Autodesk, Inc. (ADSK) September 9, 2026 Conference Transcript

Hey, everyone. I'm Tyler Radke, Citi's co-head of U.S. Software. Thank you for joining day two of Citi's global TMT conference. We're happy to have Autodesk back, who's a regular attender of the conference and excited to have a great discussion with a practitioner, I would say, at Autodesk. We have Bob Bray, the VP and GM of Autodesk Tandem. Bob, thanks for making it out to New York City. I know you've got a big conference in Las Vegas next week you're probably busy prepping for. But for the benefit of the folks in the room, give us a sense of your background. What are the things that you do day-to-day at Autodesk just to set the stage?

Yeah, I mean, I live and breathe in the trenches. I build software. I run teams that build software. I've been doing it for Autodesk for almost 30 years, all through our architecture, engineering, construction business for a lot of that time, was part of the foundation of our construction technology business as we did some foundational acquisitions there. And now I am kind of leading the way into the operations business for Autodesk, so that next pillar of growth for us. So excited to be here.

Yeah. And what is Autodesk Tandem, just for folks that may be less familiar with it?

You can think of Tandem as an operational intelligence platform. Audit S for years has helped customers design and build things. Tandem is more about how do we help customers operate those things effectively, efficiently, reduce downtime, increase operating efficiency, whether that's throughput of a factory, whether that's helping people get through TSA lines more efficiently, whatever that may be. It's really about driving efficiency in operations. And if you think about the flywheel that creates, that knowledge from operations then feeds back into helping people design and build things better in the future. So it really connects the value chain for Autodesk from end to end and for our customers.

Just to add that, the operations market term is about $40 billion. That's about four times the size of the construction term. But as Bob said, there's a couple of interesting attributes about it. One is it's a decades-long business, whereas our construction business and design business is a three- to five-year business. It's a massively longer-duration business. And the second thing is it opens up a whole massively bigger data set for us because it's not just the new build where you get the data. It's the existing built environment of which there's vastly more than the annual new build as well. So, for example, a very large global retailer in Q2 signed an EBA with us to digitize their entire estate. And it's not just their new build. They're digitizing their entire existing estate. So that's pretty exciting using Tandem.

And when you say it's a, I think you said 10 or 20 year opportunity versus a three to five year, what do you mean? It's actually a 50, 60, 70 year opportunity.

So think about, so you design a building, you're going through the architecture engineering process for maybe three to five years. You're building that building, you're going through five years of process. Typically, that's 10 years of subscription for us.

Sure, sure.

Whereas if you're maintaining that building, you're maintaining that building for 50, 60, 70 years. That's a long lifetime of subscription.

Yeah, sort of thinking about the duration of the project and the use case. That makes sense. You know, within sort of the building and construction design side, Autodesk has long talked about the BIM standard, I think probably for the past decade, maybe not 30 years. But how is that, you know, where are we in the adoption of BIM and how is that driving growth for Tandem and some of the operations products that we sell?

The adoption of BIM is really about data interoperability and data flow from early stage planning, through design, through engineering, into construction, and then into operations. So keeping that data consistent and having a smooth set of data all the way from end to end. So a concrete example of this in the real world, a large contractor in Europe, BAM, they are now digitizing every project from beginning to end with the end deliverable being a tandem digital twin for use and operations. And they also do, they have extended their business in the last decade into facility management services. So they'll sign a 25-year contract to actually maintain a building after they've built it. And they use Tandem all the way through that process from beginning into operations.

And they've already called us about having a conversation about maintenance as well.

No, that's super exciting. As you just kind of look at the lay of the land within the ACOS, AECO segment, like how are you seeing the puts and takes obviously data center demand construction projects are off the charts but you know uh there's certain areas certainly offset areas so how would you kind of blend it together i mean you've been at autodesk 30 years seen many many cycles where are we kind of in um in those ranges there's always an ebb and flow um obviously commercial construction starts are not exactly high right now but other things are high right now so there's always an ebb and flow in the construction industry uh if commercial off

buildings aren't being built data centers are or pharmaceutical facilities or other types of facilities so there's no shortage of demand for building thing yeah that's for sure yeah yeah got it and this concept of a digital twin we've heard a number of design and simulation companies talk about that in the industry like what differentiates your approach and your your digital twin product versus some of the others we might care of our approach is life cycle based the whole asset life cycle from early stage planning all the way through design simulation of how that will perform building that and tracking the twin through construction and using you know simulation of scheduling and all kinds of things through the construction life cycle and then into operations and then being able to compare is this factory producing at the rate that i simulated that it would if not why not and then being able to optimize based on that so it is really that whole life cycle approach that differentiates autodesk and our approach to this process right right got it um you touched on maintain x and obviously it's a you know big acquisition probably very exciting given where you sit within the organization.

