The U.S.A Needs Internet Speed – Can Cable Providers Deliver?
Where it is available, fiber is far superior than cable Internet service, when it comes to delivering faster Internet with lower latency, for various reasons. A major reason being that fiber can support larger amounts of bandwidth. Fiber infrastructure overall, does a better job of delivering for how we use the Internet today. Plus, all the vast amount of activities Internet service now supports. In 2023, Internet service is used widely for activities like streaming, VoIP calling, VR, AI, gaming, video conferencing and uploads (80% of all internet traffic is video). As well as downloading of large media files, from animations to HD video.
I’m also baffled why any politician would want to relegate their constituents to not only inferior technology, but to technology that has been deemed unreliable
– Gary Bolton, President of the Fiber Broadband Association
But when it comes to supporting critical Internet infrastructure for users in the United States, cable is far more available than fiber. Which is a little counter-intuitive when considering that more citizens now work remotely from home or not in an office.
In the below video organized by the Fiber Broadband Assocation, special guests discuss how cable ISPs may potentially satisy the demands of United States internet users for faster speeds, with upgrades to DOCSIS 4.0.
Fiber For Breakfast Video – Fiber, Meet DOCSIS 4.0
Fiber, Meet DOCSIS 4.0 Video Transcript
Gary Bolton: Good morning, everyone and welcome to the Fiber Broadband Association’s Fiber for Breakfast. We’re now in our 47th episode of 2022. I hope that everyone has a very Happy Thanksgiving tomorrow. Before we kick off, I’d like to thank our gold sponsor, Nokia. I really appreciate them being gold sponsor of Fiber for Breakfast.
You know, yesterday, a small group of Republican senators sent a letter to NTI Assistant Secretary Alan Davidson, arguing that NTI should not have disqualified unlicensed spectrum from be bid to being deemed unreliable broadband. NTI should be technology neutral. First and foremost, you know, it’s very unfortunate that some members of Congress remain either misinformed or uninformed on why fiber is undeniably superior and the only choice for building our nation’s critical broadband infrastructure for the future.
I’m also baffled why any politician would want to relegate their constituents to not only inferior technology, but to technology that has been deemed unreliable broadband for their critical broadband infrastructure. Fortunately, Mr. Davidson and the team at NTI are dedicated to funding and building the critical infrastructure that will serve all Americans today and for generations to come.
That brings us today’s Fiber for Breakfast session with Grant Joslin of US Telecom’s Equity Research and Credit Suisse, for an update on the cable industry’s latest efforts in a session title Fiber, Meet DOCSIS 4.0, cable’s upgrade path to 10 Gigabits. But before I formally introduce today’s guests, let me introduce Trish Ehlers Martin, who will walk us through some housekeeping items.
Trish Ehlers: Thank you, Gary. And good morning to everyone who’s joined us this morning. Before I go over a few logistical items, I’d like to once again thank our Fiber for Breakfast gold sponsor, Nokia.
Now, if you’d all keep in mind that everyone is in listen mode only. To ask a question, you can simply type it into the question box on the control panel on the right side of your screen. We’ll host a Q&A session with our panelists at the end of today’s webinar.
This presentation is being recorded and will be available to members only on FPAs website within 24 hours. You can find the recording in the Events tab under the Fiber for Breakfast dropdown option. At the conclusion of the presentation, you’ll be prompted to complete a very brief feedback survey, if you’d take a minute to do so, we appreciate your input. I’ll pass it back to Gary now to introduce our panelist and get us started. Gary.
Gary Bolton: Thanks Trish, and good morning. I’m Gary Bolton, the President and CEO of the Fiber Broadband Association. You know last week on Fiber for Breakfast we had a great session with Evan Marwell the founder and CEO of the EducationSuperhighway, discuss how state and local leaders can close the digital divide by removing barriers to free internet. We greatly appreciate Evan’s and the EducationSuperhighway’s efforts to drive broadband affordability.
Today, on Fiber for Breakfast we have the pleasure of meeting with Grant Joslin of US Telecom Equity Research at Credit Suisse for an update on the cable industry’s latest efforts in a session titled Fiber, Meet DOCSIS 4.0 – Cable’s Upgrade Path to 10 Gigabits.
