Ladies and gentlemen, please welcome Autodesk CEO, Carl Bass.
Good morning, everybody. Good morning, everyone back there. A couple of things we wanted to do today, and really two in particular. First thing we wanted to do is discuss the business model transition. We're going to spend a lot of time on how the business is transitioning. One of the things we candidly noticed was just the divergence, not only of opinion, but of models. We want to get back to arguing about opinions, not about facts. Hopefully you'll find today that we over-disclose, and you get more than enough information, at least about the way that we're thinking about the transition, where we are in the transition, what it means over both the short term and the long term. We'll spend the majority of the time, we'll be talking about that. There's a device.
The second theme that we really wanted to talk about was the expanding opportunity. As you know, the two biggest opportunities we have are in manufacturing and construction. What we really wanted to talk about was the effect that cloud-based engineering software is going to have. If you look at the rest of the enterprise, cloud-based software is taking over. It's taking over at an accelerating rate. The part of the market that remains is really in engineering software. We made a big bet starting about three-plus years ago into moving software to the cloud, particularly for these two markets. We are bigger believers today than we were then about the importance and how it expands the market.
We're going to spend a lot of time, Amar in particular, talking about the expanded opportunity that comes about not only because of the new business model, but primarily because of what's going on in these two markets. One of the things that's interesting is, I've been to a number of engineering colleges and universities around the country in the last couple of months, mostly because I've been looking with my younger son at schools. One of the things I realized, I was giving a talk yesterday at a university and thinking about today and how there seems to be a lack of understanding about the importance of what the cloud will mean, the lack of new technology for doing it. All of a sudden, I realized I was with a group of about 200 students that looked like this yesterday, all engineering students.
I realized that the software that was being taught, the software that was being used, the software that you guys cover, whether that's for documentation or 3D modeling, for visualization, for simulation, for collaboration, all of the software was written before these guys were actually born. Think about this. In what other market is the software that is the predominant software written before they were born? Just a way to illustrate how clear it is to me that cloud software is going to completely change, that there's this huge willingness for people to use modern tools. When you look at the materials and the processes, the change in hardware and software, and the connected devices that people are building, engineering software is totally moving to the cloud.
One of the big things that I hope you walk away with today is an understanding of what we saw three years ago, what we've been investing in, why we think it's so important for this future generation, and how it's going to drive financial results in the future. The agenda for today, we're going to start with Andrew, who's going to talk a lot about the programmatic aspects and then the financial implications of what it means to move to a subscription-based model. You heard on the last call how we talked about how we're accelerating that change and the things we're doing to move that more quickly. Amar's going to talk about the expanding opportunity. Steve's going to cover all the issues surrounding our various sales channels.
All right. I'm going to start where we started moving forward. Two years ago, FY 2014, we came up here, and we started to tell you we were going to begin the transition to a subscription-based business. The thing we wanted you to pay attention to back then was the billing throughout the whole transition, and look at the billings as we go through the transition. We told you there was going to be a 12% CAGR from FY 2014 to FY 2018. The fundamental things that we're going to be driving the billing, and by the way, this doesn't change, and I'll get into that in a little while, is more value for each account and more subscriptions. Basic. More value for each account, more subscriptions.
We gave you absolute growth numbers of 20% growth in the value of the account over the timeframe and 50% growth in the subscription base over the timeframe. I know a lot of you are expecting some of these things to change today. Yes, they will be changing today. Before I go into that, what I want to do is go through how have we been performing relative to the model we rolled out two years ago. Let's start looking at some of those stats. Since we rolled out the model, this is from the first quarter of FY 2015 to the second quarter of this fiscal year, we've averaged 70% constant currency billings growth. Okay? Well above the 12% we were targeting in the model.
Many of you who've been looking at us and tracking our subscription growth, you've noticed that the subscription growth we've been delivering is well above the 50% target we were trying to deliver out in FY 2018. That's going to continue. That's continuing this year. It's going to continue out in outlying years. We're already performing well above the trends that we originally rolled out to you guys. If you look at where that subscription growth is coming from, it's coming from all aspects of the business that we said it would. We've seen growth in maintenance. We saw growth in the new term model, the subscription model. We saw growth in the cloud subscriptions as well. On every one of those measures, we're doing exactly what we said we were going to do.
Something's very important about that model we rolled out in FY 2014, and I'm going to remind everybody about it. That was a hybrid model. The assumption from FY 2014 to FY 2018 was that we were going to continue to sell both perpetual licenses and term licenses moving forward. That's a pretty important distinction because as you know, last year we started talking about moving to an entirely subscription model. As of the middle of next year, fiscal FY 2017, 18 months earlier than what we had in that original 12, 20, 50 model, we will be an entirely subscription-based company. What that means for the conversation we're having is that billings are no longer the measure of success as we move through the transition. This is not the thing we want you to pay attention to as we move through the transition.
Subscriptions will continue to be one of the measures of success, I want to start talking about a model that's probably a lot more familiar to some of you. It's a model that Adobe and companies like Intuit have looked at, it's really the kinds of things we want you to pay attention to. It's pretty simple. It's based on annualized recurring revenue. Now that we're moving to a fully subscription-based model, we want you to pay attention to the dynamics around ARR and how it's driven by subscription. I'm going to give you some new numbers to pay attention to and some new time frames. Then I'm going to talk about which one of these numbers are important and remind you again where the growth is coming from in each one of the numbers that drive those.
Here's the new set of numbers I want you to pay attention to, 24, 3, 20. Let's just make sure we dissect these a little bit because I want to make sure that it's perfectly clear what we're looking at. All of these are CAGRs now. These are compounded annual growth rates over the period from FY 2016, this fiscal year, to FY 2020. What you're seeing is a 24% compound annual growth in ARR, 20% compounded annual growth in subscriptions. I know a lot of you are looking at that 3% growth in ARPS or annualized revenue per subscription. I'm going to pause and talk about that a little bit, because what we're going to see is a dynamic no different than what Adobe or Intuit saw in their business.
To help you understand that, let's just take a look at the mix a little bit. Today, FY 2016, this is a representation of our current subscription mix. I notice this goes from low ARPS to high ARPS. You might notice a few things. Cloud subscriptions are over at low ARPS, and one of the reasons why they're over in low ARPS, we have a very high volume of subscriptions that are only about $30 a year. Whereas we also have a moderate volume of subscriptions that are near $1,000 a year. It kind of drives it down because of the volume. This is what the mix looks like roughly today. As we move out to FY 2020, the mix looks much different. Right?
