Autodesk, Inc. (ADSK) Earnings Call Transcript & Summary

Okay. Good afternoon, everyone. Thank you for joining us for our last company fireside chat of the design software conference this year. I am excited to host the Autodesk team today. And we're going to have Simon, who heads up their Investor Relations side of things, first, introducing the safe harbor statement as well. Simon, do you want to go ahead before I introduce Stephen as well?

Sure. This is going to be the least exciting bit of the presentation. So we may make forward-looking statements during the course of this presentation. Please refer to our SEC filings for information on risks and other factors that may cause our actual results to differ materially from these statements. Back to you, Gal.

Awesome. Thank you. And with that, going to the exciting part, as Simon called it, we got Steve Hooper, who is the VP of Design and Manufacturing at Autodesk. Stephen and I go a while back now, having had a few chats over the years. And Steve, I'm so glad we got to host you today and get a little bit of an update on the manufacturing side of the business.

Yes, likewise. Great to be here. Thanks for having us.

Thank you. Steve, considering you're kind of leading the Autodesk Fusion 360 and the manufacturing efforts, interestingly, when I -- when investors contact me around Autodesk and want to talk Autodesk, a lot of times, they focus on a well-publicized Construction side and the other parts, like [ Vector ] side of the business. But over the years, Autodesk's been very, very strong in kind of the maker community, going both from the design all the way towards CAM and then manufacturing. It feels like, at least within the company, even when we talk to Andrew, for example, say, we had to invest a lot of mindshare and a lot of capital from a mind perspective and also financial capital into getting the construction in place where we're happy with. But at the same time, we're kind of refocusing some of that incremental effort into manufacturing. And that's been happening with -- arguably with Fusion 360 for almost a decade now. Would you please help us understand how you at Autodesk, when you look at the manufacturing from your perspective, how do you see the opportunity and the market potential for Autodesk to kind of come in and potentially kind of consolidate some of these very dispersed things that are floating around that are 20, 30 years old?

Yes. Yes, exactly. I think the first thing I would say is that, that area of the market has been fairly underwhelming in terms of disruptive technology. Like you said, for the last 30 years, most of the kind of traditional categories of software that we've known have remained substantially, I think, the same. So if you think about computer-aided design, computer-aided manufacturing, simulation, not much has changed in those categories. Our belief is that we're now at a phase where the market really needs to see a lot of those technologies converge together in order to drive better automation across the whole of the manufacturing process. And that includes going down onto the shop floor. So our perspective is that providing point solutions that are effectively tools to solve 1 piece of the product development process isn't going to yield much of a benefit for customers. If you actually think about the type of benefit you can deliver in terms of productivity, even if your software is best-in-class, you may be able to displace a competitor with maybe 4% or 5% productivity improvement. By converging these tools together, we can start to introduce automations. And those automations, if you think about downstream deliverables that people need to actually manufacture the products they design, some things like drawings or machining strategies can take days to produce. So if we can automate those processes, then we can yield benefits of maybe 200%, 300% productivity improvement. And we think that's what gives us the opportunity to really disrupt that segment of the market. In terms of market size, there's roughly $38 billion, we believe, by FY '26 in terms of market term with 31 million users. I think the interesting thing for us, though, is that in the past, the design side of this business has traditionally been larger. And I think at this point, we're actually starting to see an inflection where the manufacturing processes are becoming larger due to digitization. So we actually estimate that $20 million (sic) [ $20 billion ] of that $38 billion is actually attributable to manufacturing. And there's a similar opportunity there to what you mentioned in construction, which is digitization, especially in the world's current environment where so many people are having to work remotely or reconfigure their supply chains. You can't really rely on some of the manual processes that have been taking place to support that in the past. So digitization really leads the charge in terms of holding the key to the opportunity there.

That's interesting. And I think it's a great introduction to kind of set the tone, especially because of the fact, like you said, that technologies after CAD, CAD is in large -- to a large extent, fairly commoditized, right? But when you go down and you take the cut, it's not what gets to get the machine. And we know about 3D printing and sure, 3D printing is kind of a press the button sometimes with errors and stuff. But when you go to CNC machining and those kind of problems, there's a lot of menu manipulation and coding and stuff that goes on in between those processes.

