We good? All right. Good afternoon. I'm Vineet Chhangani. I'm a Managing Director of Morgan Stanley's investment banking. Just for important disclosures, please see Morgan Stanley's research disclosure website. If you have any questions, please reach out to your Morgan Stanley sales rep. Okay, great. Thank you for joining us today. We're here with David Morken, Co-Founder CEO of Bandwidth, and John Bell, Chief Product Officer. David, John, welcome to the conference.
Thank you, Vineet.
Thank you.
How's the conference been so far?
Outstanding.
Great. why don't we just start, you know, let's just go for the audience who are not familiar with this story. David, I'd love to, have you talk a little bit about what is Bandwidth all about. Talk to us a little bit about the platform, the trajectory, a little bit of the history.
Happy to, and have to start with I have henna on my hands. I'm not Mike Tyson wanting to have, like, a permanent tattoo. About 70 hours ago, I was in New Delhi when we decided to attack Iran, but I'm happy to be here today. I'm on fumes, Vineet, please be gentle. Origin story for us, founded in 1999 and began selling internet connectivity back then, but then in 2007, with Google as an anchor tenant, built out a nationwide network. We like to call ourselves the last CLEC ever built, but in all 50 states we were able to offer voice services, and then from that beginning, expanded through acquisition to now serve 65 countries with a global voice network and software platform on top of it.
We do cloud communications for enterprise customers, including all 12 of the Gartner Magic Quadrant leaders in conferencing, CCaaS, UCaaS, and then many enterprise customers. Based in Raleigh, North Carolina, and went public in 2017 because of the outstanding leadership of a certain individual at Morgan Stanley who's doing this interview.
Thank you. John, maybe one question for you. You've been with the company 14 years now. You've seen the evolution of technology. Talk to us a little bit about oftentimes people hear CPaaS, CCaaS, UCaaS. Where does Bandwidth sit, and what makes Bandwidth different than some of the other players in the space?
Yeah, it's great question. The moment we're at right now, there's a lot of things changing. If you hear CPaaS is enabling platform, a lot of people will think it's text messaging, and that's where a lot of the industry's been. We've been off in a different area. We've been very focused on the voice side of CPaaS and enabling the CCaaS providers, contact centers, the service, the UCaaS providers, contact center apps, when they go to market to SMBs, large enterprises and bundle telecom in, customers that way. We've also been selling directly to large enterprises. What's interesting about this moment now is everything used to be separate. There was digital channels, and they did one thing on their own, and there were voice channels that did things with human beings on the other side. That is changing.
Voice is becoming a digital channel and being tightly woven in with other digital channels. It's a very exciting time for us as voice AI is changing and really becoming front and center in what we see happening, and clearly in the core of decisions we see enterprises making now as they choose their infrastructure.
Mm-hmm. We should call out you cater to large scale enterprise customers. You have Maestro cloud communication platform that sits on top of your owned and operated network. Talk to us a little bit about, like, we're in a very interesting time right now. Investors have seen the CX, the CCaaS bucket, if you will, get disrupted or pressure on their stocks and valuations, but it feels like you're sitting in a very interesting spot that's powering them. Talk to us a little bit about, like, how's that strategically differentiated? What does that mean for you? Maybe, David, you start, and I'd love some product insight on it.
Our view is the next 1 billion users of the PSTN globally are AI voice agents.
Mm-hmm.
They may be performing the function that is currently fulfilled in a contact center. They may be fulfilling the function of a knowledge worker. None of them need a mobile phone. All of them originate and receive their calls from the cloud. All of them need a global network to be able to communicate. What we have and our vision is to be the network and platform of choice where you can orchestrate an AI voice agent that you built with ElevenLabs, or you built it with Sierra, or you built it with Vapi or somebody else, but you need to bring it to life around the world for whatever its mission is, and we want you to be able to bring that voice agent experience through the Maestro platform and our global network to life.
You have to have ultra-low latency, high fidelity, resiliency and reach. We have 65 countries where we're full PSTN replacement, and we own and operate our network. That is ideal for this next wave of voice, and we're already starting to see that in the voice growth rates that we've seen in 2025 versus 2024 and going into 2026.