Grant Joslin is the Vice President at Credit Suisse, with equity research covered in the US communications sector and leads telecom technology and regulatory efforts. As an equity research analyst, Grant models and gathers evidence about the telecom and media industry trends, producing reports and events to help inform investment decisions. His work to understand the telecom industry encompasses new consumer applications, household budgets, the outlook for new entrants, regulatory and industry structure and technical accounting.
Grant joined Credit Suisse in 2018 from Deloitte in London. And he’s the past winner of the AICPA’s Elijah Watt Sells Award and holds a B.S. and Masters in Accounting from Ohio State University, which that’s a good place if you want to be working towards a national championship. So hopefully, Grant, you guys will continue undefeated, all the way to the playoffs there. So with that welcome, Grant. And for our audience, please type in your questions and comments as we go, and we’ll get those into Q&A at the end. With that, I’d like to get things kicked off and hand them over to Grant.
Grant Joslin: Thanks so much, Gary, I really appreciate you having me here. It’d be a lie if I said I wasn’t worried for Saturday’s game, but I think it will be a really great one. So, the genesis of this report is that we did a lot of work over the summer and in the early fall about the DOCSIS 4.0 upgrade path. I think it was a really big area of misunderstanding for our investors, and sometimes in the trade press and things as well.
So, we’re going to spend a little bit of time talking about kind of the fundamentals of DOCSIS networks, which I’m sure I’ll be, you know, way below the level of those of you who’ve actually built them in a past life perhaps but maybe a useful refresher for everybody else. And then we’ll talk a little bit about the upgrade and what we think it means for these company’s strategies.
Similar to one of my peers that I think was on Fiber for Breakfast a couple of weeks ago, you know, we cover and rate stocks in the communication sector. So in addition to cable and fiber operators, we’ve also got wireless and satellite under our coverage, we’ve got media, music, and theaters as well. So this is just a little reference to what we cover.
Moving on to the next slide, I think I wanted to spend a few minutes setting the stage for how we look at the world for strategy. And so on slide, yeah, this next slide, we’ve got US Density statistics here, so this is kind of how we chop up the US into different places, and kind of the implications of what it means for the network. So you know, starting on the left, we’ve got deep rural places very sparse. And then and then rural, your core suburban and urban markets kind of in the middle or about the middle 40% of the US population, and then about 30% in dense urban and downtown areas. And we’ll come back to these a little bit later but they give you a sense for how the networks and the choices that operators will make are a little bit different in each.
And then the other piece of setting the stage is, you know, this is added up FCC form 477 data. So this is not the brand new data from last Friday, which I couldn’t get done in time. But you know, the prior census block level data. So this is kind of our map of the US in terms of the large operators and the small operators and how everybody intersects with each other. If you look all the way to the bottom right, what it kind of adds up to for us is that we think that currently, we’ve got about a third of the country covered by telco fiber, maybe a couple of percentage points higher because these numbers are a year or two old.
And then the last thing I wanted to spend some time on is how we look at the internet and consumer utility for the internet. So we had a nice big bump in the rate of broadband usage and traffic growth in 2020 and 2021, driven by the quick transition of everything to being done at home, right, school at home, e-commerce, work from home, e-conferencing, telehealth. And that slowed down a little bit in 2021 and earlier this year, but we think that the long term drivers of traffic growth are still pretty much intact. And we see that as on track to keep growing about 30% a year going forward.
And on the right is application use and the bitrate requirements to different application. So as a reminder, 80% of all internet traffic is video. It’s a high bitrate application, and it’s also a high amount of time spent. So when you think about internet traffic, you can almost throw away everything with video and just think about video traffic and you’ll be, it’ll be directionally right.
We did a report at the beginning of this year about the Metaverse and AR and VR and holograms and all of the kind of applications that were being posited, you know, late last year and early this year. And when we took a look at the network requirements, I think they were very demanding in terms of latency and loss and reliability and things, but still not really stressing networks in terms of just sheer bitrate or speed requirements. So even things like a really high end VR setup or holograms, we’re looking at 200 or 300 Megabits.
Which I think raises the natural question, why speed? And we attempt to answer that on the next couple of slides. So one thing is that – or sorry, so this slide it talks a little bit about the other aspects of quality besides speed. So the guys at Tutela, whose work we really admire, talk about all of their elements of consistent quality for mobile testing, and we like them just as well for fixed broadband testing as well. So you’ve got Latency, which is the round trip journey time of data, you’ve got Jitter which is how unevenly that’s distributed. Reliability for tests able to complete, the Packet Discard.
