many of these valuations are on the lower side.” In Sharma’s mind, the growth of the optics industry is “not supported by historical data because what we are about to do has never been done. The Internet as it will exist in the next ten years has never existed before, so why should the old conventional rules of measuring these companies apply?” But some have even predicted a “bandwidth glut” — where the number of networking players makes for excess bandwidth. Juniper Networks founder Pradeep Sindhu explains, “Many people make the mistake of saying that the IP bandwidth will be proportional to the population of the world. I think that it will not be proportional to the population of the world because five years from now the killer app will be computer-to-computer interactions on the network — unmediated by human hands.” In this model, there seems to be no limit to bandwidth demand. Sindhu adds, “When Moore’s Law is essentially providing a doubling every eighteen months, and bandwidth demand and supply is galloping along at 2x to 4x every year you essentially have a scalability problem where you have to reinvent architectures every two to three years just to keep up with demand.” But the crux of the matter still lies with what model the carriers will go with. If Adler and Szelag are right and the carriers choose to go with only those vendors offering more extensive solutions, many of these publicly traded niche companies may have a hard time sustaining their current business model, no matter how large the market. But the reverse is equally possible, and far more shocking. If the carriers, pressured to perform by exploding bandwidth demand, go with the dynamic innovation of niche players, then the big three could see their own businesses profoundly affected. Will we soon be experiencing the “post Cisco era?” Probably not any time too soon, but the thought is already lodged inside the competitive minds of many entrepreneurial engineers in the networking space. What is clear is that some emerging players are fast on their way to becoming powerhouses. It’s hard to suggest that a company like Sycamore or Juniper is a “point solution” — and ONI, Amber and others are probably fast on their way to becoming household names. Many of the pre and post-IPO companies competing with the big three have no intention of remaining “niche” plays. Sycamore, for example, has every intention of growing into a company with the size and scope of Cisco (see interview with Desh Deshpande, page 36). Cisco must therefore also consider a future where there is not just a “big three,” but also a “big five,” or a “big eight.” In that scenario, Cisco may no longer boast a market cap of half a trillion dollars. Looking Ahead For us, the eventual users of the network, the bloodthirsty competition within this virtual arms race can only be a source of excitement, as we anticipate the ways in which the optics of the future can potentially change our lives. But there is a long way to go. When asked how far the development of the optical network has progressed on a scale of 0 to 10, most industry players rated that we were still between zero and one. Kathy Szelag paints an interesting picture of the future according to Lucent. “At some point in the future, you will do everything you do with electrons today, but you’ll do it with photons or soletons. We have a vision that in your home there won’t be any copper at all. In fact, there may be fiber — but the fiber may end in the wall and not even come into your television. Think of your home as having optical sensors built into the walls all over the place and no wiring. So your TV sits on the table but it’s not plugged into anything except for power. You take a little PC out on the deck with you — no wires attached — and on that little PC, you can download huge images from the Internet instantaneously because the sensors in your home are picking it up and shooting it to it. That’s the vision — it’s the instant Internet, it’s the infinite bandwidth Internet, it’s the mobile Internet.” But Szelag also admits that much of the vision is unclear because it assumes the proficient use of technologies, such as optical computing, that haven’t even been invented yet. So it seems that no matter how you try to hash it out today, the race for optical superiority doesn’t end with the networks alone. The optical players will continue to slug it out for years to come. Companies with names like Cisco and Sycamore seem ready to play a big role in shaping the optical world, but many of the key contributors may have yet to find a name. At any rate, in 2010, more likely than not you’ll get siliconindia beamed to you instantly. Or maybe you’ll still prefer the print version.