MG: BroadBand Brief -- The Broadband Killer Napp (Part 1 of 2)<br /><br /><br />11:12am EST 28-Nov-00 BofA Montgomery (Shapiro, Douglas S 212-847-5676) ADLAC T<br />BANC OF AMERICA SECURITIES MONTGOMERY DIVISION BANC OF AMERICA SECURITIES<br /> <br />November 28, 2000 CABLE AND BROADBAND SERVICES<br /> <br />Douglas S. Shapiro (212) 847-5676; Industry Overview<br />dshapiro@bofasecurities.com<br /> <br /> DJIA: 10590<br /> S&P 500: 1354<br /> <br />B r o a d B a n d B r i e f <br /> <br />IN THIS ISSUE<br /> <br />In this Broadband Brief we examine the impact of the rise of Napster and other<br />peer-to-peer (P2P) applications on cable and DSL providers.<br /> <br />Our conclusion is that while many pundits have realized the significance of<br />Napster in the evolution of computing (and as a challenge to intellectual<br />property).<br /> <br />.Few have grasped that it could be the broadband killer app because, to work<br />tolerably, it essentially requires broadband.<br /> <br />Our random sample of Napster users supports the point by showing that a<br />disproportionate percentage of users, over 70%, are using broadband today.<br /> <br />P2P and other distributed computing may pose a challenge to DSL providers and<br />cable operators by taxing the assumptions under which the network has been<br />engineered, but we think it will prove more boon than burden.<br /> <br />The Broadband Killer Napp<br /> <br /> Napster is mainstream. It's one thing for Red Herring or Wired to feature it<br />in a cover story or even MTV to air a skit about Napster during the MTV Video<br />Music Awards. But when Napster founder Shawn Fanning was featured on the cover<br />of Time a couple of months ago, it was clear to us that public awareness of<br />Napster, and the concept of peer-to-peer (P2P) networking in general, had<br />reached critical mass. So what does this have to do with broadband? In our<br />opinion, a lot. In a Broadband Brief we wrote earlier in the year (see Cable,<br />DSL, and the Big Picture, dated March 27, 2000), we argued that consumer<br />broadband penetration would ultimately accelerate much faster than anyone<br />forecasts because of network effects; i.e., the wide availability of bandwidth<br />should result in a proliferation of bandwidth-intensive applications that will<br />drive consumers to adopt broadband much faster than the prospect of simply<br />getting email and surfing the Web faster. We believe Napster is the best<br />example yet. While pundits have alternatively cited Napster (and P2P generally)<br />as a revolution in computing (not to mention a shot across the bow of<br />intellectual property holders), what we believe is not as widely grasped is<br />that P2P may be the broadband killer app the application that drives users to<br />abandon narrowband and embrace broadband in droves - because, to work<br />tolerably, it requires broadband. As we describe in this Broadband Brief:<br /> <br /> P2P networking is legitimately a big deal.<br /> <br /> Regardless of Napster's fate, the genie is out of the bottle.<br /> <br /> .And, as illustrated by our analysis of Napster users, P2P may be the<br /> broadband killer app.<br /> <br /> The explosion of P2P and distributed computing will be a challenge to<br /> service providers, but we think it will be more boon than burden.<br /> <br /><br />P2P Networking Warrants the Hype<br /> <br /> What is it? As it sounds, peer-to-peer networking describes direct<br />interaction between clients on a network. Napster, for instance, enables<br />clients hooked up to the network to access a centralized database that allows<br />them to directly share music files across the Internet. Why is it such a big<br />deal? Napster is arguably as significant a development in computing since Marc<br />Andreessen and his associates developed Mosaic in 1993, the first browser that<br />made client-server computing user friendly and really launched the World Wide<br />Web. The reason is that it continues the development of computing that started<br />with networks of mainframes and dumb terminals, moved on to standalone powerful<br />PCs, evolved further to the creation of LANs over which these PCs could share<br />resources, further grew through the interconnection of these LANs to create the<br />Internet and, with the development of user-friendly software (encompassing<br />HTTP, HTML, and Mosaic), WAN client-server computing (the Web). Napster makes<br />the next step in the evolution feasible by making client-to-client computing<br />accessible (and desirable) to the masses.