You see a shift and more of a distribution among the types of ARPS that we're going to be delivering from the various subscriptions. The mix drives the CAGR there, and the ARPS is not the important number to pay attention to during the transition. Like I said, this is no different than what Adobe's seen and what Intuit's seen. They've seen kind of this wobbling around inflationary-type increases in the ARPS. What I really want you to pay attention to is this idea of the 20% driving the 24%. The 20% growth, compound annual growth and subscription, driving that 24% growth in ARR. Those are the big, important numbers. These will increase monotonically throughout this transition. If you want to measure how we're succeeding, if these are increasing, we're heading in the right direction. If they're not, you can come and talk to us. Okay?
All right. Now let's kind of drill in again where the mechanics of the growth are coming from. There's two buckets, more new subscriptions and more value per account. I'm going to recap a lot of things I've said before, but we're going to give you a little bit more detail on a couple of things, and you'll get a lot more detail as the day progresses. Let's talk about where more new subscriptions are coming from. First off, the non-subscriber base. For those of you who haven't been here in the past, let me recap what this number is. This is what we consider the active non-subscriber base. We actually have a much larger non-subscriber base, but this number tends to stay relatively constant. It goes back 5 years and includes the current non-subscribers that bought this year.
Those of you who have been here many years, you've noticed the number changed a little bit. That's a good thing because it's showing some progress we've made, and I'll talk about that a little bit. It used to be 2.9 million. It's 2.8 million. There's another thing I want to correct some people because I've read some of your reports, and I think you're operating from an assumption that may not be valid. We don't assume to hit our targets that we have to convert this entire non-subscriber base. In fact, we only assume a 30% penetration into this space over the period that I'm talking about today in this presentation. Okay, a 30% penetration. It's important for everybody to keep that in mind. Now another important source, project-based and non-paying users.
These are lumped together purposely because project users tend to be inadvertent pirates in many ways, either through trials or other types of things. The global piracy rate is still super high, 43%. The new business models let us reach into the space. We're already seeing it in our media and entertainment business in big ways. We're seeing a lot of people who were trialers turning into buyers. Some of our immediate entertainment products were the most trialed, but least bought. We're starting to see a shift in that. We're actually starting to capture these non-paying users. The other incredibly important section, or incredibly important target for subscription growth, and you're going to hear a lot of this from Amar, is the market expansion. I want to give you a sense for what the opportunity is here.
In the construction space, there's 100 million potential subscribers that are looking for IT solutions to their construction management problems, and the cloud is the only way they can get those solutions. In product development and manufacturing, there's an 18 million-plus subscriber base that's opening up because of the transition to the cloud. You'll get a lot more specifics about this from Amar, but it's important to recognize that there are markets that we are going to penetrate and grow into fairly aggressively over this timeframe, and it's an important part of the subscription growth. When we look at the more value per account, the cloud services are already getting attached to our accounts in a meaningful way. Customers like the value of them. We're not forcing them on them. The customers are buying into the cloud services because they solve a problem for them.
BIM 360, PLM 360, cloud consumption for things like rendering. The customers are buying into these, and it's increasing the value we get from each account. They're buying in because they want the services we're offering because they add value. The other thing that's important, after the end of sale of perpetuals, middle of next year, when a customer adds a new seat, the new seat's going to be a pure subscription seat. Those are at higher annualized recurring revenue than the current maintenance model. You're going to see every new seat going to be a subscription seat, which is going to drive up value per account on an ongoing basis going forward. The other dynamic that I've talked to a lot of you about is we're going to start to see migration of the maintenance base.
I want to be crystal clear about a few things because we get these questions a lot. We are not going to force our customers off of maintenance. Customers that pay maintenance can pay maintenance as long as they want. I'll try to help you understand that they are going to want to migrate off of maintenance as we move through this transition, and we're already seeing customers migrating off of maintenance with some of the offerings. I'll try to give you a sense for why a customer would choose to do that in this presentation. We're not going to force them. I think that's important for everybody to be on the same page about.
I want to give you some proof points to help you understand some of the things that are driving the growth in the model, and I'm going to give them to you in 3 categories. These are the basic things from our existing business. Talk about the current LT family performance. I'm going to look at just high level, the opportunity for AutoCAD, and drill in a little bit to what it's going to look like as we migrate the maintenance base. Those of you who were here last year, you heard pretty clearly this year we're going to spend a lot of energy focused on the LT base, trying to move them to pure term. We've been accelerating that in various places, where we've removed perpetual rights in sub-geographies. We're moving it more in some other sub-geographies right now. We move to the rest of the portfolio.
I want to give you a sense for what's the impact been of those efforts to date. Here they are. This is the impact. We've seen 35% growth in the annualized recurring revenue generated by the LT base, well above what I'm talking about in terms of the 24% target. We've seen a 27% growth in the subscriptions. We did see some growth in ARPS, but we grayed it out because we don't want to focus there because of the mix, right? Where the mix is going to come from. What you see here is the performance to date is well above the targets I'm telling you, and there's still a lot of room to grow. 40% of those non-subscribers are still LTs, and we're actively converting them right now, and we're moving towards full termination of perpetual LT sales.
You're going to continue to see this performance in the LT base, and you're going to continue to see growth. That's one proof point in terms of giving you confidence in where the model's going and what we're doing. Now let's just talk about AutoCAD. To date, we've done nothing with the AutoCAD base. If you add up AutoCAD and AutoCAD family, meaning all the AutoCAD verticals, this is the next largest non-subscriber base. If we just converted 20% of that base, remember, our models are targeting 30% conversion of the non-subscriber base. Let's just say if we converted 20% of that base, what happens to that chunk of our business? Pretty simple math. Anybody can do it. 45% growth in ARR, 21% growth in subscriptions. Again, above the targets I'm giving you on a 20% target for migrating the base.
This is our next opportunity for non-subscribers. This is where we're going to focus next, and it's how we're going to move forward in various places. Okay, another proof point. Now let's talk about migrating the maintenance base and why some of these customers might move, and why some of the customers that have already moved today are moving. First off, let's look at the offering or the model that's out in front of some of these customers. We've tried to make this slide as messy as possible. Reality is actually messier. Our customers that buy in the current state today have a confusing potpourri of products that have different models, that are managed differently, they're deployed differently, they offer different entitlements, different rights. The suites have been a great simplification, but there's 7 types of suites with 3 types of tiers.