Got it.
You wanna add to that from a product-.
Yeah, I think we're asking kind of what's changing. CCaaS customers, great customers of ours, love to serve them. There's a lot of great innovation as they are bringing AI into their platforms. The opportunity is for the enterprise that wants to go best of breed.
Mm-hmm.
It says, we wanna do something that is not within the capabilities of that current CCaaS platform, and it wants to bring in a highly verticalized application here that wants to train their people differently, that wants to do things differently, and that is the opportunity that we're enabling with our Maestro platform, being able to bring in many of these voice agents at the same time to create really differentiated B2C communications experiences for the enterprise.
Why is it that you can do this better than the others? Who are the customers choosing between? Is it more like they're working homegrown solutions? Who are the competitors in RFPs? Tell us a little bit about that differentiation.
In the fourth quarter, we announced four new large enterprise voice customers. One was a top 10 bank in the U.S., another a household name in insurance that you would recognize immediately. All four were winaways from Verizon, AT&T, Lumen, and represent the differentiation we offer with the Maestro software platform. If you're going to do agentic call flows, or in one case, if you have a Cisco environment, but you need Google's AI solution, Maestro lets you orchestrate and send a call first to the AI sentiment analysis engine simultaneously to the called party. That capability, Verizon has nothing like.
Mm-hmm.
AT&T doesn't have an orchestration layer for any of these kinds of call flows, and Lumen as well. In all four of these cases, they illustrate the primary competitive dynamic we see and how we win. That orchestration layer, that software platform on top of the global network.
Mm-hmm
... is what wins the enterprise and I think that the next generation of voice from these enterprises will largely include voice agents.
Interesting. David, maybe switch gears a little bit. Interesting business model. You know, you know, investors have always kinda looked at software versus hardware, but you kinda have a good mix and maybe now that's playing to more of the strength given your, you know, owned and operated network. Talk to us a little bit about what's the most misunderstood aspect of Bandwidth business model right now for the investors?
Yeah. Historically it was, you really have a software platform, you're just the network. It's forget about the software platform, let's talk about the network.
Mm-hmm.
Wait a minute, you don't have SaaS seats, you have usage. What, what the heck is usage? How do you predict that?
Sure.
For 32 straight quarters we've met or exceeded guidance.
Mm-hmm.
Predictability is something we've got a good handle on. We have an outstanding cloud communication software platform. For some reason, it is rather challenging to understand that we have a vertically integrated voice solution that includes the ability to engage with voice orchestration through software, but control the delivery, quality, reliability, latency of the voice call on the network.
Mm-hmm.
That's a unique model.
Yeah. Just staying with the voice, you obviously had great year. You just reported your Q4 full year results. We're seeing the growth accelerate. You're delivering on the margins. Talk to us a little bit about like what have been the key drivers, and you obviously gave great guidance going forward, good mix of growth and profitability. Tell us like what gives you the confidence, like how are things changing?
One of our segments, Enterprise, is growing at 21%, and we've seen voice accelerate from 3% in 2024 to 8% in 2025 for global voice plans, and it exited the 2025 fourth quarter at 12%.
Mm-hmm.
That voice tailwind comes from the agentic voice moment that we're in. We have installed customers that are beginning to finally scale beta and alpha and early R&D projects in voice agents, and that is manifesting in the voice growth rate. Messaging is growing at the rate of the market. We're happy with that.
Mm-hmm.
What we're most excited about is in the past we have been cyclical around political messaging in one year versus the next, and what we see going forward is double-digit cloud comms growth paired with 20% EBITDA %.
Mm-hmm.
That excites us to be a consistent grower in the future.
Got it. Staying with that a little bit, which is a key theme in our conference right now, just given what we're seeing, the SaaS disruption, if you will, and, you know, the ability to just wide code what software can do. Maybe, John, a question for you. How do you think about that in your space? You know, as you think about your defensibility around it, is there anything that could disrupt your, you know, the orchestration layer, the Maestro? Can somebody use an LLM and put a tool on top? Or how do you think about that?
Yeah. I don't think of it in a defensive way.
Sure.
