Providing advanced services and applications such as video-on-demand, interactive television, digital video, high-speed data and, eventually, Internet protocol-based voice services certainly are high on the minds of the cable industry's top technology thinkers. At the same time, a spate of technological
hurdles and barriers continue to litter that path, causing operators to blaze different trails in order to reach their goals. Does the move from DOCSIS 1.0 to
involve a simple software upgrade? When will voice-over-IP be ready for primetime?
editorial director Roger Brown, industry analyst and
columnist Leslie Ellis and
broadband editor Jeff Baumgartner recently hosted a roundtable to pore over these issues. On the call were Comcast Corp. vice president of strategic planning Mark Coblitz; Rogers Cablesystems Ltd. CTO Nick Hamilton-Piercy, AT&T Broadband
executive vice president of
engineering and telephony operations Greg Braden; Cox Communications Inc. senior vice president of technology development Chris Bowick; and Time
Warner Cable CTO Jim Chiddix. An edited transcript of the palaver follows:
MCN: Digital television, in terms of more channels
and a guide, is pretty stable. Can each of you discuss what
you'd like to phase in, in terms of client-server
or interactive applications going forward?
Our focus right now is VOD [video-on-demand].
MCN: And then what happens after that?
We are looking at a number of different things, but most of the time in our company we try and focus our people on something where you can understand where the business is and drive that forward, and so next year that focus is on VOD. There will be other things that we will experiment with and look at that have to do with advanced services and interactive advertising, and some other things, but they are more in the experimentation phase as opposed to the "let's get out and deploy it." And for the "let's go deploy it," that's VOD.
We [recently] launched WebTV on [Scientific-Atlanta Inc.'s] "Explorer 2000" as a commercial service to our customers. We've overpriced it at $19.95, not expecting too many people to come on, and we haven't marketed it other than word-of-mouth, and already we've got about 1,000 customers on. It's available throughout our Ontario systems. It's a WebTV Classic service, so it's got e-mail and walled-garden, specialized stuff and a back door where you can go out to the Internet. It's causing no technical problems.
Our emphasis is fairly similar to the one that Mark outlined. Our primary goal is to move ahead with video-on-demand. As you are probably aware, we are working with Diva [Systems Corp.], and we have, or will be, launching VOD services in several of our markets this year and next year, and then follow that with a move into interactive TV. We are currently working with Liberate [Technologies]. We'll have a trial beginning with them towards the latter part of this year, and we are planning to move into a trial with Microsoft TV the early part of next year.
MCN: It seemed that everything you've tried to do so far was great, but it was probably too much to do at once. Would that be fair to say?
No, I don't think so. We've had a lot of different activities underway, aimed at how to integrate these services together, etc. That's a challenge-there's a lot of moving parts on these things, there's no question about that-but I think our focus is clear in terms of VOD, then ITV. Now there's parts of both of those that we are continuing to work on, and continuing to get to the levels that we think they need to be before we go commercial with them, but I think they are coming along well. Tony [Werner, former AT&T Broadband CTO] did a great job, I think, of getting our focus and getting the right activities in place. I'm delighted to have the opportunity to pick up where he left off, but he certainly has put us down the right strategic path, and I'm pleased that we have the opportunity to keep pushing that forward.
You guys know, I'm sure, that we've actually announced VOD trials in a couple of our markets and we've got a full-scale rollout going on in San Diego right now. We've announced a trial in Phoenix, which is currently up and running to employees, and we will take that further.
MCN: Is that also with Concurrent Computer Corp.?
We've got both with Concurrent. We've also announced an interactive-television trial with Liberate on the S-A platform in San Diego, which is currently up and running with employees. We haven't yet taken it out to the field and consumers yet because Greg hit the nail right on the head-it's a very, very complex environment and we want to make sure it's up and running, and running well, before we do that. I hate to sound like a broken record, but the order of priority, if you look at those two, is video-on-demand and then interactive television.
MCN: In the Liberate tests, is that Liberate on top of PowerTV on top of S-A?
Yes, it is.
I don't think you can overemphasize the magnitude of rolling out any of these services. Just rolling out digital-on this call you've got people who represent cable systems that serve the great majority of the country-involves tens of thousands of employees who actually have to implement all this stuff. So getting the processes in place and so forth to first roll out digital, and once that is stable to roll out video-on-demand-those are enormous tasks. We're down the road on that, we've got digital launched in most of our systems. We've launched VOD to a couple hundred thousand subscribers in three of our systems with a lot more to come. And interactive TV is certainly an interesting future area. But the nature of the business is really unknown right now, and so, obviously that takes lower priority than getting this basic platform and this first interactive application that really does generate substantial revenue, VOD, in place.