Yes. Yes, absolutely. So that extends to other manufacturing processes as well. Obviously, you can't do injection molding, which the majority of the world's plastic products are actually manufactured from injection molding, and you need subtractive machining to produce those model and tool [ dies ]. And there's a tremendous amount of time and effort spent on programming that. So if you think about most of the manufacturing processes out there today, most of them are subtractive or fabricated in some form or fashion. So whether it's laser cutting, plasma cutting in 2 axes, some more -- the sort of more sophisticated side of subtractive machining, you see multi-access machines up to 5-axes simultaneous machining. Our belief is that we can actually automate a lot of that process. So -- especially with the advent in machine learning, there are opportunities to take what is a fairly formulaic process when you look at speeds and feeds and machine tool wear and the type of geometry that you're trying to machine, the type of material that you're using. Often, manufacturers are producing parts that are very similar to 1 another. They're not the same but they're similar. And that opens up opportunities to employ things like machine learning to automate that process. So our belief is if you combine that with moving the computation to the cloud, you have the opportunity basically to take a 3D model in its definition of the design and automate the production of the machining instructions that actually drive machine tool on the shop floor. So we've been investing a lot in that space, not just in the actual programming of the machine tool. We've actually acquired a business that does the G-Code simulation to make sure it runs effectively on that machine tool. And we just completed an acquisition of a company that produces manufacturing execution instructions for the shop floor as well. So you combine these assets together and you have the opportunity to automate the production of the machining strategy, test the posted G-Code, which is actually what's run on the machine on the shop floor, transmit it to the shop floor in a digital format and actually see the execution through on the shop floor and manage it with a digital execution tool, which can run on a mobile device. And again, very similar to what we've seen in the construction industry, which has been very transformative in that space and opens up a huge opportunity for us as a business.

I think analogy, and I used once when we had a chat in the past, was in a similar way to having drivers sort of printer and all that stuff that we used to deal with when you didn't know if you could print or not, whereas now you'd kind of log in and you expect to press a button and it will come out, the text or the picture, whatever we're trying to do. The problem is way more complex in manufacturing. But that's the vision, isn't it? The vision is to design something and knowing that it's fit for manufacturing, whatever the process might be.

I've always seen it as analogous to desktop publishing. So you saw the revolution that took place in desktop publishing. Largely, that was due to a combination of factors. The first one was most of the tools that we used either for graphic illustration or typesetting, font production, editing and layout all got combined together in a desktop publishing utility where what you saw on your screen is what you got when you actually went through the print process. Then in the physical manufacturing, although only 2 degrees of freedom for a printer, is very similar to a machine tool. So we went to drivers. So if you think about when you connect a printer to your laptop or your PC, you don't worry about the print instructions that you send to the printer anymore. The print driver just takes care of that for you so that when you hit Print, what you saw on your screen is what comes out of the printer in your home office. So we see exactly the same transformation happening in manufacturing. That's why we've partnered with organizations like Haas machine tools, where we now actually have a printer that behaves in exactly the same way as it does for your desktop machine at home with your 2D printer. You literally wire it up to your Haas machine tool. It interrogates the machine tool, understands how it operates and then can produce the code necessary to drive that machine tool automatically for you. So as you say, it's complex. It's not as simple as a 2D printer, but we're making significant progress and that progress has accelerated over the last 2 years, especially with some of the partnerships we forged with machine tool manufacturers, like Haas or Mazak and many others. So we see that as a strong opportunity for growth.

Really interesting. Those are some of the most complex problems that you're trying to solve, but the platform today and in Fusion, I always see it more than a product, right? It's becoming like a product that is in itself a platform with all the expansions you generated and everything. And 1 part of the -- 1 big part of the differentiation against the traditional CAD tools would be the fact that it facilitates the workflow management and doesn't require you checking in, checking out of the files and stuff because it does support the full workflow. Is that the reason why coming out of pandemic and through pandemic, that's one of the businesses that's really accelerated, and it's almost like giving a little push to the market, right? Is it a workflow? What has been, over the last couple of years, that has allowed you to significantly accelerate that commercial growth of number of customers? I think you're about 185,000 right now, right, as you report them in the last quarter, which is getting a decent scale of the business. And what has been the driver? Has it been mostly the connectivity and the ability to work remotely together on 1 different model? Or is it those advanced things that you introduced into the process?

