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

Welcome back, everyone. We're back to our live sessions. This is our eighth session of the day in terms of the lineup, and I'm very, very glad to introduce Autodesk and the team from Autodesk. We've got Steve Hooper, who is their General Manager of the Fusion 360 and the manufacturing efforts that they have. So it should be a very interesting discussion. But before we start, I'd like to introduce Simon Mays-Smith as well, who is the Head of Investor Relations at Autodesk. And he's going to give us just a little bit of a legal introduction.

Thanks, Gal. Just the usual one, which is 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. Thanks, Gal. Back to you.

Thank you. Awesome. So Steve, hi, and thank you so much for giving us your time today.

It's my pleasure. Thanks for having us.

And some investors that might be dialing in or logging in, whatever they're doing, might know you already. We've -- interestingly, I would say that they will probably get to know you even better in the future considering the fact that I would say that Autodesk manufacturing business is definitely getting more and more prominent within the -- becoming increasingly more important as the whole kind of investment thesis plays out as well. It's fair to say that a lot of the focus in the past has been -- especially the past few years, has been in Construction itself, which might be a little bit unfair considering the fact that you've been developing specifically Fusion 360 and -- at the same time and starting to get some really good adoption in that developer market. So based on that, Steve, what I'd like to kind of focus on first is that would you agree with my statement that Fusion 360 and the home manufacturing approach to Autodesk is an increasingly important area for you, especially as you think about kind of the future and where Autodesk is taking -- and it's an area that you're taking extremely seriously?

Yes, absolutely. So I think Andrew, our CEO, called this out at Investor Day quite well when he said beyond FY '23, we look towards manufacturing to contribute accelerating growth for the company. So it's been a high area of investment for us right now to build that foundation to create that kind of groundswell with a fast-growing user-based community, which you've seen throughout the course of this year. So yes, definitely a very big area of opportunity for us.

Right. Right. And when you look at the opportunity, especially in the maker community, and make no mistake, it's a powerful set of tools as well that they're starting to make its way all the way to the traditional designers as well, right? So how do you size that? Let's call it -- I don't know. Should we refer to it as a cloud-based, kind of SaaS-based CAD and 3D CAD for manufacturing or whatever we want to call it? But how do you look at opportunity overall? Who is your target market? And maybe you've created some sort of personas or something like that.

Yes. Great question. I think the reason people struggle with that is it doesn't fit any of the existing profiles of predetermined CAD or software categories in the design and manufacturing business. So if you think of the sort of typical predefined categories that people have operated with for the last 20 or 30 years, we've thought about CAD, whether it's 2D or 3D computer-aided design. There's been CAE, which is computer-aided engineering. That's a catchall for all the simulation technologies. There's computer-aided manufacture, CAM. And then there's PLM and data management, so product life cycle management. So each one of those categories has kind of had individual products that serve the needs in that specific area and address just one portion of the overall product development process. So what's unique about Fusion is it addresses all of them as a single platform. So it's not CAD. It's not CAM. It's not data management, PLM or simulation. It's not one of those things. It's all of those things together on a common cloud data model infrastructure. And the reason that that's attractive for a customer is that typically, if you're using these stand-alone products, anytime you contribute information in the CAD design area, it's difficult to realize that value in any other area because it's in a different file format. People can't benefit from that information because they have to translate it. And often as part of that translation process, they lose the fidelity of that data. So just like any other SaaS-based application, think of Salesforce in CRM or Workday in HR management, what we're doing is creating a single data model in the cloud. And that means that any point in the process, when somebody contributes information or data to that model, everyone else benefits from it. Not just the people in that one department. And it also means that everyone's got access to it anytime, anywhere they might be in the world. So that's the kind of powerful thing for customers. That's what attracts them to Fusion. What I would say is that if you talk about sizing it, traditionally, people would have looked at those predefined categories, and they would have said, what's the end user expenditure in CAD, for example. Well, because Fusion bridges the divide between design and production and manufacturing, we have to kind of think of the broader set of the market. So in design, which is where a lot of people focus, if you think about competitors like PTC, Dassault or Siemens, people would look at design and they would equate that to about $19 billion in end user expenditure. And that's accounted for by 9 million degree-qualified professionals. The interesting thing for us is that because we don't just exclusively focus on the design well, we also converge that with what's going on in manufacturing and production, that gives us access to an additional TAM of another $14 billion. And there's actually maybe twice as many -- over twice as many users in that space. There's actually 20 million degree-qualified professionals that work in the manufacturing environment and aren't addressed by design technologies. So combined, when you bring those worlds together, you're looking at a combined opportunity of roughly $33 billion and nearly 30 million professionals. So it creates a significant opportunity because we're able to serve that market in a way that many of our competitors don't because lots of the new generation of more sophisticated manufacturing techniques demand that you think differently about how you design. You really need a solution that converges the design technologies and the manufacturing technologies together. And that's where we're uniquely positioned to deliver that as the only cloud tender that has a single end-to-end tool chain. So that's how I think about the market for us, and that's how I'd size it. It's a massive opportunity. You've seen some of the growth that we've had over the last year, a lot of that obviously stimulated and catalyzed by what we've seen from the pandemic and the lockdowns that many of our customers have had to work through. Obviously, remote collaboration with access and connection to a single source of truth that everyone's contributing information to and benefiting from is kind of a unique opportunity for us.