But the point is that you’ve got a lot of kind of different elements of performance. And I think that, you know, some of you, your eyes might be starting to glaze over, and certainly that’s how the consumer feels as well generally where consumers I think want to shop, basically with speed, and they want to trust that if they’re getting a high speed offering that it’s coming from a good quality network that’s going to do right by them on all these other metrics.
And as a result of that, I think consumers are putting a lot of their sense of utility on speed. And so on one hand, that means that providers are provisioning more and more speed, but on the other, it also means that providing higher top speeds of service and introducing higher service tiers also unlocks ARPU growth potential. And so that’s one of the big drivers of this DOCSIS 4.0 upgrade, just like it’s been a big driver of the 10 Gigabit upgrades for fiber operators.
And lastly, I wanted to talk a little bit about the halo and horns effect, which is that if you perceive one aspect, your attribute of a product is great, you’ll tend to think all other attributes of it are satisfactory, and vice versa. If something really important about a product disappoints you, then you’re going to be pretty furious at all aspects of the experience. So this is evidence from our consumer survey, which shows that the faster that consumers rate their broadband, the more overall happy they are with kind of every aspect of it.
And, you know, I think we didn’t ask this question, but the Recon Analytics guys have even asked, you know, how did you feel about the cleanliness of the consumer of the physical stores? And how did you feel about the IVR, or the voice response system when you call it? And even those metrics are rated much higher by consumers taking faster speed tier. So, you know, faster speed tiers are good, even in the absence of an application that might look like it demands it to a network engineer, one, because they unlock an ARPU lift for operators. And two, because they’re going to make customers much happier, stickier customers in the long term.
All right, so with that, let’s move on to DOCSIS networks and how they work. So this is an overview diagram of a DOCSIS network from sort of the internet at the top to the consumer at the bottom. You’ve got a core network running out to a CMTS, which was originally where the video signal was received and is now mostly what does the electronics communication with the individual modems further fiber out from there to nodes. And so the nodes are where a lot of the previous CMTS functions. They used to live in the headend are now being done. So they’re pushing those functions out. And the node terminates the fiber connection on one end and starts the coax connection on the other. So it’s actually doing all of the work to generate the electrical signal.
Now, a node serves about 400 homes passed or something like 175 to 200 customers. As the signal that the node generates travels down the coax, it attenuates or loses strength, and amplifiers take electrical power off of the line and kind of refill the strength of that signal and regenerate it so it can physically go further. So those are the active outside plant components. There’s about one amplifier per 30 homes fast. And then there’s also splitters and taps, and splitters and taps just physically branch the signal into two pieces so that you can screw one input cable in on one side and several outputs on the other. And then a little bit more coax for the drop. And finally you’re at the modem. So that’s the end of the overview.
I guess one key point I would say, before we leave this slide is if you look at, you know, a 60 million Home Pass Network all the way in the right column. So this is your kind of typical Comcast or Charter network, you’re looking at something like 150,000 nodes, something like 2 million amplifiers and something like 20 million or even more splitters and taps out in the network.
And if that was a little abstract, the next slide is a typical single node. So we’ve got our Passives, our Splitters and Taps in that teal color. You can see they outnumber the active components pretty substantially. You’ve got a pretty good number of amplifiers 13, in the case of this node. And this is a single node so that’ll give you a sense kind of for the number of components. And then off each one of these taps, you’ll have a kind of a small handful of homes passed.
OK. So the top equation you might have seen before, but this is a little bit about how this upgrade works. So the math that the operators are doing is that their Network Capacity is equal to the Total Bandwidth that they’re employing, times the Spectral Efficiency that they’ve got, which is the number of bits that you can get across each hertz of your bandwidth.
And the Required Capacity. So this basically says you look at your Average Peak Hour usage, and you multiply that by the Number of subs on the node and even when you’re at peak hour average usage, you want to have at least your Max tier bitrate worth of extra capacity as headroom above that. I think that’s kind of the same for any type of network. Both of these are.