I like VOD better [than WebTV], but it has a lot of challenges. Content, obviously; capital outlay, because you have extremely expensive QAM [quadrature-amplitude modulation] modulators sprinkled all over the place. WebTV doesn't need any of that capital outlay. It has operational issues, of course, so you have to have customer-service people fully trained to be able to support it. There are challenges like the French-English language in Canada, which took a little bit of extra software time. But frankly, looking in hindsight, it was the least stressful and least intense of any of our new-service launches.
MCN: Doesn't that make you feel good, Greg?
It certainly does.
MCN: Jim, have you found any points of harmony between America Online Inc.'s AOL TV and what your plans beyond VOD are, seeing as how it's at least hoping to be your new parent?
Well there certainly is something there, and we've had preliminary discussions, but I'm sure that we will eventually find AOL applications to run on our set-top box, but again, the focus for me is VOD for now.
MCN: One more interactive
question for all of you: With the electronic guide as the primary resident application in the set-top box, there are varying opinions about whether or not it makes sense to enable the guide for interactivity. For example,Gem
star-TV Guide International Inc.'s slant is
to put three 'clickable' things on the screen, and [the MSO] gets some percentage of the ad revenues. Can each of you discuss what you are doing with the guide?
Our guide is a very straightforward guide, it is not encumbered with lots of advertising and other distractions. It's a guide that is supplied to us by set-top box vendors. Both Pioneer [New Media Technologies Inc.] and S-A have supplied us with guides that our customers like a great deal. In these guides, the active picture shrinks into the upper corner of the screen and then is surrounded by guide information, which lets people navigate through the services that are on. Exactly how we will then go from there to interactive services remains to be worked out.
MCN: Chris, what's your thinking on that, es
pecially with what you are doing with WorldGate Communications Inc.?
It's really a very complex issue when you look at our marketplace. We've got the S-A guide deployed in about 40 percent of our markets, so ditto the comments Jim made on that. And we've probably got 40 percent [with the] StarSight guide on the Motorola Broadband Communications Sector platform, and the other 20 percent or so are TV Guide national addressable control. So we are a little bit mixed in that regard. What we haven't liked, I think, to date, is that TV Guide has been out proposing their business case to all the various MSOs and we just haven't liked the business relationship being proposed at this point, so hence, the announcement of moving forward with other MSOs to work on our own guide independently. I don't know exactly where that's going to lead yet. We don't currently have it in trial, but we do have plans to put it in trial sometime next year, and I don't have any specific timing for that.
I think everyone knows we are moving toward a DCT-5000 type of platform as quickly as we can, and in that environment, we think it makes a lot of sense to have a presentation engine as well as an execution engine. So, embedding interactivity in the guide itself is not something that we would necessarily advocate. [In some of the] boxes that we are using today, such as the DCT-2000s, doing some of that makes sense because of the memory capacity differences. But where you have the memory capability to operate those things independently, that's really where we are headed in our strategy.
MCN: Is everybody galvanizing around the thick-client
approach versus maybe something that's more of a thin-client approach toward interactivity and interactive applications?
It depends on what you mean by thin client. If you look at the DCT-2000 as being certainly not a thick client, and then come back to the statement that I made that interactivity [is] a very specific set of revenue- generating interactive services that can be implemented and make money for us now, then the DCT-2000 and the Explorer 2000 is something that we are all pushing for a lot of applications on them. So there's what Greg said, which is looking toward the thick client, where that doesn't take away from the millions of thin clients that we all have out there that can do interactive applications. I think you are going to see a mixed world.
MCN: This all comes back to what kind of
resource you are putting in the box, and one of the things that keeps coming up is the amount of memory.
Can you guys talk about how you plan to populate the box with memory next year?
I'm not so sure I can answer that. What I can tell you is, I haven't yet put a DCT-5000 on order in any quantity, yet. So I don't know exactly what that's going to look like. I will tell you that I will be putting out an RFQ [request for quotations] for set-tops for the next two to three years, and in that process we will be spelling out what our needs are.
Web TV just about fits in a 2000. It fits a little better on a 2010, and it works better and faster on a 3000. But it's still on a svelte client, I guess you'd call it. It's running on [S-A's] Power TV operating system, not Windows CE.
MCN: What's the current status of your rollout plans?