I think it's a combination of factors, some of them technological and some of them from a go-to-market perspective. So from a technology perspective, we've reached a significant milestone in terms of maturity. So if you think about subtractive machining, it now parallels just about any other solution available on the market. In fact, we signed a deal with ModuleWorks just a couple of months ago to basically complete that transformation. So there's nothing else on the market that offers anything more in terms of capability. So in terms of maturity for things like subtractive machining, it's reached a point where it's got parity with any other manufacturing tool on the market. What I would say is that the user experience is significantly better. So it's a much more modern product. It's the most modern product on the market. And its integrated nature means that it's easier to move design data through into manufacturing, regardless of where that design data came from. And then I think if you combine that with some of the digitization that you spoke about, this is a key trend for most of our customers right now is digital transformation. It sounds a little esoteric, but if you think about how you connect your supply chain together, how you transmit data from your design organization to your manufacturing organization, whether that's within your organization or through a supply chain, being able to digitize that and the cloud has led to some significant benefits as we've gone through the pandemic. So I think that in itself has led to a lot of opportunity. And then combine that with the maturity of the extensions. So our goal has always been to unify our portfolio on that single platform that you mentioned. And of course, we want to maintain the value of the technology offerings that we have. So similar to the sorts of transformation you've seen with the likes of Salesforce or Workday, anyone in the organization gets access to a base level of capability, which gives them a single data model in the cloud that everyone can use to collaborate together seamlessly. The specialist users can obviously pay for advanced capabilities for that particular persona. And some of those were missing in the past. So now we've introduced a simulation extension, a design extension that's already the generative extension. There are extensions for additive manufacturing, subtractive manufacturing and fabrication. So now you have the full range of tools to suit everything that you require on the way to physical manifestation on the shop floor. So I think that's 1 side is the technical maturity, as you mentioned, being able to collaborate in the cloud is particularly beneficial in the middle of the pandemic. But I think beyond that is also the go-to-market model. So with that exceptionally disruptive base offering priced at $495, it lends itself well to telesales and e-commerce, where many of the traditional products that you see in the market don't. Many of our competitors in subtractive machining charge upwards of $7,000 for a single license. Same on the design side. And that's not the sort of purchase that most people just make over the telephone or our website. And so if you have that base level client offering at $495, it creates an on-ramp. And so we see many, many on-ramps in the product. $500 is negligible for most organizations. So if you only see a small piece of initial rally, right, let's take that machine tool simulation I mentioned earlier on. In the past, that could have cost anywhere up to sort of $7,000 or $8,000. So just the machine tool simulation might be enough to tempt you to buy a single license at Fusion. But once you've purchased that, you quickly then start to realize that there are many other things Fusion could do for you across the rest of your business. So you see this initial seed offering expand across the enterprise. And so what we've tried to do with Fusion is create as many on-ramps for an organization as possible. Perhaps you just want to use it for digital collaboration. Maybe it's generative design that interests you, could be subtractive machining. Maybe you want to use some more of the sort of advanced additive manufacturing tools that are in there. So there are lots and lots of different parts of the business we can run Fusion in initially even if it's alongside a competitor. And then as the customer starts to get more experience with the offering, you start to see that platform make its way across to the rest of the business. So I think that go-to-market model has also helped accelerate the growth that you're seeing.

Do you ever see -- that's really helpful. And do you ever see, based on that, that they could use more into the CAD as well, right, like not just into -- because some of those extensions are best-in-class. And they basically, each on their own, stand against CAM, generative or whatever, product optimization, but it's also a very capable CAD tool to kind of -- for MCAD. Do you see increasingly the opportunity to kind of go into the legacy CAD user base and disrupt that part, which Autodesk has been very strong, especially in academia for inventor and stuff. But when you think about the commercial market shares, again, it's something that maybe #3, #4 in terms of the market share. So there's a lot of users, [indiscernible] users effectively to go after just on the design side like you said.