And so the DOCSIS 4.0 upgrade, especially the ESD flavor of it is a bit of a spectral efficiency upgrade but also a pretty substantial bandwidth upgrade. And if you look at the bottom, you know, the DOCSIS 1.0 to 2.0 standards used up to 860 megahertz of spectrum, and most cable networks are still at 750 or 860 today. The DOCSIS 3.1 standard raises that to 1.2, which is about 50%, more than the 800-ish of the prior generation. So about 50% more capacity just for that. And then DOCSIS 4.0 either adds another 50% to that in the case of the ESD flavor that most operators favor, or it allows the first section of the spectrum, or about a third of it to be used twice. And so in either case, the DOCSIS 4.0 upgrade is about another 50% of total spectrum. So a big slug of frequency come in.
Moving on to the next slide, we’ll talk a little bit about what the upgrades actually entail. So the first step is to take the analog communication between the CMTS that’s you know, back in the operator’s headend to the node and upgrade that to digital so it’s digitally signaled fiber. And at the same time to push some of the functions that were originally done in a CMTS out to the edge where they can be done kind of closer to the user.
We think that unlock speeds somewhere in the range of three gigabits per second down, and 250 megabits per second up. So you know, three down, 250 if you’re not at symmetrical multi-Gig speeds by any means at this point, but certainly a big step up from where things were especially on the upload side, which is where a lot of cable operators are constrained today.
We’ve got a heat map over on the right in red and it’s a pretty limited upgrade in terms of what you have to touch, right? So even for one of the giant nationwide networks like a Comcast or a Charter, you’re really touching 150,000 nodes only. And then and then the code that’s in your CMTS as well.
The second step is harder, that’s where you’re going to all of those amplifiers, taps and splitters out in the network and you’re physically upgrading those. So those outside plant elements have a rated performance up to a certain frequency maps, and they can’t perform reliably above that. And so if you want to go from 750, or 860 to 1.2 or 1.8, that’s a component replacement. Those components are individually very cheap. It’s just that you’ve got to go into pedestals in the case of a buried network, or you’ve got to go up on a you know, a cherry picker truck for an aerial network and individually swap them out. And so you’re limited by just you know, labor, the number of man hours that you have available for the task and your ability to make sure you get all 20 million of those little components in.
And then the last step, fortunately, is easy. Again, once you’ve done the hard step two to step three upgrade. At this point, the CMTS is virtualized and so it’s literally a software upgrade inside the core, pretty easy to do. The nodes that were installed in step one are modular, and so it’s a, you know, it’s a little device like a hard drive and you swap open the clamshell, you unscrew a handful of screws and swap the new one in. And then lastly, to get this full, like 9 Gig or 10 Gig level of speed, you also need to replace the CPE the modem inside the house.
But fortunately, the DOCSIS standard is downwards compatible. And so that means that if you’ve got somebody who’s happy continuing to receive 5 Gig service or 1 Gig service or something, there’s no need to upgrade those. So it’s not necessarily that the entire, you know, set of modems of an operator needs to be upgraded in one go. It’s more that, you know, just for the early adopters, and then as the speed tier becomes more and more mainstream, they’ll kind of work that through over time.
So that means that step one is pretty easy and actually well underway. Comcast has said they’ve done about a third of their footprint, we believe. Step two is harder, and the components for it are just becoming available now. And then step three is, again, cheaper and easier relative to step two, but kind of not in the cards for a few more years. And I guess the last important thing before we leave that slide, sorry, is that this is a pretty flexible upgrade. And so you can stop after step one, and you’ve got a working network, or you can stop after step two.
And when we move on to the next slide, and we think about the different physical topologies and competition that we were talking about earlier. You know, we expect kind of different choices to be made in different areas. And so, if you’re in an area where you’ve got maybe millimeter wave fixed wireless and you’ve got one fiber competitor, two or three fiber competitors even, that’s the kind of area where you would prioritize it first. And as soon as you’ve got components coming in, you would love to do those upgrades.
Likewise, in a moderately dense area where you might not have millimeter wave competitors, but you would have, you know, one fiber builder plus one cable company, we expect those to get done pretty quickly. In a little bit sparser areas like a suburban area without fiber, I think that the cable companies will look at this as a defensive move and so they’ll kind of make the upgrade a little bit slower to try and scare off a potential fiber build if they can manage to add or at least to, you know, take advantage of a little bit of benefit on churn and a little bit of benefit on ARPU.
