These increases are partly offset by a reduction in cash interest, net of the benefits from our $200 million interest rate swap. This is achieved by us having already fully paid down our revolving line of credit, our only expensive debt by the end of January 2024. With regards to capital allocation, in 2023, along with providing sufficiently for our growth initiatives, we deployed $259 million paying down bank debt, $59 million on dividends, $59 million on effective share repurchases to offset dilution from stock-based compensation and $38 million on acquisitions and investments. As of the end of Q4, our net senior debt leverage was 0.5 times and including our 2026 and 2027 convertible notes full year stat our net debt leverage was 3.5 times.
We continued our strong deleveraging trajectory, delevering 1.2 times adjusted EBITDA since the beginning of 2023. With our strong free cash flow generation profile, we expect to organically delever and to increase our balance sheet strength while maintaining our programmatic M&A readiness, our dividend and share repurchases. From a rate exposure perspective, now that we have fully repaid our revolving line of credit substantially all of our debt is protected from high or rising interest rates. To read a very low fixed coupon interest on our convertible notes, or our $200 million interest rate swap expiring in 2030. We remain very comfortable with our capital structure, our leverage, our maturity profile and interest rate exposure. And we’ll continue to evaluate ways to optimize as conditions change.
Moving to our 2024 outlook, which reflects our confidence in continued favorable market conditions and in the momentum of our growth initiatives, it also assumes a slightly larger than normal range of possible ARR outcomes to account for a commercial model shift in China from ARR, lower acquisition expectations, reduced escalations now that peak inflation is behind us and the incremental upside from asset analytics as Greg discussed. Our outlook also reflects our continued margin improvement commitment of approximately 100 basis points. Accordingly we expect total GAAP revenues, based on current exchange rates in the range of $1.350 billion to $1.375 billion with constant currency growth between 10% and 12%. This is reflective of our mix with services revenues expected to grow above half of company average and license revenues growing more slowly.
We are projecting constant currency ARR growth, between 10.5% and 13%. We expect an adjusted operating income with stock-based compensation margin, reflecting approximately 100 basis points annual margin improvement. Any FX impact on our margin is expected to be significantly mitigated by our natural hedge. We expect our effective tax rate to be approximately 20%. We expect cash flow from operations to convert from our adjusted EBITDA, at a rate of approximately 80%. We expect capital expenditures of approximately $22 million. To help you with your models I also included here on this slide additional expectations on interest expense and cash interest cash taxes, stock-based compensation, operating depreciation and amortization, outstanding shares and our dividend, which we have raised by $0.04 for 2024.
And with that, I think we are ready for Q&A. Over to Eric to moderate. Thank you.
A – Eric Boyer: Thanks Werner. In order for everyone to ask a question today we ask that you limit yourselves just one. Our first question is from Matt Hedberg from RBC.
Matt Hedberg: Great. Thanks guys. Thanks for taking my question. So, I guess I had a question from IIJA funding a lot of good content in the prepared remarks. And I think you noted 35% of the $1.2 trillion of funding has been announced. I guess, beyond transportation and now water and electric grid, how should we think about the release of those dollars to additional verticals? And how should we think about just kind of the potential tailwind that could be for you guys over the next several years?
Greg Bentley: Nicholas, could you fill that please?
Nicholas Cumins: Yeah, absolutely. So 35% of the product funding has been announced. And mind you there is a bit of a time gap between when it’s announced and when it’s awarded, which can range quite a lot from one project to the other. In general, it does take time to percolate down from the federal agencies to the state authorities and our own operators to the contractors. Most of it as for transportation as you noted and we explained in the prepared remarks, we see it coming now for water, we see it coming from electric grid. The biggest issue we see for electric grid unfortunately is the permitting process in the US, because it can take many, many years for a product to get started once you have all the approvals, all the permitting done.
So this stage remains the main limiting factor right now in order to have investments in electric grid as an even stronger tailwind. Most of the activities we’ve seen for electric grid, actually all of them are about resilience of the existing transmission lines, analysis of their current state but it’s not yet about expansion. When it comes to expansion this is when you need permitting this will be a significant additional growth opportunity.
Eric Boyer: Thanks Matt.
Matt Hedberg: Thanks. Congrats on another good year.
Eric Boyer: Our next question comes from Kristen Owen from Oppenheimer.
Kristen Owen: Hi, good morning. Thank you for taking the question. I wanted to ask about your AI priorities just given the shift this year in capital allocation. Can you help us pick some parameters around what you feel Bentley needs to itself — versus partner versus maybe acquire and some of the priorities that you’ve outlined. And just as a related follow-up remind us with the success of the cell tower business over the last three years, just what those unit economics look like as an ongoing AI business model? Thank you.