You do the math, how many permutations. Our customers deal with a fairly messy environment. They don't always know what's right for them. They don't know what to buy. Then when they have to manage it and decide how to get it to individual users, it gets really complicated. This kind of complexity is an opportunity in terms of helping customers see a better way to move forward. As we move into next year, the middle of next year and beyond, the way we present ourselves to customers is going to change dramatically. We're going to be really just talking about Autodesk Subscription. That subscription's going to come. Various ways to buy.
There's going to be an enterprise way to buy, a multi-user way to buy, so if somebody wants to buy a subscription and let multiple users access it, and a single-user way to buy it, named user type models like we have today. When it comes to what these products open up, today, the enterprise offering is T-FLEX . It's all product, it's consumption-based. That's going to continue, it's going to move forward, it's going to get better. As you move down, the customers are going to have simple options. You can buy individual products on subscription, or you can buy what we'll call collections. The number of collections is going to be vastly reduced from this week. As a matter of fact, probably just one per industry. A completely simpler set of products that we'll be offering to the customer.
Same goes as you move to single users. Individuals, collections of products. Easier to understand, easier to manage, no hybrid environment, more access to products, and a better experience. You'll see a lot of proof of this as we move forward over the years. Here's something. You don't have to believe this, because we're already seeing it at the highest end of our market. Let's look at these. Oh, yes. Basically, there's just a lot more incentive to move. Now I'm going to send it over to you. At the highest end of our market, since Q2 of last year, 22 of our largest customers, these are big customers, very large contracts with us. 22 of our customers have given up their maintenance rights to move to the enterprise offering, Token Flex offering. They just said, "Okay, fine, I'm moving off of maintenance.
This is a better way for me to buy. It's a more effective way for me to engage in Autodesk. That total enterprise space has seen a 90% increase in annualized recurring revenue delivered to the company. Customers actually gave up their maintenance rights to move to something that was completely different moving forward. You're going to see the same dynamics moving downmarket. Let me tell you what that might look like as we move downmarket. I'm going to give you some numbers. As you move downmarket, you start looking at these multi-user collections. A customer moves off of maintenance to a multi-user collection. What you're going to see is a 1.25x-1.5x increase in the annualized recurring revenue from that account. Why the range?
As customers move from maintenance to a collection, they might change their seats. They might buy less seats, they might consolidate. They'll do some things like this, but they will deliver still at ARR levels well above what we're giving you as the goals for the business overall. Same goes as you move downmarket. If a customer moves from a maintenance type environment to a single user subscription on a collection, you're going to see a 1.5x to 2x increase in the annualized recurring revenue from that account. There's lots of opportunity here. We don't expect a lot of people migrating off of maintenance next year. What we expect is, as we move into FY 2018 and beyond, the customers are going to start to understand, now I see what Autodesk is doing. Now I see the offering they put together.
I see how much more flexible it is, how much more access they're giving me, and how much better the experience is in the new world versus the world I exist in today. This is what's going to create the incentive to move. We're confident in this, we believe in this because we saw it at the highest parts of our market. We're taking the ideas that translate down to the volume parts of our market and applying them downmarket. There's lots of opportunity here, and you're going to see it start to evolve as we move into FY 2018 and beyond. I want to recap a few things because I want to leave you with some important things as we head into the rest of the presentation. What are the fundamental business drivers as we move to the full subscription transition?
More new subscriptions, more value per account. Fundamental business drivers, okay? We're moving away from the billing and space measures of success. We want you to look at annualized recurring revenue and subscription over the time frame from FY 2016 to FY 2020. The whole 24, 3, 20 model, but the numbers we really want you to pay attention to is the 20% driving the 24%. You see those numbers going up? You see those numbers continuously going up? You know we're creating the value we say we're going to create. You know we're succeeding through the transition. Like I said, if you see the numbers going the other way, we can have a different conversation. Where does the growth come from? New cloud services on top of the accounts. The value is there. Customers are buying them. They want to buy them. You'll also hear a lot of new customers are buying them.
Moving forward, as of the middle of next year, all new fees are subscription fees. As we move into FY 2018 and beyond, we're going to start migrating maintenance customers and seeing them move to the new types of offerings. In terms of the new subscriptions, one of the things that's really important here is the impact of the cloud on these new subscriptions. We're going to convert the non-subscribers. You saw it's $2.8 million now, $2.9 million last year. We only have to convert 30% of them to hit our goals. Project-based and non-paying users, we're already starting to see that trend rising up. The big change here is new subscription targets. The cloud changes everything.
The new business models, the cloud-based business models let us reach the non-subscribers, they let us reach the non-paying users. The cloud is the way we are going to unlock this market expansion opportunity. What I want to do as a segue now is go over to Amar Hanspal so he can tell you exactly how the cloud unlocks the market expansion. Thank you very much.
Amar.
Thank you, Andrew. Good morning, everyone. Andrew made a compelling case as to how a shift to the subscription-based business model helps us build a better business. Now, I'm here to tell you how the shift to the cloud helps us build a bigger business. You just have to go back in time to see the correlation between technology transition and software opportunities that are created by those technology transitions. The birth of enterprise software in the era of mainframe computing, companies when the PC arrived, and not just large companies, but SMB companies joined the bandwagon of accessing and using software as it became easier to use and deploy. Certainly, with the cloud and mobile devices, software has become way easier for companies of any size to consume, and in fact, even for individuals to use in their daily lives.
The cloud for Autodesk is more than access to a larger customer base. We look at the cloud as a fundamental shift in providing us the ability to build the next generation of design and engineering software. It is about reimagining what is possible. It is about re-looking at manufacturing, re-looking at construction, and looking at an unlocking or cracking open those huge opportunities with subscribers and TAM. $100 million in construction, easily north of $18 million in manufacturing. It's really by looking at the cloud as a fundamentally different way of doing things that we're going to be able to crack those opportunities open. The time is right to do that. This is all about new markets and new subscribers for Autodesk. The time is right to do that because if you look at any enterprise function, they have all gone through using cloud solutions.
Whether you look at finance or HR or sales or service, every major function in the enterprise today has moved to the cloud. In doing so, they're not just using the same thing they used to use on a desktop, but they've been able to completely reimagine their business processes. Well, design and engineering is going to be no different. The time has come. In fact, I think we're well into the shift of design and engineering moving to a cloud-based environment. By doing so, not just realizing the dramatic gains of moving to the cloud, but letting customers fundamentally reimagine the way products are made, the way buildings are made, the way infrastructure is made. This is what we're building at Autodesk with our cloud.