And then rural and deep rural areas, I think the upgrades will be more gradual. But the step one upgrade, at least is not a super involved touch every component of the network kind of thing. And so we think that upgrade will come quicker, and it’ll just be overtime and to the extent that fiber makes it into these more remote and more rural areas that we’ll see the more difficult more expensive steps two and three done.
A little bit on costs, you know, I think that everybody’s got kind of a different competitive set in different groups that they’re worried about. But the big question, I think, for investors was, is this going to totally blow up the capital spending plans of the cable companies and really expand the amount that they’re spending every year? And we don’t think that’s necessarily the case. So to give a sense, Charter and Comcast spend kind of $9 billion to $10 billion a year and their business as usual CapEx and in this case, we think the entire cost of the upgrade over the many years that it will be done is somewhere in that $10 billion to $11 billion range.
So the next slide is just a graph that shows the same thing. And then I think the last point that I would I would make for the cable operators is that it does come with some cost offsets as well. So you’ve got, in theory, better reliability from newer components, you should have a little bit of streaming telemetry where there’s active devices in the network that can fail, which is one of the differences between an HFC network and a PON network, you know, the cable companies would have to look at the modem and the headend and not really have a great idea what was going on between them and try and use process of elimination to find where a problem was. And now each one of those active components is kind of always going to be streaming back to the network operations center and saying, “Yeah, I’ll still get here.” So that should help a little bit with reliability and speed of restoring things.
And then the other thing I would say is that there’s also a CapEx savings, because about a billion dollars a year of spending at these large cable companies is just node splits where you’ve got a node that’s just oversubscribed and too heavy of usage. And so you go and you split it into two service groups to kind of tamp down the capacity usage on that. And this big slug of capacity that’s coming all at once should kind of offset a lot of that. And so there’s a little bit of a cost savings there.
And last, before we open it up for questions, I just wanted to shout out that we have an industry distribution list. So when we go and do this work for investors, we’re very happy to share it with the industry too. We learn all sorts of interesting things that we might have gotten 90% of the way right, or you know, a different operator’s perspective on things. And so we’re very happy to provide our perspective, you can send me an email there. And don’t worry about jotting it down, these slides will be distributed later, but just contact me if you want access to that. With that, I’m happy to take any questions you guys may have.
Gary Bolton: Grant, great stuff. I really appreciate all your hard work and research. But I mean, my first question is, why bother? You know, because if you look at – man, and this is kind of more philosophical, right is, does the cable industry really want to hang their identity on kind of a dying architecture? Or do they want to really focus on being a service provider and kind of distancing themselves from the cable moniker?
Grant Joslin: Yeah, I think that’s a really interesting question. And there’s a couple important debates in there in the industry. I think you’ve heard you know, most famously, among large cable, you’ve heard Altice US that said, you know, “We just can’t get the reliability characteristics we like out of HFC. And we’re in such a competitive footprint that we can’t get by with medium good reliability. And so for us, the right answer is to go, to fiber to the home.”
But on the other hand, you know, Comcast, who we visited earlier this year, and met with, they’ve said that for a really well-constructed cable node, that’s, you know, got new components and things that you really kind of need a microscope to see any reliability difference at all. And so that that hasn’t been the experience of other operators. Right? And I think that it’s really not a national picture of what the technology is capable of. It’s really about implementation and the quality of engineering forces and things whether the difference between the two technologies is inherent are something that can be bridged. But I think that at least at some of the organizations like Comcast, they’re confident that they can still provide a good user experience.
Gary Bolton: You know, I’m just – I mean, so the cables ain’t even using it, because like the telcos, they’ve kind of gone away from that moniker. Right? Because unless you’re AT&T where you’re American Telephone and Telegraph or you’re hanging on legacy technologies, but you know, you I would say like if I – you know, my observation has been all the small, many of the small cable operators have already moved completely to fiber to the home because, you know, they don’t have that huge legacy investment that Comcast or Charter have.
But even on the Comcast and Charter, I mean, they’ve been doing great, you know, the cable industry has been doing great on the broadband grants at the state level by, you know, basically bidding fiber to the home network so that way they win when they deploy fiber. And so what we’re talking about is just areas that aren’t going to be competitive, right? I mean, because anywhere there’s going to be competitive, they need to upgrade to fiber.