Huge amount of headroom for growth for us. Yes, absolutely. We're already starting to see it. Like the majority of our design wins actually come from competitive displacements from the likes of SolidWorks. I guess what I would say about those other competitors is there's a term that's always struck me. What I would say is perfecting the irrelevant. So you can take a piece of software so far that in order to justify the maintenance that you charge, you have to continue to add features. And you get to the point of diminishing returns with the features that you add, so that the majority of users don't even see a benefit from the functionality that you continue to add. And I think many of our competitors have reached that point. And we've seen this happen before in the past. When we made the transition from Unix-based systems across to Windows, AutoCAD did this spectacularly well. And I remember, I actually worked for an organization at that time and as a mechanical engineer was using Unigraphics as a 2D system background. And my boss at the time decided to switch out the 10 seats of Unigraphics we had for about, I think, something like 80 seats of AutoCAD. And I remember asking him at the time, "Why would you do that?" Because Unigraphics is at that stage in 2D. It was like best-in-class. And he said to me, "The reason we're doing it, Steve, is because the benefit that we get by unifying the whole organization on the same digital platform outweighs the small benefit we would get from your 1 design department being best-in-class." And it's kind of like moneyball. If you play the averages for the whole company, you get a better result from a high-performing team than you do from a star player. And so the differences between Fusion's capability and design compared to something like SolidWorks are negligible, and to many users, they're not even perceptible. So the benefit that they get from being able to connect their machine shop and their inspection and their simulation capabilities altogether on that single digital platform outweighs any small productivity benefit that they may have had in design. And we're quickly closing the gap on that. That's why we introduced the design extension. We actually see the opportunity to surpass SolidWorks in many areas in design. So that's one of the key areas of focus for us right now.

In the past, I've heard -- there was just a couple of years old comment when we used to still be able to go to conferences, but I did speak to 1-person, decent-size shop, where they said, we've been buying or subscribing to Fusion recently and can do about -- at that stage, they said on the design side about 90%, but he does also about 50% more, right, which will have cost me a lot more if I were to go out and buy all these things. And more than 90% of the time, that's well enough, right, of what it was doing, but you're saying you're closing the gap on that side?

Absolutely. And the other thing is it's kind of risk-free for a customer. So at $495, you dip your toe in the water, it's negligible. And then for most of our advanced functionality, yes, we offer you the opportunity to subscribe to that as an extension. But of course, you can subscribe to an extension for a shorter duration of time, maybe just a month. And for many of our technologies, things like generative design, we also offer them on demand. So it might be that you only want to use it once or twice. You can test it out with very minimal exposure from a financial perspective. So pretty attractive, low risk. I mean what's not to win by trying it out.

And last year, you've actually decreased, right, the all-you-can-eat kind of the side of generative design used to be around $8,000, a significant drop in, I think, $1,800. Have you seen increased number of designs and things being generated because of that, because you basically lowered...

Yes. We saw up to about 4x in terms of improvement there in adoption. And again, it's just part of derisking it for customers. I think our view in the past, and if I'm honest, I think it was a mistake, our view was that, inherently, there's a tremendous amount of value in generative. And there is but it's a new paradigm for people. And it breaks established norms. And if you're going to break established norms, you have to help people into that process. And there's a couple of ways you can do that. You can lower the price, which reduces the risk in experimentation. You can also start to federate it with more traditional processes. So for generative design, today, it's a fairly big paradigm shift from something traditional at parametric modeling to move to generative. So generative requires you to kind of establish a design problem like you would do in a simulation setup, and not everyone is comfortable in doing that. So what we're now doing is extending generative capabilities towards what we call generative modeling, just slightly different. It means instead of having to define things like forces, fixed constraints, manufacturing processes, you literally tell the system where you have a gap in your product design. So if you maybe think of a complex mechanism with a linkage bar, you just say you need these 2 charts linked together. And it will go away and produce geometry for you. It will create maybe 3 or 4 different options for you. Now if you want, you can take one of those options forward into generative design and evaluate it against stress or you can just accept the geometry that it produces for you and insert that into the design automatically. So you can kind of think...