We believe, we are clear in our minds, that someday soon, nearly all of the world's buildings, products, infrastructure are going to be designed, engineered, visualized, simulated, validated, managed, made, built, monitored, and operated through the cloud. We have no doubt about it. We have been building it, as Carl said, for at least three years and maybe more. By doing this, we are fundamentally reimagining the manufacturing software opportunity, as well as creating an entirely new opportunity in construction. I'm going to talk to you about that. We believe that this is the way our customers are heading because the cloud enables them to do something that is fundamentally not possible today. The cloud is not about delivering software differently, it's about doing things differently.
For example, our customers are now able, with the cloud, to apply virtually infinite computing capabilities or resources to any problem. Not possible without the cloud. They're able to coordinate their processes across companies, across all of the individuals necessary to make products or buildings come to life in these inherently fragmented industries. Again, not possible without the cloud. They're able to connect to context. They're able to connect to devices, to data, to means of manufacturing, and make better decisions upfront. Again, not possible without the cloud. To illustrate this, let's think about just the first one of these, just the raw compute power of the cloud. You and I experience this on a daily basis when we Google something, when we're looking for an answer to a question or an inquiry in our head. Engineers are no different.
They're looking for answers all the time to way more complex problems. In the case of this 3D printer design, they're trying to figure out how they're going to manage the thermal density of the insides of this 3D printer. How are they going to lay out these components so that they don't have part failure later on and don't have to create a large envelope just to accommodate those parts? The answers come through those virtually infinite compute capabilities of the cloud and by applying CFD and finding out and making those trade-offs very early on in the design process. Not really possible without the cloud. You go to the nature of our industries, which are fundamentally fragmented. The manufacturers work across supply chains. They sometimes involve their customers in customizing products. Construction is inherently fragmented.
There's hundreds of subs, hundreds of sites that they work, dozens of sites that they're working on, or material suppliers and owners and regulators. These industries are fundamentally fragmented, and they have to work across these companies to realize their objectives or outcomes. It is only through the cloud that they're able to organize these companies into really coherent teams, share information in a structured way, and actually execute processes. They're able to connect for the first time only through the cloud, design to make to use. Or use AEC terms, design to build to operate. That's the only way they're able to execute this across companies. Then I touched on this really briefly earlier. For the first time, manufacturing companies have a full digital pipeline. They have access to the means of manufacturing.
What I mean by that is, the traditional subtractive manufacturing methodologies, CNC machines have become increasingly popular, way more capable, and way more prevalent in manufacturing lines. Additive manufacturing has shown up, or 3D printing, as it's popularly known, and in fact, being applied increasingly in industrial scenarios. The way to drive both additive and CNC technologies is through digital means. The fact that these digital means or digital manufacturing capabilities are visible and customers can connect to them through the cloud means they can design dramatically differently for the first time in many years. That goes one step forward because not only can they connect to the means of making, but through smart devices or this thing we call the Internet of Things, they can connect to the methods of use and watch how their products actually function in the operational sense.
They can start to think about delivering products differently. You already hear GE talking about delivering flight hours rather than engines. This is a major trend that's taking root in both manufacturing and the built environment. The cloud enables our customers to do things differently that were not possible before, connect design to make to use, and have a consistent, coherent approach throughout the process. That lets us go and attack those two large opportunities that Carl and Andrew touched upon. Over 18 million subscribers in manufacturing, over 100 million subscribers in construction. Perhaps the way to help you digest that is to break it down sort of sector by sector, and I'll begin again in the world of simulation, because it's relatively easy to understand how a shift to the cloud creates a huge opportunity in simulation. Simulation's been around for 40 years.
Simulation software has been around for at least 30, 35 years. It's a well-understood market. There's at least a $3 billion TAM. The thing is, after 30 years of simulation software being available, less than 10% of the engineers use it. Why? Simply put, it is extremely hard to use, it's expensive to own, and the hardware and specialized knowledge required to use it can be afforded only by a few companies, the largest of the large companies. That means it's perfectly set up for disruption by the cloud, because the elastic compute power of the cloud that I talked about earlier really lets us address that whole price-performance issue, which changes who can use or how expensive it is to use simulation.
The ease of use and integration into the design pipeline lets us make it accessible to more and more design and engineering customers than ever before. This, really, the shift to the cloud unlocks a huge mid-market opportunity that says go after the 90% of the engineers that don't use simulation software. The case for using simulation is really obvious. It is to make a better decision sooner, and every engineering company would use it if it was more affordable, if it was more accessible, if they could use a simulation run on a Tuesday and a Friday without having to commit to years and years of buying simulation software, right? The cloud changes the economics. We have a cloud-based simulation stack for mechanical, for thermal, for fluid, for plastic, for structure, for energy, for all these various capabilities that people need to answer those questions.
What we've seen is in the past year alone, the number of jobs run on our cloud has easily doubled, and it's poised to do it again. We are very bullish that simulation is an opportunity for us to disrupt incumbents in that space and democratize an entire market. Speaking of disruption and democratization, what better market to think about that than PLM? You know this. Every manufacturing company, every product design and engineering company needs a PLM system. Yet only a few can afford it. Because the economics of owning and managing a traditional PLM system from, you name it, from Oracle, from Dassault, from Siemens, from PTC, they are absolutely prohibitive. This market reminds me of exactly where the CRM market was when Siebel was the incumbent, and then Salesforce showed up and changed the economics of owning, managing, and deploying a CRM system.
That's what we're doing with PLM 360, is changing the economics of owning, deploying, and managing a PLM system. PLM 360 is a modern configurable cloud-based system, and between its ease of use, it's the best-looking PLM system on the planet. It's instant on configurability. We're hearing from our customers that they find it to be at least three to five times more affordable than one of the traditional big iron PLM systems. This opens up a huge mid-market opportunity. For us, it opens up subscribers in the supply chain. As Steve will share with you in a minute, over half the people that are showing up on our or nearly half of the people that are showing up using our PLM system are new to Autodesk. They are people who are using competitive systems or who are using PLM systems for the first time.
We're really gaining momentum here. We're getting customers, and we're gaining customers because it's delivering real results. If you want to experience this for yourself, we have this annual conference in the fall called Accelerate. We just held it in Boston in September. There was easily over 300 customers who were using PLM customers. The amazing thing was we didn't have to say a single word. They were the ones out there talking about their experiences, referencing how they were using. The question was not if PLM, how PLM, but how much PLM that they were dealing with, right? They were really trying to go beyond their first deployment to the broader corporation that they were dealing with.