That's quite interesting because especially, it would also signal that you're just kind of in the first stage of the adoption of the development, right? Is it fair to say that over the last 5-ish years or so, 5 to 7 years, you've really been focusing in developing the tool that -- or platform, actually, that enables all these things that you referred to? It's a lot more than a CAD tool. It's a PLM tool, including PLM. It's a CAM tool. It's a simulation tool. It's a generative design, if that's simulation or whatever it is, tool. So if the first stage is let's create that platform, and then the second one adopt -- let's make sure that adoption counts and gets in. And a lot of the focus so far has been on kind of universities, on education and that kind of seeding it effectively. And now we're just starting to move. You mentioned '23. Maybe that's a good time line when we start thinking about material contribution to Autodesk as a business as well. Is that fair?

Yes, that's fair. Yes, definitely. I would say that, as you correctly point out, over the last 3 or 4 years, we've been working diligently to build out that platform. So it isn't just -- it isn't just a CAD tool that's been ported as a SaaS application. It's something different because I think the problem is if you take an existing paradigm like CAD and you just move it to a SaaS infrastructure, you're really not solving any meaningful problem for your customers that hasn't already been solved. And then the problem you've got is that you're asking a customer to endure switching costs and paying nearly the same amount of money for the same solution they've already got with the only difference being desktop to SaaS. And like I said, if I go back to any other successful SaaS organization that you can think of, one core fundamental underlying principle to that success is that you create a single data model that the rest of the organization is always contributing to. So if it's Salesforce, anyone in your company that touches the customer, they record that information in that Salesforce CRM tool. They enrich the profile and everyone benefits from that enriched profile. So if you only build one element of the process as a SaaS application, you're really not delivering any sort of benefit beyond what the customer already had. And then, of course, you're asking them to endure switching costs on top of that, which makes it not a very sound commercial proposition. So for the last sort of 3 or 4 years, what we've been doing is taking a number of the acquisitions that we've made with this in mind and refactoring those technologies and their expertise to suit a SaaS application platform. And that's basically what Fusion is. So if you think about organizations like Delcam that we acquired for subtractive machining, we acquired netfabb for additive [ design ] for industrial design, we've been bringing all of those technologies together on the same data model.

And it's not -- so the point you made is important because it also doesn't require anyone to switch necessarily, right? You can use Fusion 360 alongside with your SolidWorks or what you may have?