[indiscernible] better than why you had come up with in the first place anyway?

And certainly a lot faster. Think of it as like autocomplete in texting. It's the engineer's autocomplete function. And because it behaves like a parametric feature, it becomes a lot less of a cognitive shift for a user to start experimenting with it. So we think that will be 1 secret to driving more generative success. The other is to expand the scope of generative. Our belief is that we should be getting into systems. Systems is the best way to represent a complex product, and it's the key to delivering generative product design rather than generative part design. So for us, easy to reach. Again, let's not perfect the irrelevant like we've seen with some forms of system engineering like Modelica-based solutions. Let's create something for the masses that's easy to use, that everyone can access, that represents the whole product and then apply generative technologies to it.

Do you feel that on generative specifically, you're still getting the pushback where especially the designer community sometimes feels a little bit threatened by the fact that it might be that to replace them? Would you think we've crossed that bridge to the other side where it's like this makes me a superhuman in a way, but I didn't -- I wasn't able to do it before? Or is it still kind of a mix on the way to go that?

I wouldn't say we've crossed the bridge yet but we're certainly almost there. I think 1 thing originally with generative design that put people off from me, there was a few things. Firstly, the outcomes were always additive-based outcomes. And of course, that's not applicable for many applications out there in the field. So since we introduced things like 2.5-axis, 3-axis milling constraints, it's become a lot more usable for people in real-world scenarios. And then the second thing is we made the outcomes editable. So in the early implementations of generative, you ended up with like this mesh geometry. And if you didn't like what it gave you, there wasn't much you could do with it. So people would often take the mesh and use it as a source of inspiration, which meant that they had to pretty much reproduce everything. Now what we do is give you editable geometry, which doesn't cut the human out of the loop. It just means the human can use it to create a lot of different ideas that are perhaps better than they would have come up with initially on their own. And then they can spend their quality time perfecting those ideas. I think that's been one of the keys to making generative more approachable is focus on the user experience, make sure that we can actually produce geometry that's manufacturable through traditional processes, like you mentioned earlier, most of the world out there is still using subtractive machining, and then make sure that the user's got the opportunity to actually edit those results manually afterwards. That way, we're actually having the computer assist the human use and not replace them.

Right. It's less of a black box for them as well, probably because they can kind of feed [indiscernible] not just feedback.

You see this everywhere. Microsoft has done a really nice job of this. If you use PowerPoint and you just drop the new imaging there, PowerPoint will come back to you and say, "Look, here's 6 suggested slide layouts. Pick one." You use one. You never use it as the finished slide. It just ends up being a starting point for you to then edit so that you become more productive, a lot more productive. Again, if we just shoot for sort of 5% to 10% productivity, that is never enough to persuade a user to go through the cross-boundary, the switching cost to actually change systems. You need to be delivering something like 100%-plus productivity.

Awesome. And Steve, 1 thing you mentioned as well is in the past, we went from kind of Unix to the PCs. And we've gone through -- and CAD has evolved over time, right, from workstations to now effectively dedicated cloud capacity that mostly kind of driving all of that. With every shift in this major platform, with every platform shift, we've seen a significant amount of deflation, price deflation in the CAD itself, right? These machines that [indiscernible] CAD initially on cost $60,000, $80,000, then it came down to $20,000. And with PCs, you could go to AutoCAD for a couple of grand where you could even get, [indiscernible] I don't know, $7,000, $8,000. And now you've basically pushed it so far to literally $495 to kind of get it started. What has enabled you to do that? And like is that just a normal thing that has happened so many times? And I guess going from $80,000 to $20,000 was just as inconceivable as it is now to go from $7,000 to $500 as well? But how could you do it and develop -- deliver all that value to it?