What's really important to us is our global communications network. Everybody talks. You'll see great demos, and people talk about the functionality that can be built.
Mm-hmm.
To really deploy it at scale in enterprise, it needs to be fast, so low latency.
Mm-hmm.
Right? Your voice AI experience has to be low latency. It needs to be high quality, which means you need to control the media. It needs to scale, it needs to be resilient and redundant.
Mm-hmm.
It needs to be cost-effective. You cannot do that if you're operating a software layer that sits on some legacy telco's network.
Sure.
It just doesn't work. You have no control, you can't deliver the latency. It just won't work. That's really what we enable. It's more than just the functionality, it's actually the performance. Will it work at scale reliably for an enterprise? That's what our platform enables. We continue to expand our coverage, so we can provide that experience that just works and always works.
Mm-hmm
... for our customers.
Great. Talk to us a little bit about just the roadmap, let's just call it next 12 to 18 months, both on the product side and also as you think of geos, how are you thinking of the rollout? You mentioned 65 countries. Maybe we start with the product first, then talk to us a little bit about the geos after.
Yeah. Obviously voice AI is really important to us as we look at the tech stack and where we believe we fit in the tech stack. We'll continue to do investments there. We're excited about supporting agents. We're excited about opportunities for our own agents we can bring as well. There's an area that's really important. Our Voice API, we continue to grow. We find that there are a lot of new entrants coming into the market, not just enterprises, but platforms. We continue to expand our Voice API and its capabilities. A lot of traditional customers want to work with SIP. That's great. You wanna work with Voice API, that's great. You wanna come with WebRTC, great. We continue to expand there as well.
We have a very flexible platform that lets our customers choose the technology they wanna work with.
Mm-hmm.
We continue to take a very open approach. We are 100% open. We will bring our own services. We'll also make sure our platform works for the best of breed services that our customers wanna bring as well.
Mm.
In terms of jurisdictions, we have large current customers that are pulling us into new countries.
Mm.
That's been really important in places like Brazil because it underwrites the nominal CapEx to get there, but in a way that we have a return on that investment within a reasonable period of time. That's exciting for us to expand into certain key jurisdictions, and we'll continue to do that. That's within our guidance for 2026.
Got it. David, how should we think about when you think of an international market owned and operated versus partnership? Like, how do you think about the network when you go to a country like Brazil or...
Yeah. We have what we call our network is the universal platform.
Mm.
That's a consistent set of hardware and interconnects in a regulatorily compliant manner. Within Brazil, you end up only one hop in terms of proximity from the core of the network, and that's vital for low latency and reliability, the codec flexibility that we have by managing our own hardware in country. This is very different than just reselling somebody by commercial agreement in Brazil.
Mm-hmm.
It gives you the visibility into the performance of the network and its conduct, and you can have, you can have failover and resiliency, but it takes time. You're also legally in front of the regulator in a way that's appropriate so that you can provide service. In our case, when we say PSTN service, that's inbound calls, outbound calls, a phone number, and emergency service. The 911 equivalent in that country. That's very important to an enterprise customer for us, folks like Microsoft or Google or Amazon. We have large internet hyperscalers we serve as well as large global 3,000s.
When we open up a new country, it's a real investment in time and money, and that's really what drove our acquisition back in 2020 of a company that had spent 15 years building out around the world in this fashion. That was vital for us to expand our footprint to serve global enterprise.
Sure. Sure. Just thinking through, you mentioned how, what role do you play with Sierra of the world, Decagon of the world. Talk to us a little bit about, like, how do you see your pricing, your business model evolve as you start supporting, let's just call it like the next gen of CCaaS platforms that are coming out?
You will build your voice agent at ElevenLabs or at Sierra. In a local environment, you will tailor it and guardrail it and make it observable and be very excited about how empathetic and intelligent with the right context it appears. The moment you take it out into the wild and deploy it globally on a phone network, if that network partner does not have a high quality, ultra low latency network, the voice agent will start to annoy you to death.
Okay.