Correct. That's been one of the attractive value propositions we've had, is because -- think of every one of the historical capability needs that we address. So if you're a machinist and you were using CAM, well, Fusion can address that need, that underlying need that the machinist persona has. If you're a designer and you need CAD, we can address that need. We've started adding more and more capabilities that address more and more unique needs. So for example, we introduced machine tool simulation, which lets you test machine code that your system has built before you run it on hardware on the shop floor. And that lets you avoid costly crashes of physical hardware. Think of it as a digital twin. We're carrying a digital twin of the machine tool and your shop floor, so you can test it before you run it and avoid crashing $200,000 worth of machine tool. Now that was an area of business where there are a few incumbent software companies that serve that market. We can go in there and replace them but complement other tools in your tool chain like SolidWorks or Mastercam, for example. We could sit along as alongside Mastercam and perform machine tool simulation on that output. The interesting thing is though that once a customer makes that investment for that one area of need, they then quickly realize that Fusion could actually serve many other areas of needs that they've got more cost effectively and with a better ease of use. And then we see this kind of expansion in those accounts. And we've been pretty diligent about setting that up for success. So for example, we can read SolidWorks data in associatively. So if you're a SolidWorks user, you can use Fusion to do your manufacturing. You can read your SolidWorks data in, produce your machine code, test it. And if the data changes in SolidWorks, they will update in Fusion. So as you say, it can work very nicely as a companion solution.

So would it be fair to say as well that when you take Fusion 360, what -- because the focus so much is on the maker community, and when I think about that, what they're interested in is having the CAD file, it doesn't matter if they made it or someone else supplied it to them. It doesn't matter whether it was generated within Fusion or another inventor or another CAD offering like SolidWorks. Being able to do that without the other steps in the process, which people might not be aware of, but if you talk about subtractive manufacturing and CNC machining and stuff, there's a whole lot of coding that's going on in between that. And it doesn't just happen. You have your CAD and you have your machine. You press the button and it happens. There's a lot of prototyping, a lot of playing around with little parts and a lot of green screens like coding still happening on the shop floor in the -- what's called digital factory today, right? Is the idea to ultimately build this one end-to-end solution that takes the CAD data and then you see your machine part at the end, which is not what is the case today?

Yes, absolutely.

And it's both for additive and subtractive manufacturing, I guess.

Correct. And that's the paradigm that's already well explored, tested and proven. And it happened in another industry, which was the desktop printing industry. So if you think back, think back to kind of -- if you're based in the U.K., think back to Fleet Street and printing presses and you had typesetters. You had all these areas of specialized capability and see how machine operators and typesetters and editors and everyone worked in their own area and had the hand data across. Hugely inefficient. Today's modern practice is desktop publishing. And desktop publishing really comes down to what you see on the screen is what you get when you hit print. Our job is to basically replicate that in the world of design and manufacturing. Because if you think about a 2D printer, it's actually not that much different to the 5-axis machine tool. It's just the 2D printer moves in 2 axes, and a 5-axis machine tool obviously moves in 5 axes. So basically, what we're trying to do is work on making it as easy as possible to go from the conceptual idea that you form with a 3D model to the point where you can hit the print button. Now obviously, that is grossly oversimplifying some of what you just described. There's a tremendous amount of technology hurdles that need to be overcome to do that.

It's more than just -- it's outside of printing edges.

Well, the thing that makes it complex is that most of the machine tools that are out there have different controls. So you have this constant mix of different machine tools and controllers, and people have to write what they would call an interpreter to take the machine code you've developed and translate it so it can run on that machine. Well, we're working with a lot of the vendors to actually get rid of that. So for example, we worked with Haas, who built Haas machine tools. Very popular. What we've done is work with their team to develop something that is closer to a printer driver. So you don't have to have a third party come in and write this custom bit of code to act as a translator. You can simply point the Haas machine on your shop floor and send the job directly to it in the same way that you would to an Epson printer or a Canon printer. So there was a driver rather than this custom interpreter. That's just one example of some of the things that we're working on to make that process easier.

That's helpful. I think everyone can relate to installing drivers for printers. I'm pretty sure...