I've always put it down to it's always like an order of magnitude with each shift in technology. So you see 3 components. The first 1 is the platform shift. So Unix workstation still, Windows-based workstations, almost in order of magnitude, then you see the actual go-to-market model change. But the go-to-market model changes usually because the cost of ownership changes. So where an $80,000 system would support direct sale, a $5,000 system really only works with the channel network because you start to scale the business. And then when you go to e-commerce, again, you come down on order of magnitude. And that really only works when you have the e-commerce mechanisms and telesales to be able to scale the operation beyond what you have through a channel network. So I think first off, you're looking at volume. So the actual volume that we can scale to is huge. It's absolutely massive. But what we don't want to miss is the value piece of that. And that's where the strategy of combining this base offering at $495 where the extensions is key. So extensions, they're still not as expensive as the traditional offerings. If you look to one of our subtractive offerings like PowerMill, it could have been in excess of $8,000 for a seat. The extension is somewhere in the range of $1,600. You can actually buy a manufacturing persona offering now, which is about $3,000. So still significantly cheaper than what we have with PowerMill. However, it's still reasonably well priced to realize the value that we're delivering to the market. So I think what you've got to consider is that Fusion really isn't $495. $495 buys you access to the unified data model in the cloud, and it gives you entry-level access for just about every piece of the product development process all the way through. Then what you do is you layer in on top specialist personas where you can start to realize the ACV behind the product value that you're delivering. You combine that with the extra adjacent expansion opportunities that we have. So think about the shop floor in the past. We've produced manufacturing instructions and then left it to the user to copy that onto a memory stick, take it down to the shop floor and execute it. Now what we have is this 10:1 opportunity, just like we have with PLM where for every production engineer that's producing machine code, there were 10 operators down on the shop floor consuming that. And then there's maybe 100 machines. So if we start to license Fusion to the operators on the shop floor for manufacturing execution or to the machines for direct connectivity, then we easily expand our opportunity in terms of volume by an order of magnitude again. So those kind of economics lead to a pretty robust business model. It's difficult for our traditional competitors to match. Actually, being fourth in the market, in some respects, gives you an advantage. We have everything to gain and not much to lose. And so if you think about somebody like SolidWorks -- and again, I want to be clear, cost leadership strategy is probably the worst strategy you can have because anyone can follow you down the rabbit hole. But of course, for our friends at [ The Sow ] or at PTC, that's pretty difficult to do without trashing their business. And secondly, that isn't what we rely on solely. Like I said, the key to our strategy is automation. And those automations will be the things that we really realize the value of the solution on later on in the process. So ideally, we want to get to the point where you as a user create the design. And once the design is created, any derivative information that you need to either document it or manufacture it is just produced automatically. So if you -- if you just think about out in the future a little ways, imagine a SolidWorks user who just uploads their park to Fusion, and then 30 seconds later, gets machining instructions and the 2D drawing automatically produced for them. They won't see 10% productivity gains or even 100%. They'll see something in the region of 400% to 500% productivity gain. That's really where we'll start to make our money, I think, out of the back of those automations. So you can think of the $495 as just establishing the base for the users out there, and then we layer in the financial opportunities with the extensions and automations.

Interesting. Yes, if I heard you right, basically, the way you framed it was to really have that persona we're going after, it's around 3,000, right, or whatever it might be in terms of the apps. But the point there is there's so many different stakeholders that we can sell to as well, not just the designers but actually the market itself. And I like that slide from your Investor Day when you showed, right, that if you were to combine what is in Fusion today with traditional offers, you'll be paying [indiscernible] $50,000, right, or even more than that. And that's the point because you can go from 1 to 10 stakeholder and get that value out of it, whereas others cannot consolidate back because CAM solutions that are dispersed, they don't have CAD and then they'll have others. And I guess what enables you that is 15 years of M&A and R&D and everything that you've kind of done and kind of finally bearing the fruits of that investment that has been done over [indiscernible] decades.

No, I would say, first of all, it's the vision and the commitment to delivering on that vision, which is -- takes sustained execution. And what I'd say is, think about the strategy I outlined, just converging technologies by themselves yields some significant benefits. But the end game is that full automation I spoke about, you won't hear that message from any of our competitors. What you'll hear from our competitors is we can run it in a browser. Well, so can we and we'll be delivering all of that in a browser before long anyway. Most of our customers still want to use some of their dedicated clients as well, so we'll offer the best of both worlds. You'll hear that people are perfecting machining. Like I said earlier, perfecting the irrelevant doesn't deliver or yield much value to our customers. You won't hear this message of automation coming from nearly anyone else out there in the market. So I think it's fairly well differentiated from many of our competitors, and it, most importantly, puts the customer at the center of what we're trying to accomplish because from a customer's perspective, it really is the best strategy to help them become more productive in the market and more competitive.