We are one of the unusual operators that doesn't run DOCSIS [Data Over Cable Service Interface Specification] 1.0 yet. We have 300,000 high-speed customers. We've been using a proprietary LANcity modem for about 130,000 customers and Terayon for the rest, and we are adding thousands more. We plan to launch DOCSIS using 1.1, and not go through 1.0.
MCN: Was that simply because when you started there was no 1.0?
Yes, and because the goal was to get market share and not let technology get in the way. We thought that the local telephone company, Bell Canada, would have been far more aggressive with ADSL [asynchronous digital subscriber line]. Luckily, they've stumbled, as you'd expect, and we've just kept going.
MCN: What kind of competition are you guys beginning to feel from DSL in the various systems in which you've launched data?
Certainly there's competition, but we are still in a position where it's all we can do to keep up with demand for additional installs. Even in Portland, Maine, where we got an early start and have a market share in excess of 20 percent, our connect levels are as high as they've ever been. Right now, the demand is still really unmet in aggregate by both of us.
We ended the third quarter with about 888,000 high-speed Internet access customers. Those are both @Home as well as Road Runner customers in our footprint. We're aiming to hit 1.1 million [by the end of the year]. We are doing about 3,700 net adds per day, and did almost 200,000 net adds in the quarter. We are seeing tremendous growth in this business; we are running hard to keep up with the demand. But we are seeing very aggressive efforts by Qwest [Communications International Inc.] in some of the markets in which we operate.
We've got 13 percent average penetration across the whole country now and there's absolutely no sign of it dropping off. In fact, it's hockey-sticking the other way. So this is probably a 30-percent-plus phenomenon when we are through.
MCN: Does the presence of DSL as a competitor just create more awareness of the category?
In Canada, DSL is identical in price to us. There's a marketing campaign where they talk about shared service and how useless it is, and we talk about living far away from the central office. So you slam each other in the marketplace, but the net result is that more customers come on. So I think its actually adding to the awareness.
I would agree with that assessment. When you consider the difficulties that DSL does have (with) installation with regard to footprint and capability, raising awareness on DSL, and then finding frustrated customers who can't get the service would certainly bring them to us quickly. Nick was talking earlier about proprietary modems, and we've got the whole range of proprietary modems out there too, so we're simulcasting both DOCSIS and proprietary modems and will for the foreseeable future. We're still growing. [Note: Cox reported 398,816 high-speed customers at the end of the third quarter.]
MCN: What about the transition to 1.1? How painful will that be? Is it really just a software upgrade?
: It's never as simple as "it's just a software upgrade." But in general, we have planned as much as one can possibly plan for that to happen. We had a number of proprietary markets, but everything we have been launching this year has been DOCSIS 1.0 with the plan to go to 1.1 as soon as there are certified modems and qualified CMTSs [cable-modem termination systems]. The modems that we have been supporting are the kind that could be shifted to 1.1. We have also been simulcasting. For some of the 1.1 delivery, particularly something which might focus on IP [Internet-protocol] telephony, we may even have separate spectrums that we supply both data and telephony for as we migrate into the full 1.0 movement to 1.1. So there are issues there but we want to deliver IP telephony, and to do that, we have to have 1.1.
Our path is very similar to the one that Mark outlined. We have an embedded base of proprietary, [and] we will have DOCSIS in all of our markets by the end of this year. We are looking forward, obviously, to being able to migrate to 1.1 via software upgrades. We know it's not going to be perfect, but that's the kind of plan of record that we are executing against at this point.
We have DOCSIS everywhere, but we've also launched a voice-over-IP trial in Portland, and it's been interesting that the bugs we've worked out there largely have to do with the gateway, and its functionality rather than DOCSIS itself. The DOCSIS that's there is not certified 1.1 but its well down that path.
We've all got a modified DOCSIS that allows us, in trials, to have quality of service on top of 1.0. It hasn't been certified or qualified, but it allows us to see how that's really going to work. It's been our experience that it is working fine.
With IP telephony as a driver, CMTSs have to be a much more rugged, carrier-grade part of the telecommunications system. Another thing that one has to consider, is that when we started launching data services, we had very high asymmetry between the downstream/upstream paths. Recent research shows that's getting much more symmetrical because of applications like Napster. So one has to say, if I'm going to 1.1, shouldn't I also use these so-called "turbocharging" technologies like Digital Furnace [now owned by Broadcom Corp.] to get myself more upstream capacity?