Right. And it should be so transparent. But it's like Windows on Mac OS. But that driver is just installed for you. It recognizes the printer on your network. It installs the driver, and you never even have to see it. It happens, but it happens transparently.

Steve, how far away are we from you to say, okay, where we have most of the -- we cover 80%, 90% of the machining world today and we are the section where you can press a print button and it will?

I mean the ultimate vision of being able to press the print button is still some ways away. But even today, what I just described is available. You can create a 3D model, [ depending on ] what model of Haas machine you've got, have a digital twin inside the Fusion that will simulate the machining of your part in that Haas machine. And then we can connect directly to the Haas machine. You can send the machine code straight to the shop floor, straight to the machine controller. So that's available today. What we'd like to do is take that even further and start to automate the production of the machining strategies themselves. So that's where the real productivity comes in because then instead of a production engineer spending half a day programming a part, the part is automatically programmed in 10 seconds. That's the kind of productivity differential that we think we can deliver with the cloud. So the first part of our strategy was to connect all of the different parts of the process on a single common cloud platform with a single data model. Yes, the benefit is that everyone is contributing to the same source of information and everyone is benefiting from it. But the real benefits come next when we start to apply automations. And you can only apply those automations when all the data is hosted together in that single cloud location. That's where the real productivity gains come in. That's what you see from areas of technology like generative design.

Yes. Very helpful. And when we touch on generative design, basically what you're trying to achieve there is if -- if I'm correct, you're effectively taking a job that's fairly artist-like, fairly creative and taking the concept of, okay, we -- as people might consider 3, 5 different options when we start our design. And then we narrow down and we go with one, and then we run away with that one. And if it works, great. If not, we have to go back. Whereas generative design will start from the other premise like, okay, let's take 100,000 of different options and see which one might work for you. There's no one. There's 10 or 20 of them, but you can pick which one we think would be the best based on, a, what your manufacturing process is. It's not all 3D printing. Is it -- some of that might be subtractive. And then B, the forces and the limitations that you gave us, right? And throw in as well different materials that we can use. Is that -- that's just too complex for a person to -- for a designer to kind of put in. We are not able to process, even though we have great computers in our heads, like it's very, very hard to do that, right? So do you believe the world is understanding that generative design is something that's more powerful than a little design team of 2, 3 people trying to come up with a solution? Because -- I'll give you a piece of feedback. When I was -- in 2015, '16, when we were just kind of releasing the early stages of generative design, I was at a DEVELOP3D conference, I remember, and I asked one of the attendee, so what do you think about this? Would you do it? And he was a designer. And he said, "Are you kidding me? My boss first would ask me like why didn't I do my job; and then secondly, wouldn't believe what I gave them; and then thirdly say, well, if it works, why are you still here?" So the question is, where is generative design today considering the fact that it is objectively better than what people designed but 99.999% of designs are still done the old way?