Eric Boyer: Greg, we cannot hear you.
Werner Andre: Yeah, Greg. You’re on mute.
Greg Bentley: I’m sorry. I realize I’m also invisible, which we’re working on here. But Nicholas maybe you can speak to the first part of the question and then I’ll talk about capital allocation, and asset analytics.
Nicholas Cumins: I will be happy to. We see two main use cases for AI in the infrastructure sector. The first use case is actually asset analytics where we use AI to understand the exact physical condition of an infrastructure asset in its context. We can detect any corrosion, any crack, we can detect vegetation and we can trigger engineering workflows based on this. We can also trigger by the way commercial workflows to understand the full utilization of a cell tower and if there is space for more equipment if necessary. These are the — this is the first use case for AI in infrastructure sector and it’s quite advanced. And then, Greg can talk about the kind of financials we see there. But the second use case, which is also quite promising, is the use of AI, when it comes to design and using AI as a copilot for engineers to be able to do more.
I mean that engineering resource capacity gap that we talked about is indeed widening. We’re simply not as a full ecosystem, not creating enough engineers fast enough to cope with the demand. So it’s all about making them more productive. And we think, AI has a tremendous potential to make them more productive. And here our investments are around site and engineering using AI to automate the proposal of site layout to automate doing production. It’s still very early stage. We’re very much at the cutting edge here, but we’re working together with a lot of representative users, a lot of representative accounts. We are fully endorsing the direction given its potential. Now, across both AI for Asset Analytics and AI for Design, there is a true platform opportunity.
At the most fundamental level, it’s the digital twin platform. When we do AI analytics, we actually create a digital twin of an infrastructure asset. When we power infrastructure engineers to do better design through [indiscernible] capabilities, we also leverage digital twin technology. So across both, we have the iTwin platform. And then, when it comes to Asset Analytics specifically through our investments with iTwin Ventures, our corporate venture arm, we have looked at so many companies in that space and we realize, there is definitely a lot of redundancy. They all build the same capabilities over and over again. So we see a clear platform opportunity for us on top of Digital Twin technology in order to do Asset Analytics.
Greg Bentley: Well, I’ll only add that Kristen and I realize it’s been three years since I talked about the cell tower opportunity as a forerunner, for what should become digital twins in every infrastructure asset type. And the AI has gotten better and better and the value has gotten more and more over that period of time, but the business has had the fits and starts of the tower cos wanting digital twins, but their own ecosystem of providers in their own internal enterprise systems not being ready. So that has come and gone to where what’s going on now however is it’s in prime time with the household main providers of broadband infrastructure on board with this potential as they remake 5G networks. And our role is behind the scenes provider of the AI and the AI processing and reality modeling and we’re generating three digits per tower and participating in procurements that will help institutionalize finding.
So you will have taken – rest of this year on top for three years, but he’s forerunners to become – to establish the potential of Asset Analytics on very much the believer that we’re at this starting point now.
Eric Boyer: Thanks, Kristen. Our next question comes from Clarke Jeffries from Piper Sandler.
Clarke Jeffries: Hello. Thank you for taking the question. Greg I’m reflecting on your comments about no longer being able to rely on user volume growth and maybe an expectation for longer-term application accretion expansion. Are we at the point where you expect that to be a pretty linear cadence each year? Or is this a comment about the longer term that you’d expect continued expansion? I’m trying to File Thank you for taking the question. Greg, I’m reflecting on your comments about no longer being able to rely on user volume growth and maybe an expectation for longer-term application accretion expansion. Are we at the point, where you expect that to be a pretty linear cadence each year? Or is this a comment about the longer-term that you’d expect continued expansion?
I’m trying to understand, if we’re at the point where user growth is not growing net per year or if we’re trying to increase the monetization up level of the individual employee with the application mix as soon as 2024?
Greg Bentley: Well, first I might say that our observation about user volume is particular, where we see it most in the E365 universe, where we literally charge for application day. And it’s those accounts, the large ones that have the latest data. We saw one out of every 10 positions is open and they face retirements faster than they can add new colleagues. So they’re literally emphasizing workflows that save user days. And we’re delivering those through blueprints and more specialized applications. I think that can inflect even faster because the – these firms are the same ENR, top firms who spend 1% of what they build on software and realize it will be smarter for them to have more productive hours and that’s the only way for them to grow their business and meet their backlog.
So I don’t think it has to be only a linear progression from what is spent now per infrastructure engineer to what could be spent and is already spent by other types of engineers on software. We’ve been talking about that comprising our TAM if you like for a long time. And it’s – the pressures are causing it to accelerate now I think.