But how do you, I guess you touched on some customer anecdotes, but what's sort of been the customer response and how are you sort of envisioning where that product goes versus where Tandem goes going forward?

Great question. I mean, we're super optimistic about this acquisition. Customers have started engaging with us significantly around this acquisition. And why is because if you think about what we built with tandem we've built the system of intelligence so we can understand when a product might fail or a component might fail in the building now we have an application that can we can basically assign a technician to go and fix that problem and also provide them context to what the root cause might be or what the solution might be so it's again connecting those systems together intelligently so we go from understanding there might be a problem what is the problem and now assigning someone to go take action to fix the problem. So that connection through the lifecycle is really important. As Simon said, we've had a global contractor who's been using Tandem for a few years now call us recently because now they see that opportunity to connect those systems end to end. That retailer also very interested in their refrigeration units is where they started with Tandem, understanding when failure might happen because highly expensive to throw food away. Now they can assign a technician to go and fix that problem.

And so there's sort of two threads of opportunity. One is sort of the owner thread, and owners want a complete view of their assets end-to-end, both of their existing estate, and then that informs also their new build. And then you've got the sort of service companies that are servicing them to help them build it who are doing the sort of same thing. And we are serving both of those constituencies sort of end-to-end. And that then opens up the opportunity, which is, you know, in AEC, and as we've told you, there's already, you know, so Maintenance has primarily been in factories and manufacturing, but any asset, whether it be a factory, is a building with stuff in it, as is an office block, as is a piece of infrastructure. They're all, you know, stuff with stuff in it, buildings with stuff in it, basically. The other opportunity is in enterprise. You know, Maintenance has primarily been in small and medium-sized business. And then the sort of third opportunity is in sort of internet. And so all three of those are significant opportunities for us to take MaintenanceX to help them expand. And then the sort of longer-term one, and then as we've been talking about, is the data, both in operations as a sort of standalone, but then also taking operations data upstream.

Right, right. And as we think about just the MaintenanceX business and what type of customers, what industries they're the strongest, and maybe this is also in conjunction with Tandem, But where do you have the most success doing operations? I imagine, you know, doing operations for retail is very different than a data center or a complex, you know, manufacturing plant. Like, give us a sense of where you sit in kind of those.

The highest impact and the customers that are adopting are those that have critical assets, where if the asset fails for any reason, they're losing revenue. typically so factory industrial um health care uh airports yep um data center um and even into retail um you know one of the funny things about maintain x that i learned early on is one of their earliest customers and their largest early adopter was actually car washes now think about that if car wash fails they're not making money yeah it makes perfect sense so that notion of the asset being critical at its time being critical is where we see adoption.

So talk to us a little bit about the data side of that, right? I mean, because these are really complex environments, the mix of operational data, you know, IoT data. And we've seen approaches from here at PTC, for instance, with some of the acquisitions and then subsequent divestitures they've made in terms of IoT. But where are you getting all this data from and how do you kind of ensure that you're able to have that full view of, you know, really critical and complicated environment?

We made a very small acquisition a couple of years ago, small enough that didn't make radar, of what it was an integration platform. And it was a hybrid integration platform that allows us to both connect to enterprise systems in a bidirectional way. So an IBM Maximo or an SAP, systems like that, but also allows us to connect to technology that runs in the OT network within the building. So that could be a building management system that's controlling temperature in this room. It could be an MQTT system or a SCADA system that's controlling the equipment in a factory or an industrial facility. And so we have that ability to connect to various systems and ingest that data from just about any system today. And that is a tool that is low-code, very easy to deploy, and we're starting to bring AI into that to basically create the citizen integrator, if you will, to make that easier and easier.

And the critical difference between what we're doing and what PTC is doing is that we are taking an owner view, which is of the entire asset. As you see with Tandem, also Maintenance is the same thing. Whereas PTC, things like ServiceMax is much more of a vendor view. So it's only a subset of the assets that are in the building, which is less useful to the owner.

Sure, sure. Okay, that's helpful. And, you know, I guess as you think about the AI opportunity, specifically within MaintenX and Tandem, I mean, what are some of the key features that you've released thus far? What are some of the things that you're working on? I'm sure we'll hear a lot next week at AU, but talk to us just how you see it.