Well, first, that was an amazing description of generative design. I think you described it better than I would have described it. So I mean I think -- I agree with you. And I agree with that observation. And I think there's part of it that is still true today. So is it understood? Probably not. So people often confuse it with a similar branch of technology called topology optimization. And what that does is you take an object, and you apply loads to it and weight to it, and you tell it where it's supported. And then the software nibbles away -- it takes away any of the material that is surplus to requirement. So you're just left with a structure that can support the load adequately based on its space envelope. So what you end up with basically is a slightly less bad design than you already had, and that's the problem with topology optimization. You are potentially taking a bad idea and making it slightly less bad. So what generative design is aimed at accomplishing is just what you said. It's like a very large simultaneous equation. So if you think of everything you just described, material type, machine tool options, different loading conditions, different constraints, let's say there are 5 variables that you could play with. And each variable has got maybe 6 values. What you have in the solution universe is 5 to the power of 6 potential solutions, right? The problem is, as a human being, we'll do what you just described. We'll come up with one idea, and then we might iterate 2 or 3 variants maybe if we've got time. The trouble is that those variants will be very close to the initial idea that we have because that's the way human beings were. What generative design does is it allows you to explore the entire solution universe and make the right appropriate trade-off decisions between performance, cost, manufacturability. So I don't think everyone fully understands that yet. So that's, I think, partly an awareness problem. I think technically, it doesn't solve every problem we need to solve yet. So at the moment, it's excellent for things like part design. So GM, for example, employed it for a few part design capability challenges that they were running. And it delivered fantastic results. But we all know that products are made up of assembly. It's not just singular parks. So we need to be able to look at systems. So that's another problem. And then perhaps the most profound problem right now is that because it's a branch of computing that was derived largely from simulation, it still looks and feels a little bit like a simulation tool, which isn't that accessible and approachable to a designer. So what we're focusing on right now is reimagining generative design to become -- I'll give you another analogy. I want generative design to be much more like auto complete when you're texting somebody on your cell phone. I want it to be able to make an inference about what you're typing and complete it for you before you finish typing it. So I still believe the initial version of topology optimization that we have, which is kind of simulation focused, is valuable and will continue to grow. But I think this kind of focus on automated design, where we can finish people's thoughts for them, that will be more ubiquitous. More people will be able to gain benefit from it, and it will aid productivity in a greater number of use cases. So I'll give you an example, if you're designing a structure, you need a few cross-beams in the structure. The system just realizes that for you and puts them in place for you. You can tweak it and modify it, but the system is kind of 1 or 2 stacks in front of you, just like it is when you're texting on your cell phone. So I think that's probably one of the futures of generative design. The other I would say is, remember, I said it's only solving a simultaneous equation for you. That's basically what it's doing. There are many other application areas for it. So in our entertainment and media solutions group, we've just implemented what we call generative scheduling. So that takes the schedule for a film production. And it basically looks at all the possible variations, how you could resequence the production schedule. And it basically uses generative technology, not to come up with an optimized shape but to come up with an optimized schedule. Now think how unbelievably valuable that would be to a factory operator who's trying to schedule jobs on the shop floor. So I think the future of generative is a lot broader than just generative design. It includes assemblies and systems. It becomes an automation technique for just pure design, and it also starts to touch on other areas like scheduling or production.

Got you. Yes. And you've also started moving into other areas. Like you said, media entertainment is one of those. But you also -- you acquired Spacemaker as well, which does a similar type of thing in the area of architecture kind of landscape design and things like that, right?

Yes. And I think for us, from a business perspective, it's just -- it's broadening the addressable market. So the broader the appeal, the more units we can sell. That's the hard economics of it. So that's what's really like driving us. We want to better serve customers with a much broader addressable market. So I think generative design as it is today will continue long into the future. It will get more sophisticated, and it will be able to deal with systems. But we will derive that technology in a way that lets everyone benefit from it in a much broader set of addressable opportunities, and that will create more economic opportunity for us.

Awesome. Now I know your monetization efforts around generative design is slightly different to the kind of start-up platform the way I can basically start trialing, I have a Fusion 360 for free very easily, if I'm not making any money out of it. But I did speak to a number of your customers and one conversation, in particular, kind of struck me. It was about 1.5 years ago now. If I remember, it was one of your not smallest customers by any means. It was a medium-sized customer and said, "Fusion 360 is really changing the way we procure CAD software. We tend to be open in anything. We have inventory. We have SolidWorks we have other areas. But we've really been mostly procuring Fusion 360 recently." And I gave you one feedback for them. And when you get the chance to ask them and -- it's interesting. And it was -- the question was like, how can they provide all this value for -- and it was bizarre because I don't hear that very often, but he thought it was way too cheap. You played around with pricing a little bit. It tends to be between all $250 all the way to $1,200 kind of per year or whatever. Right now, we're somewhere in the middle-ish, around $500. And then -- but that's not the whole story. How are you trying to really monetize and making sure that you're not leaving too much money on the table? And I understand that from a product perspective, you had a lot of technology. Autodesk, over the last 15 years, did a lot of acquisitions. And those were great acquisitions. You mentioned Delcam as an example, right? It's one of the market-leading CAM solutions. Now you're starting to put that into Fusion. You don't put it there for free, right? So how do you ensure that -- again, that you don't [ leave ] money on the table?