And as a result of all that, what we've just talked about on the top line specifically, I guess what happens is Fusion 360 and the manufacturing as a whole becomes a more important part of that contribution to growth over time, right? I think you've shown that at last Investor Day as well. How you're thinking about just conceptually the '24, '25, '26 and kind of going forward, you don't think that what we've experienced over the last 2 years, which was potentially kickstarted by the pandemic and a little bit more demand happening, is just the beginning of that millions of users that could be on the platform.

Yes, and it's all part of a sequential strategy. So the first thing that we seek to accomplish was to build a cloud platform, which is, hence, the investment in Forge. Second 1 was to create a fully integrated environment on that platform for our manufacturing customers, which you're seeing with Fusion. That convergence leads to the automation opportunity that I spoke about, and that's something we've talked about for somewhere in the region of 5 or 6 years now. The next opportunity after that is the convergence between the industries. So if you think about some of the trends in the construction space, which is why we chose to go there first, industrialized construction is something that we see as a huge opportunity in the future. So there's something like 70% waste in many construction efforts. And that's because they're really applying one-off tailored manufacturing techniques that in the manufacturing space died out somewhere in the region of like 150 years ago. So if you can start to decompose building structures into smaller, more manufacturable modular elements that can be reconfigured to create different types of building, you can start to pursue off-site manufacturer and actually start to use some of these traditional manufacturing tools to produce. But of course, you can't compromise on the principles of architecture either. So by connecting the 2 together, and we really are the only player out there that has the opportunity to do this, we can actually start to take architectural designs and ready them for off-site prefabricated manufacturing in a factory, which then obviously yields less waste out there on the construction side, higher quality and predictability for the consumer at the end of the process. So that convergence opportunity between manufacturing and construction, we think, is another key source of growth that sets us up beyond FY '26 as we move out into the future. So we're quickly investing in that space, too. So it's not like you see this as a one-and-done strategy. There are many levels and layers to this strategy as we continue to build out toward the future.

That's interesting. On the bottom line, I guess, the way this plays out is it really starts contributing to your growth and doesn't add much to the cost base. Because like you said, a lot of these tools are -- well, the tool itself and the extensions, they're predominantly e-commerce-based, right? So you've -- Autodesk used to be 87% of revenue, used to be generated by the channel, right? I would imagine that when you look at something like Fusion 360, the proportion of channel revenue and the total revenue is probably flipped the other way around-ish, which is -- but it's not direct sales force, it's mostly through e-commerce and self-service model, which is very easy to [indiscernible].

I think the important word you just said there is proportional. That's the thing not to lose track of. We expect to suggest as much, if not a lot more business come through our channel. But proportionately, we'll balance things out because of the volume that we can push through e-commerce. This is actually a very good thing for our channel because the more Fusions there are in the market, the more opportunities there are for those partners to go and upsell extensions and all the service offerings that go around them. So yes, absolutely. The ideal blend between our historical channel go-to-market methodologies and the new more e-com-focused go-to-market opportunities is through that extensions framework. So yes, I think there's an opportunity for both. You're absolutely right. We would expect to see a much higher proportion come from e-commerce. But that's probably in terms of kind of units out there in the market. We'd expect to see a strong growth of extensions through our value-added channel partners as well. Of course, we see that happening in e-commerce as well. You'd be surprised, when we actually started through this process, the reason that Fusion was priced at $495, just below $500, was our assertion was that, that was just the right sweet spot to fit on a corporate credit card without having to go through any additional approval processes. But what we're actually starting to see now, many purchases. I think we had a customer last week who bought somewhere in the region of 8 manufacturing or machining extensions and bought them all on a credit card. So it does happen. And I think you're seeing shifts in that direction. What we hadn't anticipated at the beginning of this process was obviously pandemic, which has tremendously accelerated our go-to-market efforts. As you can imagine, people running around in cars and planes trying to sell products in the traditional way has taken a serious setback over the last 2 years, whereas e-commerce and telesales hasn't. So it's been a good sustaining growth throughout the pandemic. But we don't see a return to what things were prepandemic. Of course, I think we'll go back to a more balanced environment where travel is free for everyone again. But I think you're going to see certainly a decentralization of organizations, and that lends itself well to cloud technologies and e-commerce.