You'll hear a lot of things at AU. But in terms of tandem and operational intelligence, we think about AI in three areas. One, helping customers take the massive data that they have, which is really what they They have a lot of fragmented silos of data and turning it into structured information. AI is exceptionally good at that. Helping them onboard and create their digital twin, one thing. Second thing, you hear a lot of customers that complain about dashboard fatigue. I don't want to pay somebody to look at a dashboard. where do I want the system to understand where the problem is and tell me where that problem is? So again, using AI to infer from data, produce the right context, and then inform the customer what the challenge is. And then long-term, it's taking action. So using agentic AI to actually take action within a customer's facilities. Now, this is, there's trust issues there. Customers need to trust that before they're going to have them control critical facilities in any kind of building. But customers are asking. yeah and there's those barriers are starting to come down yeah in maintain x it's you know understanding a maintenance schedule is one thing or like a manual will tell you this piece of equipment needs to be maintained every month but if you use utilization you can create a predictive maintenance schedule that's much more optimized so using ai to infer from data and then create optimize schedules provide root cause analysis for the technician basically give the technician at their fingertips what they need to know to fix the problem yeah solves a lot of problems and it it basically improves efficiency the technician right right and it sounds like this

maintenance and uh maybe tandem to a lesser extent like the the primary customer is is the target customers the owners right is it um how would you if we look at the autodesk business today, right? Clearly dominant still in design, which maybe is less on the owner's side. How would you frame how big the owner's revenue or install base is today? Is this, and the reason I ask that is understanding, is this more of a new logo or a cross-sell opportunity?

I won't rant on the term owner, but there's the capital project teams. We have deported capital project teams that honored us for many, many years. So that's a cross-sell opportunity into their operations business. At the same time, there's new logo opportunity in supporting operations businesses across the spectrum. The retailer that Simon mentioned earlier used to be an EBA years ago, walked away. They're now an EBA again. And it's really because we introduced Tandem and an opportunity to digitize their operations business, and now they're working back upstream into our design because they want to digitize their entire portfolio. So it's both new logo and expansion for us.

Okay, okay, got it. I guess as we sort of look at the AEC business broadly and maybe manufacturing as well, but sort of separating out maintain X and tandem, where are you seeing kind of the greatest interest just as customers are looking to deploy AI for Autodesk? Clearly, there's, you know, a lot going on in the industry with some of the new, you know, model releases, Astra, et cetera. But at the same time, you know, you have massive backlogs of projects with a lot of manual processes that still haven't been automated by AI. So where do you kind of see the, you know, the opportunities out there?

For AEC firms and AI, it's about automating the mundane. They do a lot of things repetitively and getting rid of the mundane constraints. We talked about that today multiple times, just having a constraint solver that will automate constraints is a huge time-saver for it. Just automating drawing production, automating the design of things that have a consistent shape to them. I hear architects complain about designing bathrooms all the time.