Right. There's a lot of aspects to this. So let me try and break it down into a few areas. So the first, I think, is market evolution. There have been kind of 3 phases that we can see so far. When I started in design, everything ran on a Unix workstation. So I used to use a version of unigraphics that run on a Unix workstation with a McDonnell Douglas interface. And that whole system cost about $35,000. It was so expensive. We could only employ 10 of them, and there were 25 of us that were in design. So we had to share those systems. Now because of the price point, everything was direct sell. Extremely expensive account managers that managed accounts directly. You could afford that with the type of margin that, that produced. And then Windows came along, and you saw products like SolidWorks and Autocad emerge. And their price point was an order of magnitude cheaper. So instead of $35,000, it's like $3,500 or $4,000. And obviously, at that price point, you can't afford the sorts of margins you do with a complete direct sell model, as was the case for the Unix kind of world. And so they moved to channel partners because they could obviously gain scale through the channel partner network and the lower margins worked well for that kind of sales model. Now you switch to the cloud, suddenly, things come down by an order of magnitude again, or at least they should. So we've gone from 5,000 down to 500 at the entry point. So again, sales channel changes. So that's where we're very focused on our e-commerce and our telehubs. So we can spin up telehubs exceptionally quickly. E-commerce is highly scalable. And it's exceptionally useful when you're in an environment like we are right now. So direct selling over the phone or through e-comm has been uninhibited by the current pandemic, I would say. In fact, it has actually seen a scaling effect. We've actually exceeded our growth goals for last year, and we expect to exceed them again this year. So I would say that the first thing is what you lose for the individual ARPS on an individual subscriber, you gain in the volume, no doubt. We're also in a highly competitive market with very entrenched competitors. So having an exceptionally disruptive price point is an advantage. I would have been worried if the customers said we were too expensive. So we have everything to play for in this market, and we are quite prepared to go the disruptive route because we think this industry is about 15 years overdue. So that's the kind of basic economics at the low level. Now as you mentioned, we acquired a lot of sophisticated technology from Delcam, netfabb, Moldflow, lots of different organizations we've acquired to build out this cloud strategy. If we folded them all into the base $500 offering, we'd be grossly undervaluing those acquisition assets. So what we've been focused on is our new model, which introduces a concept called extensions. So basically, you subscribe 500, and that's the price point of entry. What that gets you is of course set of capabilities that are multidisciplinary and take you from design to production. Most importantly, it gets you access to be able to read and write to that common cloud data model that I mentioned. And obviously, that carries a lot of benefits with it by itself. But if you're a specific persona, if you're, say, a machinist, what you can do is you can subscribe to an extra layer of capability, which is the machining extension. Now that machining extension doesn't change your user experience. The user experience stays exactly the same. What happens is you subscribe, you turn on extra capabilities inside of the UI that are specific to you as a machinist persona. To gain access to that, you pay another additional $1,600 a year. So combined, you're looking at about $2,100 a year, which is still massively price advantaged compared to some of our competitors, the likes of Mastercam, GibbsCAM, hyperMILL. Most of their technologies are between $8,000 and $16,000. So still a huge price advantage. But above that, and more importantly, you are still contributing data to the same single source of truth in the cloud, so you can collaborate remotely. Now our plan, and we have, is to introduce a range of extensions for different personas. We'll be introducing a design extension for designers. There's one per additive manufacturer. There's one for sheet metal fabrication. There's one for machining. There's even one for generative design, which allows you an all-you-can-eat model for generative. And so these extensions allow us to not compromise Fusion's core values of a single data model with a single user experience in a single brand, but they allow us to target specific personas with increased functionality appropriate to them and only them. So we don't overload everyone with the same capability, and we monetize the value of the technology that we're delivering without compromising the brand promise of Fusion. So that's our extension strategy. Now added to that, we've maintained the unique go-to-market model. So unlike any other product, you want to access one of those extensions. You run Fusion. In the top right-hand corner, you go to the extensions menu. And directly from within the product, you can purchase the machining extension for a day, for a month or for a year. So it's ultimate flexibility, priced at the point of demand directly inside of the product. That's pretty unique. Now we also create an opportunity for partners to deliver services around that. So what's interesting is while we're landing accounts with the 495 merchants through e-commerce and telesales, our partners can be upselling those customers to vertical extensions and delivering the training and implementation that goes around those in the form of services. So it's a good opportunity for a hybrid model and one that we think will increase our ARPS and yield a pretty significant shift in ARR as we move forward into the future. It doesn't compromise anything that we're doing from a next-gen sales, go-to-market model or what we're trying to deliver in terms of brand promise for Fusion. So a bit of a lengthy answer, but hopefully, it covers what you're getting at.