I would add to that the OSS [operations-support system] support that you need when you put the 1.1 platforms in place and [you] can do voice-over-IP and other applications. Understanding how that will work from an OSS perspective is also critical for all of us.
I want to add to what Nick said because I think people tend to look at what we are doing as just best-effort data. We are building robust IP platforms for a series of services that need higher data rates in aggregate, but also need availability, because people are going to depend on us for an awful lot of things, including telephony. I would say we need to get to [higher levels of service] availability because people will become more dependent on things that are presented to them by us. 1.1 starts as a set of technology for the delivery of product but has to continue into the kind of support that we need in order to deliver the future services which are going to be on the IP platform.
MCN: When you talk about voice-over-IP, are you thinking in terms of second line or teen line, or are you thinking in terms of lifeline?
We are actually talking about both, and the timing might be different.
Today, you can take your computer and make telephone calls. So you can think about it maybe on steroids. Then you use IP as a transport mechanism and soft-switch technology to deliver true primary-line telephony. Certainly that's in our plans. I think you'll see voice-over-data application first, which doesn't have all the same requirements to it, but we certainly are planning to go all the way down the path.
I would agree with that. Our whole broadband strategy is based on doing whatever we can to displace an ILEC (incumbent local-exchange carrier). So we believe the platforms that will allow us to provide VoIP have to have a degree of robustness and ruggedness so that we can match up with first-line quality requirements that customers have. But, we'll see some second-line applications, or voice-over-data type applications as Mark described, which we are also going to be pursuing, but our strategy will not be built and is not built around a second-line-only type of view.
We are in the switched camp at this point, and are moving down that path with quite a lot of growth. [Note: Cox had 205,676 telephony subs at the end of the third quarter.] That's not to say we are keeping our heads in the sand. In the last several weeks we've had every IP vendor known in house to see what the state of the fleet looks like now in voice-over-IP, so that we can better understand what our migration path may be. We, too, are feeling that you need a carrier-class type capability, 911-capable service with all the operational support systems behind it, at least right now, but you've got to start somewhere with voice-over-IP, and that would certainly bring up the second line initially as a possible launching point.
MCN: Chris, if you've had all those people in there talking to you about VoIP, what's its status?
I don't think it's ready for primetime yet, and I don't know that anybody would tell you it's ready for primetime yet. When will it be ready for primetime? I don't exactly know that answer yet. We are getting various points of view from the vendor community. But I think you will continue to see trials through the first half to three-quarters of next year before you see any real serious voice-over-IP launches.
With voice-over-IP, are you talking about end-to-end throughout? Or are you saying, if someone has a 5ESS [switch] and a GR-303 interface and IP in the access network?
It depends on if you've got switches, or you don't have switches, and then what your strategy is. Chris used the term 'ready for primetime' and that of course brings a lot with it, and the question is: Is it ready to be the primary line telephony solution? Then I certainly wouldn't disagree with Chris. If it's: 'Can we get this stuff working so that you could deliver real product to people? 'then it depends on what your definition of product is. I think I would like to start differentiating. I know we use the term 'second line, 'but that's not actually what it is. In other words, voice-over-data is not a second line. You don't hear the players in the marketplace today saying, 'I'm selling you a second line.' and there's a reason for that. Because the second line you get from the LEC has all the characteristics of the primary line. Lifeline is actually a $4-plus charge to someone who doesn't have enough money to pay for it. It is a primary line, it's just a different way of packaging and selling. And what we are talking about is something much more akin to someone who brings up a thing on the screen on their PC that can do a telephone call except we might be using a real handset or a regular handset, etc. When the circuit-switched guys like Chris talk about a second line, they are talking about a real second line. It's delivered with all the same characteristics, just as the LEC does.
When you ask the question, are you going to deliver a second line, I might answer 'No, 'but I am going to deliver voice-over-data.
MCN: When I refer to a second line, I mean non-lifeline.
I think that means that you're not going to drop your primary service provider. You're going to add this service provider with voice-over-IP. In the context of the question, I was taking it as, 'Would you drop your primary provider?'
That's different, then the answer is no.
Like Chris, we have a large base of circuit-switched. We [recently] announced that we just passed our 400,000th telephone customer. All of those customers in the broadband arena are circuit-switched customers, and we'll continue to grow on circuit switched.