We feel very bullish about how PLM is turning out, disrupting and making a market for us, gaining new subscribers, gaining new accounts, adding value to accounts where we are already present. We are disrupting in simulation, disrupting in PLM through the cloud. The one other place that is ripe for disruption because it's more than the cloud that's at work, is the way things are made. I touched upon this earlier. Subtractive manufacturing going to CNC, additive manufacturing being digital from the get-go. Carl mentioned that, in fact, new materials are showing up, which means that people really have to think about how they're going to make products differently going forward involving all these advanced materials. That's creating a huge software opportunity for us. Why? Because for the first thing, we have more CNC shops.
We have all these additive manufacturing fabrication facilities that can be really new subscribers to Autodesk because we build software for additive manufacturing, we build software to drive CNC machines. It also creates demand for design software. In fact, our competitors' customers are turning to us for design software solutions. Why? I'll oversimplify this a little bit. Many years ago, when laser printers showed up, it created a demand. As capable as laser printers were, it created a demand for desktop publishing software because something needed to describe the page in the level of detail and accuracy that laser printers were capable of. It's the same thing is going on in digital manufacturing. As these machines become more capable, it's only creating the need for design software that's capable of describing what can be made at a higher level of accuracy, specificity, and capability.
There is where we're finding that customers are turning to us because they can't get solutions from their existing vendors to design differently. A dramatic example of that is what's going on with generative design. That's a heat exchanger. A heat exchanger looks like this because it's been cloud-optimized with the knowledge of the means of making. Additive manufacturing capabilities optimized with the least amount of material and the greatest amount of effectiveness have given the designer this option. Then the designer can iterate through to getting the answer that they're looking for. Here's an example of designing differently with the cloud at the center and with the knowledge of the means of making.
This is the kind of thing that our customers look at and say, "That's my next generation of products, and I need to be able to go to somebody that has a different design stack than the one that I'm using." This is creating opportunity for us to take competitive share, as well as go to fab shops or CNC shops and add subscribers to our mix. Let's touch upon IoT for just a minute. You know, IoT today is a new pixie dust term. People sprinkle it on everything. Sounds like a huge opportunity. For our customers, it's a very practical opportunity. Many of our customers are looking to shift business models. Just they're trying to do their equivalent of flight hours. They're interested in making smarter products, smarter machines, smarter factories that have higher uptime and greater reliability.
They need infrastructure in order to be able to do that. That's a perfect problem for the cloud. How do you connect to all these data and devices? You connect to them through the cloud. This is what we did with our recent acquisition of an IoT platform, is we've added that layer to our cloud infrastructure so we can collect data, analyze data, and let customers build applications for preventive maintenance, for smarter system design. When you're looking at not just a single part, but you're trying to optimize the entire experience like a smarter building or a smarter data center. You get all of the components necessary to do that. What you're trying to do is create a feedback loop between design and use so that you can improve your designs informed by what's going on in the actual case.
IoT is a natural extension of our cloud strategy. The thing that we don't get about what our competitors are doing is they're doing a non-cloud approach to try and get what they call service, like SLM or whatever. I don't know how you connect to devices from behind the cloud in a reliable and consistent way. That's what we're doing, the opposite. We're leveraging our cloud infrastructure to build in extensions, and again, connect design to make, to use, and expand our opportunity, get to new subscribers in the service area, but also go back and help. Just like the way advanced manufacturing lets people design differently, the connection to smart devices, again, lets customers design differently. Again, we believe this will give us a leg up against our competitors, even in the core design space. We've been touching on core design all this while.
What does the cloud do? We've talked about PLM, we've talked about simulation, advanced manufacturing, how things are operated and maintained. What does the cloud do for core product design and engineering in the world of manufacturing? Well, it lets you fundamentally move beyond the modeling paradigm that's been around for 30 years and connect everything from form to fabrication. That's what we're doing with Fusion 360. I want to be clear, calling Fusion 360 the next generation CAD product is doing it injustice because it really does industrial design, mechanical design, in-built validation and simulation, in-built CNC, additive manufacturing. It's really the stack that a company can use to go from concept to execution. It's sort of like CRM collapsed a bunch of categories into one. Fusion is the way to reimagine the way products are made. Who's using Fusion?
We've seen a huge spike in the number of people using Fusion in recent months. Many of them are SolidWorks, NX, or Creo customers who've kind of grown past the modeling paradigm. Don't want to do just 3D modeling, want to do design differently, design for the next generation. Many of them are new product entrepreneurs, companies that are starting up for the first time. This bears repeating. Hardware is the new software. In other words, there's an increasing investment going into hardware startups. Last year, it grew by 76%. I've lost count of how many hardware startups are showing up in the Bay Area alone, accompanied by incubators, accompanied by manufacturing specialists. There's an intense amount of entrepreneurial activity happening on these hardware startups. First time inventors, entrepreneurs. Just like the students that Carl talked about, those guys are looking for something completely different.
They're not going to start, if you will, on their father's 3D modeling system. They're looking for something that is contemporary, that works for their way of thinking of, there's no real separation between industrial design and the means of manufacturing. It has to be a very tight pipeline, they're looking at something like Fusion to get it done. Again, Steve will share with you that over 80% of the accounts that are using Fusion are new to Autodesk. We're not going to existing customers and giving them a new tool. We're really going to people who are coming off of competitive systems or who are using systems for the first time and showing them Fusion. This is all near-term traction.
Between all those categories that I talked about in manufacturing, simulation, PLM, advanced manufacturing, IoT, then what's going on with Fusion in terms of reimagining design and engineering, we're looking at easily north of 18 million, perhaps over 20 million subscribers. We're getting near-term traction in all those categories, especially with Fusion. Look at the world of construction. That alone worth 100 million subscribers. Why? Because this is a place where we have the perfect storm taking place, which is a burning need with the right platform, with the ideal software solution. Let's go through that. The burning need is, look, construction for however long, ever since the Renaissance, has been stuck in this. They're struggling with the efficiency of their methods and the economic structure of their industry, where risk is distributed across multiple players.
You get into all of these numbers, just one I'll pick out is 40% of the materials that show up on a construction site are wasted. They're thrown away. That is real money. When you look at the opportunity for software automating or eliminating waste, just that one line, even if we did nothing about the other five lines, that one line alone in an industry that is worth $7 trillion of economic output today, is a huge opportunity. Construction employs 200 million people and above worldwide. When we think about the number of subscribers we could get to, the TAM we can get to, construction represents a huge opportunity. It's a burning need. The right platform has arrived, which is the mobile platform. Up until now, I'd say PCs limited the access to all those field workers.