How many times do I need to design a bathroom? right there's certain things that can be automated in a repetitive way and that's where ai is an early adoption within within all a forms because it takes away that kind of repetitive work and allows the architecture to do the things to do well let me tug on that thread a little bit as you mentioned the manager word but the so this is going to be a longer answer but a few things so firstly our customers really care that buildings don't fall down and wheels don't fall off cars etc so you can't just say yeah let's give it a go and and try something it has to work so So that means to build something that works takes time, and they're not just going to throw stuff out because they've got a whole bunch of processes to make sure the buildings don't fall down and wheels don't fall off. So they're not just going to rip it out, put it in and stuff. So it's going to be a workflow-by-workflow adoption to make sure that they have accuracy, which then ensures safety and a bunch of other stuff. So that's the sort of first thing. The second thing in terms of adoption is that there's sort of two bits of value that you can extract from AI. One is around money. but the other is around data. And those often can be conflicting objectives because if you charge more money, then people use it less or you have less distribution, which means you get less data. So what we're trying to do is to sort of maximize both of those, which is, again, using the sort of task workflow systems automation that we use at Investor Day, task being a narrow scope of work, workflow being a broader scope of work and the systems being the broadest scope of work. So primarily focusing on task automations, which automate crap, which people don't want to do, but the other attribute of a task automation is a small model, which means the compute can happen on the edge rather than in the cloud, which means there isn't a lot of marginal cost attached to it. That means that we can deliver it as extra value to our existing products and capture value through price as we would with a typical subscription. So that's how we capture value today, but what it also does is it makes the model available to all of our customers, and they use it more. So why is that important? Using an example from auto-constrained and fusion, at the beginning of last year, when we launched it, it had an acceptance rate, which is demonstrated value of about 30%. By the end of the year, because we were able to iterate with usage and data, the acceptance rate was north of two-thirds. A massive improvement in the value that we were delivering as demonstrated by the usage of it. So that's the sort of second thing. And then the sort of third thing, because you mentioned Astra, and to sort of talk about it. Again, so we talked a couple of weeks ago, as you know, on our earnings call, that there is going to be no single model that does everything because no single model will have access to the right data and the right context. So the way to think about it, particularly in our world, which is very complex and has high stakes of things that go wrong, is that a decision a customer has taken in any given point will reflect the combination of needs around accuracy, speed, and cost. And those three things are often competing constraints. So something that is very accurate is often slow and expensive to generate. And so the choices you make between those three things will depend on the context in which you're taking that decision. And that context will then be contingent on a bunch of other things that are dependent upstream and downstream on then what happens once that decision is made. And so that context and the decision-making will be determined by the harness layer. And the harness, which is for us Autodesk Assistant, then pull on the right resources, which will be a combination of our own AI, neural CAD. It'll be frontier models. It'll be open models. It'll be our customers' own intelligence. And also it'll build on a bunch of functionality to enable accuracy, speed, and affordability to take the right decision within the flow of work. So where Astra fits into this is that we talked to Investor Day, as I said, about tasks, workflow, and systems automation. And as you move along that spectrum, the lead on the work shifts from being a human being to being a machine. As we said, Investor Day was exciting because machines generate a significant new value in two ways. One is because they can make inference that human beings cannot because the human brain cannot conceive of it. And secondly, they generate new monetizable time because they work 24 hours a day, 365 days a year. But you can only do that if the machines can intelligently take action. And so that is why Astra is important because it enables that shift of to generate that new value using our tools and draw significantly new value and more value through our ecosystem that we're pretty excited by. Right.

And sitting on that theme, it might just be like slightly different definitions here, but one of the themes we were discussing yesterday with some of the private companies, heard this from some publics, is this idea of agentic labor, which I think, you know, in your industry, you still need an agent can't build a building, right? Certainly they can help with design, the automation of tasks, but ultimately there is sort of like this human bottleneck. Is the contractor going to show up? Are they going to have the right materials, et cetera? So how do you think about the sort of agentic labor opportunity for Autodesk? Maybe it's more just agentic workflow or agentic system automation that you talked about. And, like, how much, as you think about your customers, like, how much can that really accelerate their business, given that there are, you know, physical and true human constraints, even if you did everything perfectly on the design and coordination?

So this is where an economics degree is going to come be valuable again, you know, make economics valuable again. So, you know, so this is sort of lump of labor fallacy, Jevons paradox, for those who want to sort of look it up, which is that AI is not about doing what you've always done with fewer people. It's about using the people you have to do a lot more and a lot more defined as more volume of work. So solving capacity constraints, which are endemic in our industries. And secondly, do stuff which is previously inconceivably hard to do. So if you're not trying to do more and harder stuff and are only trying to do what you did before, you're going to die because we're going to take all of your business if you're one of our competitors. So that's the sort of first thing. This is about doing more and harder stuff. It's not about doing the same as you've always done with fewer people. But the second thing, and this is important, is the way that humans interact with machines is going to change quite significantly. from a sort of human-led with a sort of computer enabling the human to the computer doing a bunch of, actually, a lot of the heavy lifting with the human defining the goals and validating on the sort of back end. So the way that humans interact with machines is going to change quite significantly. So you're already seeing this in R&D and in the software development role with things like spectrum development and AI factory processes. That thinking is going to propagate across the entire economy. So we're very optimistic that there's going to be a lot more done, a lot harder stuff. We're going to be able to solve a bunch of problems, not only things like climate change, which are big problems that need solving. There aren't enough resources to solve that problem. AI will enable that sort of stuff. But it is going to require people to work differently with machines.

You'll see people engaging with products differently. Instead of engaging with products, they'll engage with a conversational agent that will take them into their product workload. And that's going to change.

And you can see that already in how we're thinking. If you look at what Forma is doing, that is already, if you know what specter-driven development is, that is already a sort of specter-driven, whereas the human is saying, okay, what do we want to try and achieve? And then Forma is doing a lot of the compute behind that.

Right, right. And you touched on Forma, and so I wanted to hit on that sort of data modernization, and I'm sure we'll hear a lot of updates next week. But where are your customers at just in terms of, you know, modernizing, you know, the traditional kind of file-based models and migrating that data to the cloud?

I mean, let's just say it's a normal distribution curve. Some are sort of right at the cutting edge. Some have barely begun. And then everyone else is sort of strung out in between.