So what that means for us as we think about IP, there's probably three paths of migration to IP that we think will come into play for us, and we will use a mix of those three paths as part of that migration. One path is, once a full end-to-end IP solution is in place that uses soft switch versus a gateway into a circuit switch, to deploy that type of foundation in markets where we may not have launched circuit-switched previously. We will also use what we call our IPDT solution, which is what we will be testing up in Boulder [Colo.]. That uses IP as your access technology through a gateway. And then the other piece will be migration of the circuit-switched HDT backplane to support both IP-originated calls as well as circuit-switched-originated calls. There's going to be a mix of that for us as we move into a full-IP environment, but we think that we can do that in a logical fashion, in a fairly elegant fashion. That's one of the advantages that we all have in operating with the type of broadband networks that we have, so that the risk of stranding capital in your circuit switched environment is minimal to none.
That's exactly what our plan is.
MCN: Regarding network power versus battery backup, what do you as operators do when the power grid goes down? What's the latest thinking on that for circuit-switched and IP?
We have a slightly unique situation because we have a national cellular operator, which we own as well. One of the flavors, which is under consideration, is actually having a cellular chip as part of the modem terminal adapter, so should the power disappear or the signals disappear, immediately you have a cellular connection giving you at least lifeline services to that home. One approach we're looking at is exploiting our cellular infrastructure as the backup for the wireline, or the cable infrastructure. If a customer says, 'I'd like to have your telephony service, 'we can deliver a cell phone, a portable cell phone to that person within two hours or less, so immediately they are on the payroll, and then later on bring the wireline connection.
We do network powering today. That is our standard and we will continue to move forward with network powering both in the circuit-switched as well as IP arena. In some of the former TCI properties, we do have some phone customers that operate off of an LPSU, or a battery pack at the house, and we are working over time to go back and eliminate that, and put them on network power. All of the work that we have done clearly indicates that lifecycle costs of network power are much better and performance is much better than battery power.
MCN: Are you satisfied with the powering needs of some of the in-home devices that would use IP for telephony? For example, what if you used the
DOCSIS modem built into the set-top to handle telephony? We've heard some discussion in the past that set-top boxes and cable modems, as specific examples, were never designed to run in a low-power mode when the power was off.
No, and we wouldn't expect those devices to be the source of IP telephony in a non-powered mode. Cable modems, set-top boxes-they all plug into the wall. We really envision that we would have a network-powered network interface unit at the side of the house, just like we do in a circuit-switched arena today that has a low-power-consumption DOCSIS modem in it that will allow dial tone to continue to flow to the house, even if you've lost commercial power there.
We've got an IP telephony trial up, and we think that (IP) is in our future. I think we are perhaps more skeptical than the rest of the industry, or at least some of the rest of you, about the economics of what is called lifeline telephony, power backed-up telephony, circuit-switched telephony, which we think have very expensive cost structures. Given the unknown voice market, the unknown characteristics of the voice market in the future, putting a lot of capital against in voice product seems imprudent. We'd much rather have a voice product on a low-cost basis and- regardless of what happens with voice, with all the competitive factors of IP from various sources and wireless-we can still have something that adds little incremental cost and still can be marketed at whatever the market will pay.
MCN: What are the
concerns and challenges regarding home networking? Where is the business there?
We have to think about home networking from two different perspectives. The first perspective is the one that I think many of us are already doing, which is through our online businesses, hooking PCs together and helping our customers' network and making a high-speed-data connection more valuable. What it doesn't do is create ongoing revenue streams, other than the one for the high-speed connection. There's another world which asks what happens when all kinds of things are connected to an always-on IP network? And if that's the term of home networking, then the basic challenge there is the definition of what's really interesting to a customer and then taking the characteristics of what we have in cable, probably most represented by the work we've done in DOCSIS and in PacketCable, and extending that kind of capability through the home networks, so that might result in things like quality-of-service, and various forms of signaling among devices, and how those networks are monitored and service provided to our customers, which are going to take time to develop. We have the here and now, which adds value to our customers in the PC world, and then we have the future in which we look at all kinds of devices, not just PCs, as being members of a network in the home connected to the outside world and also connected to each other in the home. There are lots of challenges there.
MCN: We've been hearing a
lot about the so-called residential gateway. Where do you see that going?
Where will this device reside once all those different services are flowing through it-inside or
outside the home?
I don't think we have a strong opinion on where it is, but what it does. Given the fact that our strategy is built around a full suite of voice, video and data services, and knowing that some of those services, if we are going to displace the incumbents, will require things such as lifeline capability for the voice service. That tends to say that we are going to have to have at least some part of the functionality on the side of the house. Whether it's all there and then you have a different type of gateway approach inside the house, I don't think we fully understand yet.