That voice agent, however, has an enormous number of things going on in the background. The media of the call from the voice agent to you needs to also simultaneously be routed to a sentiment analysis engine, a fraud engine, a speech-to-text engine, a transcription service. Each one of those legs of that call for us is one minute of usage. What used to be one minute at $0.002 per call is now $0.10 per call.
Mm
... because we're doing five things simultaneously. It's not just that we think the next billion users of the PSTN globally are voice agents. Each call is a multiple revenue opportunity for us because we are very expertly handling the media and the signaling to not just the called party, but every aspect of the voice agent stack that needs to be invoked. By the way, your time budget to round trip the inference and reasoning stack is like 300 milliseconds. That's also your communications budget before you hang up.
Mm-hmm.
if you're at 600 total milliseconds, you're gonna abandon the call.
Right.
Suddenly our very traditional, incredible voice infrastructure is now suddenly extremely important at a very competitive dimension of latency.
Interesting. Are you seeing those use cases as you've seen some of the recent RFPs? Like, what are you seeing?
We're seeing hospitality partners like Wyndham.
Mm-hmm
take to market a voice agent for the concierge desk that does precisely what I just described in routing a call to a voice agent that's also doing transcription, sentiment, and other things simultaneously at great effectiveness.
Is it fair to say that David Morken's view is even if call center reps go down because of the voice agents, the volume is actually going up, which is a driver for the business?
Do you know the number of voice agents in the contact center that are going to exist? It's infinite.
Mm-hmm.
It's as many as are needed, spun up in real time, 24/7, as empathetic as Dr. Oz and as smart as a Nobel laureate, aware of your entire consumer context. You will demand to talk not to a representative, but you'll be talking to a representative, and you'll be saying, "AI agent, AI agent," on the voice tree.
Interesting. Okay. Let's switch gears to messaging for a second. You published the State of Messaging Report back in January. We've always heard about RCS, but it seems like SMS still sort of owns majority. Tell us a little bit about, like, what you're seeing and what's gonna be the impact on your business model going forward.
Yeah. RCS is exciting. I think what you're seeing now is it's still the early days of people understanding how to use it.
Mm-hmm.
People look at it and say, "Oh, I could make my existing...
Transaction look a little better. Well, it costs more, and they do the ROI, and it might not be worth it. I think what we're excited to see is people starting to realize that you can do more with RCS. Like, it can do more mobile, maybe mobile advertising lead into commerce, whereas a lot of text messaging is super transactional. It was a one-way information message. It's really getting in the hands of the right people. We're investing in helping people develop content for it because you didn't have to really develop content for a text message or a picture message before.
Mm-hmm.
With RCS, it's actually much richer. It's actually challenging our customers to do more for it. There are many different ROI opportunities they have with different use cases, and that's the excitement. It's brand new, so it takes a while for people-
Yeah
to realize what they can do with it.
John, just staying with that, like, one of the things that we're hearing is commerce itself. Let's just talk about commerce for a second. Commerce itself is going through an evolution with AI. You're seeing new protocols. OpenAI and Stripe came out with ACP, and then Shopify and Google do UCP. The idea is that the front end would be some chat-based prompt. You find something, and you can make the transaction happen there. Alternatively, in the past, you could do some of that on your SMS, and there was messaging, and you'd get paid from your merchants. What are your thoughts on that? Like, do you think the volume and the transaction can move there away from SMS? Are you seeing anything or-
I mean, it's still we're both going to say yes. Also from apps as well, right? Because so much of this was locked up in apps in the past, and there's a lot of friction using an app as well. So there's a lot of opportunity as the interface becomes more natural.
Mm-hmm.
Like the way you do things in a brick-and-mortar store with your voice talking to somebody.
Mm-hmm.
There's a lot of opportunity for communications and commerce to remove a lot of this friction that's existed in the past.
You still get to be the backbone of. Like, I'm just trying to picture where Bandwidth gets advantage of this change, or is it a disadvantage? Like, how should we think about that?
I think of it as we are operationalizing AI and AI-driven communications. Across all of the channels, making sure it's consistent and persistent conversation between the consumer and the business, that's the value of the CPaaS platform of the future.