Eric Boyer: Thanks, Clarke. Our next question comes from Jason Celino from KeyBanc.
Jason Celino: There we go. Good morning.
Greg Bentley: Hey, Jason.
Jason Celino: Maybe one for Werner, does 10.5% to 13% ARR growth guidance for the year, I know you mentioned it a wider than normal range but any way to unpack like what that headwind is in China, the lower maybe M&A contribution and then the third piece that was mentioned in press release?
Werner Andre: Yes. Well, on acquisitions we had 100 basis points in the outlook in 2023. And we just expect that this is coming down from that level. Actually we have in 2023, the contributions were less than 100. We were at 70 basis points in 2023 and 2022. And so we’re taking expectations down also underline what Greg mentioned. Between China lower escalations. We don’t really want to break it out clicking down further. That’s a right range of potential outcomes. And satellite momentum for our key growth drivers is very consistent, micro trends are pretty consistent. And there’s incremental upside from the asset analytics business that we talked about. So it’s a wider range of outcomes.
Jason Celino: Okay. Great. That was helpful. Thank you.
Eric Boyer: Thanks, Jason. Next question comes from Michael Funk from Bank of America.
Michael Funk: Hey, good morning. How are you doing? So Greg, thank you again for all the color on digital twins really helpful. And good to hear the updated commentary on the tower business. As we’re thinking about how that business scales, it’d be helpful though to get maybe some more metrics there. So Greg, there might be approximate about 160,000 towers in the US, for example. Can you give us a sense of how many towers you’re actually servicing today and how that’s ramped the last couple of years? And a related question, are the tower construction companies like MasTec are they also a gating factor is the telecos might have contracts with them for maintenance? So something about how quickly that business can ramp based on those factors?
Greg Bentley: Well, I think it’s mainly a business for the existing towers. And our subscriptions have gone up and down over time, based on the preparedness of the ecosystem to manage the Digital Twins. The business it starts the year in seven figures of ARR. I’m very confident, we’ll end the year in eight figures of ARR, but there are very sensitive procurements going on at the moment that preclude me saying, more than that. I believe there is equal opportunity in Blyncsy, which is a — which is proposing the crowd source based AI for maintenance condition detection, which can run every day or multiple times per day, to multiple of our DOTs and we can make that — those AI insights even more valuable by relating them to the ET and the IT regarding maintenance.
It’s very exciting to put them together. We will have — Michael, I think a quarter from now, a full announcement with a new name and so forth for that business. Right now, we’re busy winning it we’ll get to communicating more about it over the course of 2024.
Q – Michael Funk: That’s great. Just for clarification, Greg. From the tower business, you’re charging per asset and then for the Blyncsy’s business you’re charging per mile. Can you give us a sense of the economics there? I mean what you charge per tower monthly, quarterly, annually and the same for the Blyncsy’s business?
Greg Bentley: For tower it’s three digits per year, per tower and we have standard pricing. For roadway mile, with Blyncsy it also is a standard pricing, but it depends on which AI insights you want and how frequently you want them. But we’re trying to make all of this standard price book. We want to have many partners selling for us the engineering firms to be offering this to their owner-operator, accounts where they can add their proprietary analytics on top of ours and where they won’t have to worry about doing the back office processing, because we’re going to be very proficient in doing that at scale and economically and at high quality.
Q – Michael Funk: Thank you. That’s all very helpful.
Eric Boyer: Thanks, Michael. The next question comes from Jay Vleeschhouwer from Griffin Securities.
Q – Jay Vleeschhouwer: Hi, good morning. Thank you. Greg, in your comments about the 2024 developments, you noted the asset analytics. So the question there is — would you put that in the context of the industry solutions concept you’ve spoken of previously for example at YII? And if so, could you speak more broadly about some other critical product deliverables that you foresee for 2024? And is there any reason to believe that customers might be more inclined to adopt new technology these days, more quickly than they might have in the past?
Greg Bentley: Well, I think everyone is conditioned to say what can AI do for me now? And that’s the difference with asset analytics is we want an instant on entry point so that you don’t have to say, well, you start with the whole concept of digital twins and that’s from the beginning of the life cycle and so forth. If you take the example of Blyncsy with crowd-sourced imagery, you can have it the next day. There’s not even any special surveying that needs to be done for that. I would say, the principal difference from industry solutions is here with asset analytics. We really have an ecosystem in mind. We’re trying to beat the platform to be ecosystem friendly, so that the ecosystem can get into this AI opportunity without having to start from scratch, with bringing together the ET and the IT and the OT.