Smartphones, smart tablets, they really have become a form factor that works out on those job sites. That's working. The software stack we've been building, BIM 360, has really been working because it's a very practical solution to the everyday problems construction companies face. Problems like planning. Right upfront, before they want to do any physical work, construction companies want to find out what is going to go wrong. If I install the air conditioning system, is it going to run into the structural element that the structural engineer just handed me? You think that's a conceptual issue? This happens hundreds and hundreds of times on the job site.
Just walk through the lobby downstairs, you watch that mini construction project that's not even that complex, think about how many issues those guys are dealing with on a daily basis because they're trying to orchestrate information and work across all these multiple companies. BIM 360 solves that problem. It has a scheduling component, it's a planning component, there's field execution that lets you get into quality, inspection, equipment handover. All of these things are very practical solutions that let a construction company gain insight into what's going on. It creates an opportunity for us, not just for software, but for big data, to get insights into the execution of the construction projects and how to do it better. That's why we're so bullish on the opportunity in construction.
Between what we're doing with building information modeling and what we're doing in construction, we're watching customers, again, just like the PLM case, see real benefit, reduced risk, greater safety, economic benefits. This is really playing out in a big way. BIM 360 is growing 70% in terms of subscriptions year-over-year. It's going through the roof. We're doing really well and are off to a good start in terms of what we're doing in construction. Now, the one thing I haven't been able to convey to you through the slides is what's going on in the perception of Autodesk. I talked to you about each of the opportunities, large subscription opportunity in manufacturing and construction, and the early traction we're getting.
The one thing I can't emphasize enough is our bet on, or our belief actually, more clearly, in the cloud letting us reimagine the way products are designed and made, and the way buildings are designed and made. It's really resonating with customers. We're talking to the who's who in manufacturing and AEC. Companies that used to think of us as, "Oh, yeah, those guys make AutoCAD," they're coming to us, visiting the pier, talking to us about how they're going to remake their product development pipeline. The conversations have changed dramatically, we really are talking to the absolute top of the top. Many of them are our competitors' customers, they no longer see them as the effective solution providers. This is something that is really starting to change in the industries we serve. I'd like to leave you with three final thoughts.
The first one is, the opportunity that we're looking at at Autodesk is very, very large. The shift of the platform to the cloud really lets us go after that larger manufacturing opportunity, 18 million+ subscribers, by disrupting existing markets like PLM and simulation and creating entirely new ones in advanced manufacturing and IoT or reimagining the existing design and engineering stack. It lets us go after construction, 100 million subscribers, and the software market that's being created because the cloud really lets us address the fundamental issues of that industry by connecting all the right players. We think the TAM is at least $25 billion. It's, again, 100 million subscribers, 80 million subscribers. Very large opportunity. Second thing I want to convey to you is this is not some long term, on the other side, there'll be this great, amazing nirvana. We are experiencing this now.
We're seeing real traction with BIM 360 Construction, Fusion 360 manufacturing. We have green shoots with PLM, with simulation, with what we're doing on advanced manufacturing. This is real. This is happening now. Finally, the story of Autodesk is much more than the Business Model Transformation. I know you all are paying a lot of attention to that, this story is way bigger than that. Business Model Transformation helps us build a better company. The cloud shift helps us build a bigger company. We're in the early stages of a major reshaping of each of our industries. Each of our customers are reimagining the way they're making products, they're making buildings, they're making infrastructure. That's redefining who the market leader in these industries are. Autodesk and our cloud-based approach, we have established a clear and early lead in the reshaping of these industries.
That's what I wanted to share with you, and I'm sure we've blown your minds this morning, so it's time for a break. What time is it now? 9:30. Dave, are we taking till 10:00? 10:00 o'clock. You have 30 minutes, and then after the break, Steve Blum will come on and speak to you about our sales strategies. Thank you.
Ladies and gentlemen, please take your seats. We're about to begin. Thank you. Ladies and gentlemen, please take your seats. We're about to begin. Thank you. Ladies and gentlemen, please welcome Senior Vice President, Worldwide Sales and Services, Steve Blum.
Good morning. Welcome back. Hopefully, you had a good break. All right. I'm going to share some interesting information with all of you. Again, as Carl mentioned, I'm going to give you some insights into how we're going to market in each of our segments, but I'm also going to provide you with some results, some proof points of how things have been going since we were together last year. Andrew talked about our customer segments, and we basically have three different customer segments. We have enterprise accounts. I call them named accounts very often. You'll hear me use that name interchangeably. We have small and medium-sized businesses, and we have very small businesses. Our go-to-market approach varies for each of those three segments. Now, Andrew showed you how we're reconciling and simplifying the offerings that are appropriate for each of those three customer segments.
What I want to do is show you how we're going to market in each of those segments and how we're gaining traction in driving growth in subscriptions, as well as getting more value from each customer. All right. Start at the top of the pyramid, our enterprise accounts, our largest accounts. As I said, we refer to these as named accounts. For named accounts, we take an Autodesk-led approach, and we're leading the sales motions, the renewals, and adoption motions. We're using our consulting services and enterprise priority support, and we're leading with our Enterprise Business Agreements, as Andrew mentioned. These are consumption-based models that make the entire portfolio available to the users. All users within the company have access to that software overall.
We work closely with those customers to help them adopt and engage and use the software and hopefully win more business as a result. We do have partners helping us in some stages of the process, especially in some adoption-related processes. When they do participate, we pay them service fees since this business is transacted directly. Last year, I showed you some progress we were making as we were moving away from an old enterprise model to the T-Flex consumption-based model. We had about nine customers that had moved. We've now had 24 customers move from the old model to the consumption-based model, and we are getting 3x the number of monthly active users or subscriptions as a result of customers moving to this higher-value, fully accessible offering.
We're very, very pleased about that, and our customers are, and they're taking advantage of more of the offerings and using more of the portfolio as a result of that. Andrew used this slide earlier, and I wanted to reinforce it. Customers moved to T-Flex from our old enterprise model, but they also moved there from the traditional perpetual license and maintenance mode as well. 22 customers since Q2 of FY 2015 have moved over from maintenance to the T-Flex model. Across all of our named accounts, we're seeing significant growth in value from the customers, and our ARR has grown 90% since Q2 of FY 2015. This is exciting to me, and hopefully to you, because the opportunity in our named account space is still significant.