But one of the points of home networking is that perhaps it doesn't matter where it is.
When you think about it, our customers are doing it, and they are going to do it with us or without us, and Mark asked earlier, how are we going to make money at this? That's something that I think we're all struggling with and trying to figure out. Internally, we have a group that's tasked with trying to figure it out from a business-modeling standpoint and, to try and figure out how we help our customers right now with their current home-networking needs. Then we need to figure out how we can evolve that into a business, longer-term, with a so-called residential gateway.
You could ask is this business a service business, or is it a technology business? Do you think customers buy their things, and you provide the service now so they can hook things together and manage them for the customer? Or are we talking about this as an extension of our normal service? I think that's the business rationale that one has to examine.
MCN: So it really sounds that this is more of a business issue than it is really a technical issue?
: Well, there's clearly a business issue overlay. There are a lot of people who are building technology and there are going to be lots of choices as to how to hook up things together, but some of them provide positioning within a business model that might work, and some of them do not. At the end of the day, it is all going to be driven by what business models actually work. I expect it will be in each of those places, depending on what it is that we need to do. There are gateways that are little tiny boxes that can sit somewhere and do a certain set of applications, then there are gateways that are complicated servers, and more complicated than a PC. One of the issues will be, 'Where is stuff stored? Is it stored in the home on a separate device? Is it stored in a gateway? Is it stored in our headend? Is it stored on the Internet? If there's an application running, who makes sure that that application actually works? And if there's multiple applications, who makes sure that the multiple applications actually work? 'So I think all of those issues are significant challenges for home networking in general once you get beyond hooking a couple PC's together, which is relatively straightforward today.
MCN: Could each of
you step through your
rebuild/upgrade progress that you've made to date and how that compares to what your goals have been. Is there a time coming shortly when the spending spigot is turned off?
We are essentially done with the properties that we owned when we announced our upgrade project. In the last year or two, we have acquired a bunch of new properties through trades and partnerships, and those are all in one stage or another of upgrade. But our 20 percent of the country is by and large upgraded, and we are mopping up the remains. Having said that, we do have work going forward both in terms of expansion and in terms of this mopping up process next year, but our spending on the plant itself is going to taper off beyond that and shift over to the terminals that go at the end of the plant and help us realize that the business that this all enables: modems, set top boxes and so forth.
We're in somewhat of a similar situation. We were also on an acquisition binge a while back, and, if you look at all of our systems-including the acquisitions, many of which were not upgraded-we'll be on target for this year [for] what we were projecting, about 70 percent completion by the end of this year. We are shooting for something upwards of 80 percent completion by the end of next year. If you looked at our top 15 markets, which represent 70 percent of our customers, then those numbers increase by about 10 percent each, so its 80 percent this year and 91 or so percent next year. I don't see a significant decrease in capital spending next year. As a result, we've still got a lot of upgrade activity to do but after that I can begin to see it trail off.
: I think we're all that similar position, in that we've done acquisitions and we're actually trading out [you're welcome, Greg!] some upgraded systems for un- upgraded systems and we've been doing this periodically and the last group of them, which will be probably in the next year, which are part of an AT & T Broadband swap, we'll have to continue with those. We're going to end this year with almost 8 million of the 12 million homes passed that we finally end up with being fully upgraded and an additional group that have been upgraded to a slightly lesser degree. The target for 2001 is that 11 of the 12 million homes passed will be done and then I think you will start to see some tapering off once we've got all of the things that we've either acquired or traded for completed.
: As of the end of third quarter, about 78 percent of all of our plant had been upgraded to 550 MHz or higher. The vast majority of that is 750 MHz. About 73 percent of our total plant is two-way activated. We will continue with a strong push next year to complete most of the balance, with a little bit of that moving into '02, but one of our key focus points is to continue to get the upgrade and rebuild done just as quickly as we can. As we are doing that, of course, we are rolling new services on to that just as quickly as possible to begin generating revenue. In the third quarter, we reported that we grew, we added an incremental 600,000 RGUs [revenue-generating units], which cover digital-cable customers, high-speed Internet access customers, and telephony customers. So as soon as the plant's ready, we're selling. [Spending] will begin to taper off in 2002. We are also acquiring some properties, so we've got work to do there.
The [systems] we've acquired in the Shaw [Cablesystems G.P.] swap haven't been rebuilt. I think our money's going to go into getting further segmentation for the traffic, putting servers, maybe QAM modulators for VOD out into hubs and all that type expenditure rather than the old classical coax-fiber type of expenditure.