David, you know, again, just a question on, like, disruption, and this is my last one on that one. Talk to us a little bit about the moat. The customers that you have, you know, we've heard from some of the other software players talk about integration and, you know, the compliance risk that you would have. Talk to us more about as you think of next 12, 18, 24 months, your retention numbers were amazing in the last quarter. You know, as we think of the next 24 months, like, what drives that retention even higher? What keeps Bandwidth the platform of choice?
yeah, Net Revenue Retention of 107%, and that goes up with the use cases in voice that are expanding and that AI voice tailwind that we're talking about. The competitive dynamic or the moat regarding how we continue. I should say we have basically zero logo churn.
Mm-hmm
... our customers, once they're with us, stay with us forever.
Yep.
The usage that they have with us will continue to grow, depending upon those use cases. The moat that we enjoy, even though John doesn't wanna talk about it as a defensive thing, is really important because it would take an enormous amount of new capital and an enormous amount of patience because you open up each of these country jurisdictions and even the 50-state Public Utility Commission processes for a CLEC, you move at the speed of government.
Mm-hmm.
That means eons. It's a glacial pace. It took 15 years to get to the 65-plus countries we have today. Many of these countries aren't talking to new entrants anymore. I don't think it's anything that we will see, nor have we seen a new entrant in over 10 years.
Mm-hmm.
In terms of the defensibility, again, there's a lot of regulatory compliance. There's much international work that has to be done domestically. There are, as I said, 50 jurisdictions. You have to be all over TCPA and the robocalling rules and be vigilant as heck about gatekeeping the right actors, and there's a lot of vigilance involved. All this to yield us as a single partner for our existing customers to come to market. In some cases, like the hyperscalers I mentioned, we're serving 30 different products for them.
Mm-hmm
... today that they bring to market. That's exciting for us. I'll just back up for a minute, Vineet, and touch on texting. The last, you know, decade plus, certainly since the iPhone in 2007, has been dominated by texting. I think the payload and content potential of the voice channel in this AI moment is going to really be fundamental in shifting many, many behaviors from text to voice-
Mm-hmm
... and even, UI to AI because of voice, and that's exciting for us. Most of our revenue, most of our business, we've specialized in voice for a very long time. That's exciting.
Great. One question I wanted to ask is, imagine we're here next year. Let's fast-forward 12 months.
It's been a year in a row, so it's not too hard to imagine.
Yes. I'm sitting here. Tell us what do you think we'll be talking about? Everything is about AI right now. You're at the front row. You're seeing this whole communication landscape sort of go through these big changes. What, what do you think happens? Who's doing what? Give us a picture.
I'm This and $5 will buy you a cup of coffee at Starbucks. we've As I look back six months to where we are now, just seeing the pace, I'm very much a true believer in the value and extraordinary the incredible impact of these foundational models. When tuned appropriately, they change lives and they change work. It's an amazing moment. I was registered the first Bandwidth website in 1994, so saw the web. Mobile, we did an MVNO we spun out of Bandwidth and sold to DISH, so saw mobile. This is unlike anything we've seen. If I answer your question directly, one year hence, I think we look back and are amazed, absolutely amazed at how much benefit and redemptive power there is in this technology.
Hmm.
I'm totally excited and jazzed, but I couldn't tell you specifically anything other than I hope we are the network of choice for the next 1 billion users of the PSTN.
That's the right answer. John, you wanna add to that?
Yeah. I think a year from now we will have some very concrete examples of voice AI driving mission-critical properties for name brand enterprises. It will not be abstract anymore.
Mm-hmm.
It will be real, and it'll become clear to everybody how the actual tech stack actually has to look to enable true mission-critical applications, and that's what we're excited about.
Yeah, being in that tech stack.
Okay. Just a couple more questions now on just the financials. You guys have been great steward of capital. You recently bought back some converts that was due in 2028. You are guiding to double-digit growth. You had record year-end profitability. Help us understand, like, what are your priorities from a capital structure perspective? How do you plan on deploying capital, think about growth, and mix of profitability?
Very excited about the repurchase of the converts.
Mm-hmm.
going from having $600 million of debt to now $150, that's it.
Mm-hmm.