<br /> <br />The Genie is Out of the Bottle<br /> <br /> Will Napster survive a legal onslaught? Last month Bertelsman decided to<br />break ranks with the RIAA and its fellow plaintiffs by agreeing to extend a<br />loan to Napster and drop its lawsuit (contingent on the company developing a<br />fee-based file-sharing network), suggesting Napster may beat what seems like<br />heavy odds. But whether Napster itself survives doesn't really matter. The<br />genie is out of the bottle.<br /> <br /> Consumers have sent a very clear message: they want it. In July 2000 Napster<br /> issued a press release stating that its client software had been downloaded<br /> over 20 million times, with the company surpassing 5 million users in<br /> February, 10 million by April, 15 million in early June and 20 million in<br /> July - arguably making it the fastest growing information technology ever.<br /> (By contrast, cell phones took 9 years to hit 10 million users, fax machines<br /> took 22 years, cable TV took 25 years, and phones took 38 years.) The tally<br /> now? Over 38 million users - with 5 million users commonly logging on each<br /> day. The consumer appeal was further hammered home the day after Judge<br /> Marilyn Hall Patel issued an injunction against Napster and traffic spiked<br /> 70% on the service as users clamored to get their last shot at downloading<br /> songs; that day Excite@Home saw a 30% boost in traffic across its entire<br /> network. And Napster has spawned clones and offshoots that include Amster,<br /> Gnapster, HackStar, iNapster, Jnapster, Knapster, Rapster and Wrapster,<br /> among others.<br /> <br /> Other applications will be much harder to stop. Strictly speaking, Napster<br /> arguably isn't a 'true' P2P application because it still makes use of<br /> centralized servers to manage data (such as users' registration data, file<br /> lists, logs of searches, etc.). That makes it efficient and relatively easy<br /> to manage (it also makes it theoretically possible to collect subscription<br /> fees one day) but it also means it can be shut down. But other applications<br /> have been developed that have no centralized components at all, most notably<br /> Gnutella. The Gnutella client software actually enables each node (or peer)<br /> on the network to act as both client and server, by simply relaying messages<br /> from peer-to-peer (not unlike one of those old Brett shampoo commercials:<br /> '.and they told two friends, and they told two friends.'). As a result,<br /> there is no centralized server and consequently no definitive way to stop<br /> it. It too has spawned numerous spin offs, like Gnotella, Gnewtella,<br /> Bodtella, Ntella, and CuteMx, all of which add some functionality. Aimster,<br /><br /><br /><br /> for instance, effectively combines Gnutella with AOL's instant messenger,<br /> enabling users to only swap files within a defined group.<br /> <br /> Music files are only the beginning. Although Napster is geared only to<br /> sharing music files, other applications enable sharing of any type of file,<br /> including entire CDs, photographs, games, or any other software for that<br /> matter. Besides Gnutella and its offshoots, applications like FlyCode and<br /> Zden also aim to permit sharing of several types of files. And another<br /> disruptive technology is waiting to hit critical mass. DivX (or, as it is<br /> officially written, DivX;-)), is a pirated version of MPEG-4 that enables<br /> full length DVD's to be compressed from multiple GB (a DVD holds up to 5GB)<br /> down to less than 700MB, small enough to fit on a CD or be feasibly swapped<br /> over a broadband connection. The technology therefore has the capability to<br /> result in the 'Napsterization' of the movie business, particularly once DVD<br /> burners become commonplace.<br /> <br /> .As is file sharing generally. File sharing is perhaps the most obvious and<br /> commercial application of P2P, but it isn't the only one. Another broad<br /> category of P2P includes sharing resources, such as processing power or even<br /> storage, or what can be broadly called distributed computing. For instance,<br /> SETI@home is a project of the Search for Extra Terrestrial Intelligence<br /> (SETI) that enables volunteers to download a screen saver that signals SETI<br /> when their computers are idle and enables it to potentially tap significant<br /> distributed computing resources. (Reportedly hundreds of thousands of copies<br /> of the software have been downloaded.) Among the numerous startups trying to<br /> harness commercial applications of the idea include Centrata, Cosm,<br /> Distributed Science, Popular Power and United Devices.