Awesome. The other topic I wanted to touch on today was just a convergence of manufacturing construction. Autodesk, amongst the -- pretty much any company that we look in the space, is the only one that is scaled in both, and each of those size are bigger than most of the other individual companies if you look at it that way, right? So you're pretty well positioned to understand both parts of the equation, I guess. And when we talk about things like prefabrication, right, it's different to just understanding the building. So you almost need to understand the manufacturing because when things are done off-site in a factory, they are done with the precision of the manufacturing methods, right? So suddenly, tools like Revit and Fusion start being relevant together and whatever we've heard previously from customers is that if architecture customers are starting to adopt Fusion because they need to, because they need that level of even inventory at times because they need that level of detail, not for the job side but when the things get constructed. My question here is, is that -- does that lend itself to kind of just a natural integration of Revit and Fusion at some level? Or is there an opportunity for Fusion to kind of move more into the architectural space over time as well and kind of maybe become Revit or Revit introduces a product that's more Fusion-like with extensions and stuff as well in the future? What are some of those kind of connections between you and the other parts of the business?

There's a lot to unpack there. Let's maybe take it...

We've got 3 minutes, so yes, sure.

Let's do short term, long term. So short term, we've introduced a technology called FBX, which is our exchange interface between all of our products. So absolutely, Inventor remains just as viable a solution in the space. It's a very mature product and it connects exceptionally well to Revit. So FBX basically enables you to use the Forge platform as a broker in the middle. So imagine a Revit user and an Inventor user, Forge in the middle. The Revit user basically publishes their building information model to Forge. Now the Inventor user doesn't want to pull the whole building information model. It just want pieces of it. And so Forge acts as a broker in the middle to allow them to selectively import elements of that into Inventor. This is absolutely ideal for things like factory planning. We see a great opportunity for Autodesk manufacturing and factory because it combines BIM 360, Revit for building shells, structural steel work. It combines immersive technologies for virtual commissioning, factory line planning, everything you can think of. And it's a great leading indicator for this convergence between industries. So that's one of the things I would say makes us stand out at the moment. But obviously, there's a lot further for us to go and where we're going to go next is on the automation side. So helping architects package elements of their building and turn them into manufacturable subsystems automatically is something that we very much got our eyes fixed on. Now in terms of Fusion, the important thing for you to understand there is that Fusion is built on Forge. That's the company's platform for our cloud architecture. What we're going to see over the coming years is the introduction of 2 other offerings for vertical customers. So 1 will be [ Former ], which will serve the folks in the AEC industry, and the other 1 will be Flow, which serves the needs of the people in the entertainment and media sectors. Now the intent here is that all 3 of those vertical offerings, those customer environments, Fusion for -- sorry, Fusion, Flow and [ Former ] are all built on Forge, so they're vertical instantiations of Forge. And because all of the information is stored in the same information model in Forge, it means you can move information seamlessly from the AC environment through into the manufacturing environment and execute on it. So to answer your question directly, no, I don't see Fusion becoming a tool for architects. That will be [ Former ], though [ Former ] and Fusion are basically twin brothers. They're built on top of the same Forge platform and they have the same data model underlying. So a slightly different customer experience tailored for an architect versus something tailored for a manufacturer. But of course, the 2 are basically still -- you can think of them almost like extensions. One is a persona extension for an architect. The other 1 is for a manufacturer. And of course, the 2 operate together on the same foundational data model. So it's like a superset of what we've offered in Fusion. And that, we believe, will be the key to industrialized construction in the future.

Awesome. I think that's a great way to wrap up. Thank you, Steve, for joining us today. As always, very, very, very interesting talking to you, [indiscernible], but thank you so much.

No problem. Absolute pleasure.