If you're wondering, we've only converted 9% of our customers in the enterprise space in our named account program to T-Flex models. This was at the end of FY 2015. What's interesting to note is that those 9% of customers who have moved actually contributed 25% of the total billing that we've done with named accounts. As you can see, when we move them to the new model, we're delivering a lot more value to them, and we're recognizing more value as well from those customers. Again, we still have a lot of work ahead of us as we'll be moving, my goal, all customers in our named account program to these Enterprise Business Agreements over the next several years. We're going to move down to the small to medium-sized business space.
As a reminder, this is our biggest space from a volume perspective. This is where the most billings and revenue, ARR, subscriptions are going to come from and have been coming from. In this space, our go-to-market is partner-led, and our partners take the lead on the sales motions, the adoption, and the renewal motions. We have programs that help our partners specialize in particular industry and industry subsegments. I have territory sales reps that team up to help drive demand, as well as inside sales resources that are helping us get more reach in driving more customer acquisition overall. Again, this is our partner-led model, and our VARs, our Value-Added Resellers, are the ones that do most of the business in this space.
I know some of you have been questioning our VARs and have been coming up with your point of view on are our VARs endorsing and supporting and excited about the new business model? I'm going to give you some facts. Our VARs, our value-added resellers, are gaining lots of traction with selling Desktop Subscriptions. In fact, this is the last 6 quarters, and you can see Q2, our last quarter that we closed, 21% of all new seats sold by our VARs were Desktop Subscriptions. You can see this trend is moving in the right direction. We have a lot of adoption, and by the way, they're ahead of where my expectation was for them. I didn't think we'd get this far till the very end of the year. In fact, they're even above that in Q2.
Our VARs are leading with Desktop Subscription and providing that choice to our customers. In fact, across all of our partners, not just our VARs, but our volume channel partners, distribution partners, all the partners combined, they're now accounting for more than half of all Desktop Subscriptions sold. That you can see as we cross over into the new fiscal year, 52% in Q1 and 58% of all Desktop Subscriptions sold were sold by a partner. Our partners are absolutely engaged in selling these offerings. That's very exciting for several reasons, and I put together this slide, and it really has two different stories to it, let me explain it to you. What I'm showing you by quarter are the number of customers that purchased new Desktop Subscriptions. The bars actually represent a customer count, customers who purchased new Desktop Subscriptions.
You can see it is growing. It's trending in the right direction. What is really exciting to me is that if you take a look at the first half of FY 2016, over 40% of all of the customers that purchased Desktop Subscriptions were brand-new customers to Autodesk. Desktop Subscription is helping us expand and grow our TAM and helping us with new customer acquisition, and that, of course, bodes very well as we continue to go after the new market opportunities that Amar talked about earlier. Now I'm going to go to the very small businesses. Lots and lots of customers here. They're not very large, many single-seat purchases. As a result, our go-to market is a very low-touch customer service or self-service-oriented eStore and volume channel partner type model. Much of our customer acquisition actually begins through marketing campaigns, marketing-generated campaigns overall.
This is an area we're investing heavily in as far as providing more value through the eStore experience and really enabling our volume channel partners to deliver more value to the large number of customers that have many transactions that they're working through. I wanted to give you an idea of how we're doing as far as gaining traction with the new model here. What you're looking at now is the progression of traction we've been gaining with Desktop Subscription with our volume channel partners. You can see we have made significant strides. In fact, 39% of all new seats sold by our volume channel partners in Q2 were Desktop Subscriptions. They see this as an opportunity for them to grow their business and add more value to customers, and we expect this trend to continue.
This probably isn't a surprise, but I wanted to highlight it. What I'm showing you here are the percentage of new seats sold on our eStores that are Desktop Subscription. Think about this. This is now where a customer is coming on and making his or her own choice by themselves. As you can see, a majority, a significant majority of customers coming into the eStore and making a purchase are purchasing Desktop Subscription. 87% of all new seats sold on our eStores in Q2 were Desktop Subscription. What's very interesting and really pleasing to me is that the volume of business through our volume channels, so it's our eStore and our volume channel partners around the world, has been growing dramatically. In fact, in the first half of this year, total volume in those volume channels was up 68% compared to first half of last year.
There's three other things that I wanted to give you an update on that cross all of the customer segments. They're not specific to any one, but I think they're all very relevant and something that you should be aware of. The first one is going to be how we're doing with our new offerings, I want to talk about what's going on with the channel mix as we move forward because I know that's a topic that comes up very frequently. I want to give you some insights into how we're preparing for the end of perpetual license sales. Amar talked a bit about this. What I'm showing you here are our new offerings, our cloud-based offerings.
Fusion 360, HSMWorks, and PLM 360 are three significant offerings that we have in the manufacturing space, which are enabling us to grab more share and also drive TAM expansion, like Amar talked about earlier. The bars represent the percentage of customers that were brand new to Autodesk. Fusion 360, 86% of all customers buying Fusion 360 in the first half of this year were brand-new logos, brand-new companies for Autodesk. HSMWorks, which is our CAM offering, 57% brand-new customers. PLM 360, almost half, 46% of all customers were brand new to Autodesk. These offerings are helping us to grab share and drive expansion and grab more value as we're delivering more value to our customers. The bottom two are BIM 360 and InfraWorks 360. These are our cloud offerings in the world of construction.
The first takeaway is, it's a little bit smaller, but there's a positive there because we have broad adoption of these offerings within our install base in construction. Even in the world of construction, these new offerings are exposing us to new opportunities and to get to new users like never before. We are seeing 8% of customers in both of those offerings that are brand-new customers to Autodesk. I'm going to talk about the channel mix shift, a topic that I know comes up frequently. These are not totally accurate. You don't have to measure them and try to figure it out to the percentile, but they are visually correct as far as showing that between FY 2015 and FY 2020, our direct business is going to grow faster than our indirect business.
And the but is very important, both our direct and indirect business is growing. In fact, the total volume of business that will be indirect in FY 2020 is larger than the total volume of indirect business in FY 2015. Both grow. Direct is growing faster than indirect, but both will be much larger in FY 2020 than they are in FY 2015. Of course, I want to take this as an opportunity to reinforce our partners have and continue to play a very important and strategic role in our go-to market, and they will continue to have that very important role out into the future. Okay, the last topic is how are we preparing for the end-of-sale of perpetual licenses. I wanted to give you some insights into the first one that we did, because we did have a pilot of sorts. Andrew mentioned it earlier.