That covers everything. Absolutely. And I think it also shows the depth of analysis and how you -- this isn't a trial and error kind of a situation. This is segmentation. And I guess the most important, like you said, is there's different types of users that will be interested in this type of application. And now everyone will be interested in CAM or everyone will, so why charge 5,000, 7,000 for the whole pack where you can segment them better.

And that's the mistake that traditional vendors make because the software market is rife with organizations that, through a process of progressive development, allow their products to be deployed with unnecessary functionality that none of their customers want it. And when that happens, customers stop paying maintenance. So our view is that we're providing targeted capability at the persona just where they're needed, just when they need it.

Awesome. Steve, you mentioned that Fusion has done really well over the last year or so, and again, expecting that to happen. Again, you've actually started disclosing a little bit of paid user data as well around the earnings calls recently. Andrew has referred to north of 20,000 users -- new users in Q3. Q4, again, you're talking about 140,000 total paying subscribers, which is huge considering the fact where you came from, right? And also, the way I estimate on the proportion of share of the market that's available out there, you're now starting to grab material amount of those new users. What has been the key driver of it? Has it been part of that go-to-market motion but also collaboration? What's been the kind of the killer application for the adoption of Fusion 360?

Yes. I wouldn't say there's any silver bullet. I think that unique go-to-market model and product offering is interesting. I say it's unique. Shamelessly lifted the idea from vendors like Hulu and Amazon. If you think about the way you consume TV, you can subscribe to Hulu. You get a base level of requirement. If you want to watch HBO or NFL or Major League Baseball, you pay an additional amount of subscription every month. Your user interface stays the same. You get access to new content. So these are ideas we've lifted from other industries, and I wouldn't pretend otherwise, but they've been effective in those other industries. So no reason to not copy them. So what I would say is that, coupled with the unique sales channel, coupled with an unprecedented demand for remote collaboration, that's a trifecta right there for solid financial growth over the year. And I don't think that those influences will change as we move forward. I think customers now knew that they were going to move to the cloud, but they thought they would go there in 5 to 10 years. They actually went there in 5 to 12 months. So I think that won't change. Now what it did is it overcame some of the resistance factors. People mistrust in cloud. Does it really deliver value? I think people have experienced all of the benefits and none of the downside. And I think people will be very cautious that they don't expose their business to the same level of risk again in the future. So I expect those factors to maintain. I'll also add that we have -- or had the incorrect perception that Fusion was a maker tool, which is not at all the case. So we deliberately wanted to work with the maker community for a couple of reasons. One is an exceptionally scalable and capable marketing organization when you think about it. Every maker -- and our Facebook group has like, I don't know, in excess of 60,000 people in it, and there are many different Facebook groups around the world. That community does a far better job of marketing the benefits of Fusion than we could ever do. So for us to give them access to free capability in exchange seems like a fair trade, one that many of our competitors are unwilling to make. We are committed to continuing our work with makers because we see it as a mutually beneficial arrangement. They offer a great degree of marketing and promotion, and in return, we offer precedented value in the technology. We also did the same thing with the educational community, which was to make sure it was free for students and educators and make sure that the value proposition -- the principles they were trying to teach. Now both of them are calculated investments to ensure that we get a much higher degree of awareness and market preference. Many of those makers are actually professionals that are using other competitive products in their workspace, at work professionally. They maybe use Fusion at home. And they use it with their children, to teach them engineering. Eventually -- and we've got many forms of evidence at this. Eventually, those users take Fusion into their workplace professionally. So as the products matured over the last 3 or 4 years, we've seen that happen at faster and faster rate. So I think that has also led to some of the growth that you see. There's no one silver bullet. I think it's many of those factors together, new sales and go-to-market model, unique product offering with flexibility through a unique subscription and extensions model, the investment we made in the maker community, the investment we made in the education community and then just the value of a SaaS-based platform versus an individual product because like I said, if all you do in the world of SaaS is just deliver a point solution like it was on the desktop, you're not really solving a problem for anyone that wasn't already solved.