Very much in control of our own destiny when you look at our EBITDA and free cash flow and what we're projecting in 2026 and growing EBITDA 30%. Even with the stock buyback, which is meant to mitigate dilution and the repurchase of the debt, we're also investing a record amount in R&D.
Yeah.
-while we're doing that. We've wonderfully achieved the kind of free cash flow generation that affords us the opportunity to both be stewards of capital, be responsible to our equity investors, but also to put more money to work in creative ways than we ever have before. All three things are in play simultaneously, and I'm very grateful to our CFO, Daryl Raiford, who has pioneered the debt repurchase and saved us $80 million in doing so. There should be a bronze in our lobby of Daryl. And it's exciting to have the discipline of operating responsibly at this moment when there's so much opportunity to create new.
Okay. Organic, inorganic, how do you think about M&A going forward?
Organic.
Organic?
Yeah.
Yeah. You've had great results with, like, 1+ million customers. You talked about you're not gonna slow down on R&D. Tell us more, like.
Yeah.
OpEx side.
We did more enterprise $1 million+ deals in 2025 than we did in 2023 and 2024 combined. We're gonna do even more in 2026. The pipeline is larger now than it was in 2025.
Mm-hmm.
That's exciting. The deal cycle, because we've expanded the channel, is shorter than it's ever been. Those are both really favorable for the 26 guide, which is 16% growth top line.
Yep.
Really healthy EBITDA %. Again, hat tip to Daryl for making sure that we have a capital strategy that's robust. John's product roadmap, I think, is terrific for the sales team, and they're excited about it.
Perfect. I love how you have maintained and, you know, cultivated this culture at Bandwidth. I'd love to hear your thoughts on, like, how are you getting your employees, you know, not just giving them comfort, getting them excited in the world of AI when you see Block just let go 40% of their workforce. Talk to us a little bit about that.
We made a declaration during COVID that we would be five days a week in person, and we lost, you know, 20% of the team immediately. We're still five days a week in person, but we made a different declaration right now in this moment of AI, which was, "We're gonna give you every tool you need across all the major models," this was a year ago, "to embrace, use every day, and we're gonna make a commitment to you will not lose your job because of AI. You may be learning something new, doing a new job, but it will not replace you." That was a firm declaration we made in confidence, and I think that it's proved to be really effective in getting adoption and creativity without fear, which is hard to do in a moment like this.
I think that we'll make good on that. I don't see any issue with that, and I'm excited about repetitive, uncreative work going away.
Mm-hmm.
I'm fired up about it, but it does take a leap of faith to be able to make that kind of commitment to your team.
That's amazing. Let me take one minute now, see if there are any questions in the audience. I think we got one here.
I think there's, like, some very good tailwinds with AI for your business, right? Like, 24/7 support centers, like, multiple streams in Maestro. I have a concern around voice minutes. You know, if I call my bank right now, I'm waiting 15, 20 minutes, and you guys are getting paid for it. If I call an agent, it's picking it up, you know, immediately, and it's getting it resolved in two, three minutes. Can the voice minutes are pressured. How do you handle for it? Like, do you think it's a headwind for you?
I don't, because abandonment in the 20-minute queue is acute. There's no abandonment if you can answer every call immediately. You end up net positive in the total number of calls and the call durations. The conversations are incredibly effective, and so the callback rate becomes higher. All of us in this room loathe having to call, and that's changing because these voice agents know who you are, what you talked to them about last time, every detail of your account, and indeed actually interpret if you're hangry based upon the last time they talked to you, or if you might have a cold based upon how nasal you might sound. That kind of intensity in their interpretation from the voice signals you're giving them allows them to have a very wonderful dialogue with you, driving overall engagement from text and chat and app to voice.
I'd also add to that the idea of the long hold times, we've supported a lot of our customers in doing the callbacks, right? You've probably seen it many times where agent won't be available for some minutes, type in your phone number, you do it, hang up, you get a text back later. I think a lot of that has been worked. There's been opportunities for enterprises to already improve that today.
Yeah.
Great. Any other questions? Nope. Great. Well, David, thank you so much.
Thank you.
John, thanks for taking the time.
Yeah.
Thank you all.
Thank you.