<br /> <br />As Shown by an Analysis of Napster Users, P2P May be the Broadband<br />Killer App<br /> <br /> P2P is a seminal development in computing and, based on the overwhelming<br />consumer response, legal battles or not, it seems destined to proliferate in<br />some form. But what does this have to do with broadband? As we wrote above, in<br />our opinion, a lot. That's because P2P could be the broadband killer app - the<br />application that drives users to adopt broadband widely. The reason is simple:<br />to work well (or sometimes at all), all of this stuff requires broadband. Even<br />sharing relatively small MP3 files on narrowband is almost intolerable; forget<br />about anything more bandwidth intensive, like software, games, or video files.<br />And other advancements, like distributed computing, won't be feasible without<br />widespread packet-switched ('always on') connections. (We'd point out that, at<br />least conceptually, widespread P2P activity on narrowband networks would even<br />further threaten the ISP all-you-can-eat business model by taxing already<br />overburdened modem pools, perhaps eventually leading to financial difficulties<br />or higher monthly prices, which would also drive users to broadband. Although<br />that could take a while.)<br /> <br /> To test our point, we performed a simple analysis to examine whether the<br />distribution of broadband and narrowband connections among Napster users<br />differed significantly from nationwide averages. We conducted 20 sample<br />searches (ranging from Andrea Bocelli to Busta Rhymes to Eminem to Madonna to<br />Sinatra to Streisand to Zeppelin). Each search returned 100 results, yielding<br />2,000 total observations. (We conducted the searches over a weekend to minimize<br />the number of users attached through corporate networks.) Each observation<br />pointed to a specific file on a connected user's hard drive and, among other<br />information, included the name and path of the file, the bit rate at which the<br />song was encoded, the length of the song, the user's ID and, most important for<br />our purposes, the self-selected connection speed of the user. (When users first<br />install Napster, they select a connection speed from a drop-down list that<br />includes T3 (or higher), T1, DSL, Cable, ISDN, 56K, 33.6K, 28.8K, 14.4K, and<br />'Unknown.')<br /><br /><br />MG: BroadBand Brief -- The Broadband Killer Napp (Part 2 of 2)<br /><br />As shown in Figure 1, defining 'broadband' as ISDN speeds or higher, the<br />results clearly show that the vast majority of Napster users are using<br />broadband. For those who selected a speed when they installed the software, 71%<br />of the users in our sample were using broadband. As shown in Figure 2, this<br />proportion is (obviously) substantially higher than broadband penetration<br />nationwide. In addition, we note that even this analysis may understate the<br />actual proportion of broadband users. Anecdotally, many users who installed the<br />Napster client before they acquired broadband connections have not changed<br />their profiles since. And some have intentionally understated their connection<br />speed to minimize the likelihood other users will attempt to upload many songs<br />(and tax their upstream capacity). The result confirms what's intuitive: to<br />work tolerably, Napster essentially requires a broadband connection. So while<br />everyone seems to know what Napster is and the consumer demand has been<br />overwhelming, we think the less obvious implication is that it can't take off<br />without broadband.<br /> <br />Figure 1<br /> <br />THE VAST MAJORITY OF NAPSTER USERS ARE ON BROADBAND CONNECTIONS.<br /> <br />Source: Banc of America Securities LLC estimates, based on 2,000 sample<br /> observations.<br /> <br />Figure 2<br /> <br />.FAR MORE THAN NATIONWIDE AVERAGES<br /> <br />Source: IDC, Banc of America Securities LLC estimates.<br /> <br />P2P Should be More Boon Than Burden<br /> <br /> Most forecasts assume that narrowband will remain the dominant residential<br />connection for the next five to seven years. If broadband only offers the<br />ability to get email faster or reach Yahoo! sooner, we think they are right.<br />However, we consider that assumption myopic because it presumes the Internet is<br />static - which it isn't. We think the enormous and sudden popularity of Napster<br />is the most tangible evidence yet that the widespread availability of bandwidth<br />will result in a continuing evolution of new, bandwidth-intensive applications<br />that drive demand further. As such, Napster, or whatever form file swapping<br />finally takes, could be a big boon for service providers.