Okay. A little bit of collaboration maybe, but that's about it, right?

That's about it, yes. And to be fair, listen, if you've got a SolidWorks and you want collaboration, you can work with Dropbox. It's not that stronger differentiator, right? The differentiation comes when the whole organization contributes to enriching the same data model that everyone benefits from. And that is a core principle of Salesforce, of Workday, just about any successful SaaS application you can think of.

When I think about -- you've had very successful -- obviously, not market-leading but very successful...

Not yet.

Program with Inventor as well in the past. And that's been really popular in education. One, is there a risk of cannibalization as Fusion becomes really, really big, kind of taking that away, but then you say, "Oh, but the SolidWorks and the rest of the market is actually such a big -- much bigger opportunity that outweighs that risk at the same time"?

That's classic innovators' -- it is a risk, yes, disruption.

Okay. And then are you seeing any -- considering Inventor has been such a popular product, especially within the universities, is that now shifting towards Fusion 360? Because it does so much more, and people are kind of starting to incrementally more being brought up on...

In the universities, there's a little shift. So I'd start by saying any organization that isn't willing to disrupt itself is going to get disrupted by somebody else. So it's a -- there's no point in debating it because if you want to stay alive and meaningful, you're going to have to do that. So we made that decision long ago. What I will say is you can do it in a way that best serves the customer because really what you're trying to do is you're trying to maintain and grow customer base. If you disrupt your own products, that's fine, but do it in a way that's measured and effective. So that extension framework that I mentioned, not only does it deliver unique value to customers. It also gives us a vehicle to help customers start working in Fusion without losing ARR. So for example, machining extension, combine the base 500 with the machine extension, you get to $2,100 a year. That is an entitlement to our PowerMill customers, which is the old historical stand-alone desktop product. That historical stand-alone desktop product is still incredibly valuable. It still serves a large community of users in a very meaningful way. But those users now get an entitlement to Fusion with its machining extension. So they have an extra benefit that would create a collaboration and design projects and fixtures and everything else. So more benefit to the customer, same value to Autodesk in terms of ARR. Now eventually, let's suppose, years and years and years and years out in the future, customers start to use Fusion more as the primary solution. Basically, you've just offered them a vehicle to be able to move to the platform without disrupting your own business. So that extensions framework is quite an elegant way of being able to maintain a user base through a period of change without disrupting them and actually offering and adding additional value to them every step of the way, without charging them more and still maintaining your own ARR. So I think that offers us some mitigation against the potential disruption. But make no mistake about it. If disruption is required, we will do it. Because there's no other alternative, if you want to remain effective and meaningful in the market.

That's -- I think that's a great end to our time, which has just run out. And Steve, again, I'd like to thank you for your time today. I think that had been extremely informative, and I hope investors enjoyed it as much as I did, and really looking forward to hosting it at more events in the future as well. Thank you.

Yes. Likewise.