I think the interesting thing about Autodesk is we had started that process pre-COVID. And COVID actually caused customers to move faster to the cloud. And so we saw a significant adoption. They didn't go backwards. They moved to the cloud and they stayed in the cloud. And that allows flexibility in their workforce, flexibility in where their people work, whether it's home, office, remote offices, engaging workforces in different parts of the world where labor is cheaper. I think that has helped our customers a lot. And so we were early in that process. And so customers are not going backward. They're going forward.

And I think one of the underappreciated things about Autodesk is that the work that we've been doing over the last 10 years has set us up really well for the coming decades. And a lot of our competitors have not done that hard work. There's a bunch of stuff around business model transitions, around our platform, around the investments we've been making in AI for a decade, et cetera, that set us up really well for the future. I know they've been quite painful, some of them, you know, integrating more directly with our customers, another example. But all of that stuff and that hard work that we've done over the last decade have set us up really well.

Yeah, yeah, got it. And as you think about just the broader product portfolio strategy and M&A, I think, you know, certainly you've overseen quite a lot of acquisitions in your tenure at Autodesk. And, you know, Construction Cloud started off with the marquee, you know, acquisition of PlanGrid and then you added Building Connected and Assemble, all those things. How do you kind of see the additional consolidation opportunity out there in terms of what categories you want to bring in to really have a dominant platform across the broader maintained operations market?

Yeah, I mean, I think the operations playbook is following the construction playbook, right? So MaintainX was that cornerstone acquisition for us. It was the plan grid. Plan grid was a number two. MaintainX is number one on the market. So we bought the number one player first and foremost. And then what you'll see us do is continue to do back-talking acquisitions to expand that portfolio. If you think about engineering and what we've done in engineering over the years, it's all about having data and having context. Operations is no different. So we want to build rich context in operations. That means having data in different workflows that really enrich our asset intelligence that help us basically build that lifecycle end-to-end from planning to operations and close that loop.

We've been able to walk and chew gum, as you see, which is we've been able to make acquisitions, but also continues our capital return program, spending roughly 50% of our free cash flow returning to reduce share count over time.

And Simon, I think you mentioned earlier the operations ham it's 40 40 billion or so so clearly massive opportunity um but i guess you know maintain x is hundreds of of millions right so how do you is that is that all sort of that tam is it more than a green field in terms of automating uh labor manual systems how do we kind of capture that um where does that where do all those dollars sit where maintain x started was more lower into the market so basically digitizing paper processes yeah amazing number of customers today still operate and do maintenance on pen and paper um i've seen major airports in

the u.s that still do maintenance programs on paper that's a little frightening shocking that I might scare you just a little bit when you take your next slide. But obviously, what Autodesk can bring to MaintainX is our enterprise business and our ability to tap into the enterprise. So there is tons of growth potential for us in MaintainX, both in terms of bringing them upmarket into the enterprise, but also globally into the market.

But it's not just the operations business. It will also help our art design business and make business as well.

To pull through. Right. Got it. Maybe in the last few minutes, I know you don't want to front-run a bunch of announcements next week, but what are some themes investors should be looking out for at AU? What are some things you're excited about, Bob?

There's just a ton of announcements around AI that I think you're going to see that are really exciting for us, pushing boundaries, helping customers do more with less, you know, do more things and automate more tasks. That's been a big part of our focus and continues to be a big part of our focus. changing the way customers work and how they approach their projects, focusing on outcomes. Another big thing that you'll see, the whole operations announcement. Obviously, where I'm most excited because that's the world I live in.

And again, what we said at earnings, we'll also put more flesh on the bone of what we said on earnings as well.

Yeah, a few things I can't say because we're holding those announcements.

Yeah, got to keep some surprises for the keynote. And maybe, Bob, just as we wrap up, let's talk about goals. Like, what are your top three priorities over the next year? And then with this MaintainX and essentially other businesses, like, how do you think about the vision where you want to see the broader MaintainX operations business a few years down the road?

Yeah, I think for us, it's very much about building context around operations. So moving from system of intelligence, which we've been building organically with Tandem, into a system of action with MaintainX, taking all of that information and feeding it back upstream into our design and construction business so that people can basically design and build better in the future. And if we can close that loop over the next five years, I think that's a huge flywheel for us, expanding that operations business long term, but also just continuing to build our core businesses in architecture, engineering, construction, manufacturing. Yeah.

Great. Cool. All right. Thank you, everyone, for attending the session. Bob, Simon, really appreciate you coming out for the conference. Thank you.