<br /> <br /> Of course it could pose challenges too - particularly to the assumptions<br />under which cable operators and ILECs have engineered their networks. For<br />instance, cable engineers currently provision bandwidth to a node (or, for that<br />matter, a headend) based on observations that peak simultaneous usage on a<br />system is only 11% of subscribers. Similarly, ILECs and DLECs are<br />oversubscribing bandwidth at the central office (CO) to the tune of 50-100:1<br />for residential connections. But if users are leaving their PC linked to<br />Napster continuously (making it a near-permanent node on the network) or other<br />P2P applications take off, peak simultaneous usage rates will obviously rise.<br />Gnutella and similar decentralized applications are even worse because they<br />compensate for the lack of centralized databases by burning bandwidth (for<br />instance, by having each client send separate messages to many other peers<br />rather than search one database). And should other types of distributed<br /><br /><br /><br />computing become widespread - should consumers decide to subsidize their<br />broadband connection cost by effectively selling idle PC processing time, for<br />instance - the problem could get worse.<br /> <br /> How big will the problem get? In our opinion, it will require active network<br />management and it may require cable operators to tap some of their flexibility<br />to increase bandwidth per home sooner than they expect (see Broadband Brief -<br />Is 750 MHz Enough?, dated June 30, 2000 for more detail) or DSL providers to<br />provision more bandwidth at the CO than they do today. But we don't think<br />rising network usage will disrupt the economic model for either. For one thing,<br />per-bit network costs will continue to decline as optical networking technology<br />improves, helping offset the costs of rising demands on the network. In<br />addition, as they do today, service providers will likely subtly limit<br />bandwidth to users if they feel the network is being abused.<br /> <br /> But most important, applications that drive usage could be a blessing in<br />disguise should they cause users to upsell to higher levels of bandwidth. Both<br />cable operators and DSL providers are moving to embrace tiered service pricing.<br />DSL has the capability today and the next iteration of DOCSIS standards, DOCSIS<br />1.1, will also give cable operators the flexibility to sell tiers of service.<br />(Charter, albeit not using DOCSIS 1.1, has already started experimenting with a<br />256 Kbps service for $24.95.) As we've written before, we think the evolution<br />of cellular may be the best model. In cellular, price per minute declined<br />rapidly over the past decade, initially pushing down ARPUs, but they stabilized<br />and then climbed as users regularly moved to higher packages of minutes<br />monthly. We expect the same will occur with bandwidth. Per bit prices are<br />already dropping today as ARPUs decline, but we expect that just as the<br />elasticity of demand for minutes resulted in substantially higher usage, we<br />believe (a possibly greater) elasticity of demand for bits will drive bit usage<br />and, eventually, ARPUs, higher. In fact, one of the biggest revenue<br />opportunities for cable operators may not be in-home networking, home security<br />or any other number of other esoteric next-generation services, but simply<br />selling more bandwidth.<br /> <br />BroadBand Brief is a periodic publication devoted to emerging technological,<br />regulatory, and market issues concerning local access broadband deployment. It<br />is distributed via First Call and e-mail. To be placed on the distribution<br />list, please contact your Banc of America Securities salesperson or e-mail me<br />at dshapiro@bofasecurities.com<br /> <br />First Call Corporation, a Thomson Financial company.<br />All rights reserved. 888.558.250<br /><br />Nov-28-2000 16:13 GMT<br />Symbols:<br /> US;ADLA US;CVC US;CMCS US;COX US;ICOM US;CHRT US;TWX GB;5513100<br /> US;GMH US;AMER US;YHOO US;COVD US;NPNT GB;CMCSK0N US;ICCI US;CHTR GB;CHTR0EB<br /> GB;GMH0EB US;AOL GB;NWFX GB;AOL0N US;ATHM US;RTHM DE;AND XE;YHOO CA;AOLN<br />Source FC First Call Research Notes<br />Categories:<br /> BRI/55131 BRI/49428 BRI/59960 BKR/MG FCIN/BROCTV FCIN/PRTPUB<br /> FCIN/TELECM FCIN/ELECTS FCIN/COMPUT FCIN/DATAPR FCSU/IND GEO/US GEO/NME FCSU/MMN<br />