Grant Joslin: I think that in rural areas, the big difference is that, you know, a PON network, the signal can sort of go any reasonable distance and arrive perfectly intact. And with a cable network, you’re talking about a lot more amplifiers and splitters and taps for customers, you’re just talking a lot more power and costs and a lot more active components that could fail. And so to the extent that there’s a pretty big difference in performance in a deep rural area, like Charter’s build, they’ve talked about should be $5,000 per home pass this fiber to the home build that they’re working on, right? So very different from, you know, a major metro areas, downtown core and entering suburbs where you would have much fewer components. But I do think that deep rural is kind of a worst case scenario for HFC. And that’s why you’ve seen the cable operators very happy to build fiber when they’re doing that.
Gary Bolton: But like everywhere, in your charts where you have on those red, active components, I mean, every time you see that on a network diagram, that’s maintenance costs, operations costs, that kills your sustainability. And so until you can get rid of active components, you’re really going to have a hard time getting to zero carbon footprint and getting your OpEx down to anywhere close to fiber, right?
Grant Joslin: I think there – Yeah. So I think it’d be good to talk about what differences between fiber and HSC, this upgrade does and doesn’t fit. So I think that the advertised speed gap will narrow or close, the reliability of HFC should improve, because you’re going to have streaming telemetry from the amplifiers, the modems and the nodes. It should not improve entirely to the level of a fiber connection unless, you know, unless the cable engineers can really work their magic and do a great job on kind of like finding and eliminating the last few problems. So you will have kind of a difference there that should be diminished.
And then I guess the last thing that I would say about this upgrade that we completely didn’t realize until we went deep into the industry to look at this is that the nodes that are being put in today, as part of that step one piece of the upgrade are modular nodes, there’s a technology called GAP, which is the Generalized Access Platform. And if an operator decides that, you know, they’re, it’s no more use throwing good money after bad, or that they just don’t see any more lifespan in the DOCSIS technology. It’s just a module swap. And of course, you need to replace the physical coax with fiber as well and pay for drops. But you could potentially have a node where half of the node is supporting 80% of your customers on coax. And then you’ve got power users or really demanding customers or customers with the network characteristics are different. You’ve got a fiber to the home card that you screwed into that node for that 20%. And maybe that percentage increases over time.
And so it’s actually also it’s a, it’s kind of a more elegant way to get to fiber to the home, if you think, you know, everyone will be running fiber to the home in 30 years or something. It’s kind of a more elegant way than let’s revamp the entire network, burned it all down and put a new fiber to the home network then.
Gary Bolton: Yeah, 30 years, I don’t think we’ll know what cable or whatever, remember what cable was, I think more like two or three years. But the you know, so we get a lot of questions. I mean, so frequency is not all created equal. And so as you get these higher and higher frequencies, then it gets more difficult. Right? And I had a question in here, like, what is the frequency limit or impact on the coax network? You know, higher frequencies are harder to balance the network, at what point does the cable at the actual coax need to be upgraded?
Grant Joslin: Yeah, so I think it varies a little bit, but what would I – and my understand…
Trish Ehlers: Go ahead and answer Grant, I think we’re having some…
Grant Joslin: Yeah, I think we might have lost Gary for a minute, he must not be on a good fiber to the home connection. So my understanding is there’s about one more of these big frequency upgrades left so the drop cables that are widely in use by cable operators can go up to about three Gigahertz. And so that’s like another you know, 100%-ish boost in capacity and then some additional increase from spectral efficiency as well.
And spectral efficiency, to get a linear gain in spectral efficiency, you need an exponential improvement in signal to noise. So at some point that is just not worth pursuing anymore and you’ve kind of reached the max. But it seems like there should be another decade of service or more after the DOCSIS 4.0 upgrade if you ask the cable network planners.
Trish Ehlers: Wonderful. Well, it looks like we lost Gary. So I’m going to go ahead and close this out. Thank you so much Grant for joining us. I want to remind everybody to join us next Wednesday, November – let’s see, November 30th at 10am. Eastern for Fiber for Breakfast. We’ll have Workforce Development, The Clock is Ticking, with representatives of FBA and WIA’s workforce development team. So be sure and sign up for that.
Grant, thanks for joining us the day before Thanksgiving. And we hope our Fiber for Breakfast audience has a great Thanksgiving. Thanks and we’ll see you all next week. Bye-bye.