We ended the sale of perpetual LT in Australia and New Zealand during Q2. We took a contained market, and we ran through that process. Two data points that we learned from that I wanted to share with you. One was that if you take a look at the total number of seats that we sold in Q2 of LT in Australia and New Zealand, perpetual and Desktop Subscription combined, the seats grew 19%. Very strong growth in total seats. Just as important and quite interesting is that after the end of sale of LT perpetual licenses, from that point forward all the way out, we've had a 21% increase in net subscription sales per month. The monthly growth has been 21%, on average, of Desktop Subscriptions, which is now the only offering for LT in that market.
We recognized that we knew how to go through getting the communication out to customers, how we manage that process with our partners, and we feel like we learned a great deal through the process and are actually quite pleased with the overall results from a metrics perspective. We are going through the next set of pilots and learnings this quarter as we're ending the sale of perpetual LT in all of the countries in APAC, other than Japan. China and India and all the countries in ASEAN will have an end of sale of LT family perpetual licenses at the end of this quarter.
It'll give us another opportunity to learn as we prepare for the end of sale of all perpetual licenses for individual products at the end of this fiscal year, and of course, the end of perpetual suite at the end of July of next year. In summary, we're feeling really positive and quite confident about the fact that we know how to drive more subscriptions across all of our customer segments. As a result of doing what we're doing, leveraging our new offerings, and delivering more value to customers, we are also getting more value from those customers. Now our channel strategies are working. We're working really collaboratively with our partners, and we're positioned to drive growth in both our direct and indirect parts of the business over the next several years.
As we continue to work closely with our partners and engage closely with our customers, we feel like we're very well positioned to work through the end of sale of perpetual licenses and moving into a subscription-only environment. On that, I thank you, and it is my pleasure to introduce to you Scott Herren.
All right. Good morning. The piece I'm sure no one has been waiting for this morning. It was a year ago at this event that I was announced as the incoming CFO, and a month later, I started. I'm 11 months into my tenure as CFO at Autodesk, and I will tell you, it's a very exciting time to be here. It's exciting because of what you heard from, a little bit from Carl, from Andrew, and Amar about the generational platform shift that we're making. Then the proof points that you just heard from Steve, that both our customers and our partners are adopting that new platform. It's a great time to be here, and what I want to do is now talk you through what that transition looks like financially.
Before we do, and you saw the safe harbor first thing this morning, I just want to cover, in case you didn't see the press release, no change to our guidance for the year. All right. Three things that I want to do in the next, let's say, 30 to 40 minutes with you. First is we've listened, and we've heard that improving your visibility through the business model transformation, both during the transition and afterwards, is critically important. I think Carl alluded to the fact that we may be overly disclosed today. We're going to walk through a lot of detail on what we think that looks like. The second is providing you expectations for what we think the financial trends look like out through time, and then a particular focus on cash and capital allocation, which is another topic that, again, we've heard repeatedly.
Before I jump into the financial models, though, let me step back, because we've talked about this a lot, but I think it's a great context as we lead into it. This is a generational change. Don't think of what we're doing. We talk about the BMT. In fact, it is maybe partly on us that we fueled that. We talk about this as the business model transformation. It is a business model transformation, but it's a transformation that's based on a fundamental platform transition that we see happening in engineering and design software, where you can go digitally from, as Amar said, from idea all the way through fabrication and even beyond with IoT.
The cloud enabling that, and some of the new things that are coming out in additive manufacturing, which will have a significant impact on both of our key markets, on both manufacturing and AEC. First and foremost, our business objective is to accelerate that platform transition, and remember, this is truly a platform transition. As we do that, of course, it brings to bear a different pricing model. The world of perpetual license pricing and maintenance doesn't make sense in a cloud-based model. You'll see our revenues migrate from a traditional perpetual license and maintenance model that most of our peers and competitors have, by the way, to one that has a very high percentage of recurring revenue as we make the shift. Of course, that revenue stream not only grows more quickly but is more visible and more predictable, and that drives shareholder value.
Andrew talked about it this morning, and I think it's worthwhile to touch on again. The real benefit we see, of course, is driving ARR. The fundamental thing we have to do to drive ARR is drive that growth in subscriptions. We'll talk about how we do that. I think Andrew spent a fair amount of time on it. Amar also talked on it. There's three big ways that we're going to drive that subscription growth, but it's that subscription growth that we see driving the overall growth in the model. Key takeaways. A couple of things. We're creating what I think we all believe is a very valuable business model, one that grows faster, one that is more profitable, one that is based on a high percentage of recurring revenue that I think is a far more valuable revenue stream.
Fiscal 2017 will be the inflection point. It's not surprising. Right? Fiscal 2017 is the year we stop selling perpetual licenses across the board in kind of two waves. Right? We'll stop at the very beginning of fiscal 2017 selling perpetual licenses on individual products. As we previously announced, we'll stop selling suites mid-year fiscal 2017. Fiscal 2017, expect that to be an inflection point. I wanted to also pick a point that was out in the future where we feel like we've come through the transition enough that the traditional financial metrics have begun to normalize. I'll talk about it. I'll put a few stakes in the ground on fiscal 2020 that the model is yielding.
It's not to say that's the end game, that that's the end state and steady state, this is where it'll be, but that's where we're going to be by the time we get to fiscal 2020. I think it's important to put some stakes in the ground again to help everyone build out their business model. What we see, and I'll peel this back on the next several slides, what we see is a revenue stream in fiscal 2020 that's at least $3.5 billion, a little bit north of $3.5 billion, with sustained double-digit growth. Non-GAAP operating margins that are in the mid-20% range in fiscal 2020, and I'll show you what those trends look like in a few slides, going above 30% out beyond that. Free cash flows in fiscal 2020 of about $1 billion.
Don't think of fiscal 2020 as the end game. Think of it as a point where we've come through enough of the transition that the traditional business metrics have normalized, and based on that, we wanted to put some stakes in the ground for everyone. All right. Let's talk about improving visibility. Message has come through loud and clear. I think there's a handful of things that are going to be really important as we do that. Sorry, we need to scroll a slide here. Andrew touched on this. Billings and revenue are not the right way to track our progress through the transition, right? For the obvious reason, billings is going to be turbulent through the transition, not only because of the shift from perpetual license over to subscription models, but varying terms of subscription models. Some will be monthly, quarterly, annually, or multi-year.
Billings are, by definition, going to be a little bit difficult to see how the underlying business is performing as we're in the transition. Revenues have all those same complexities that billings have, and then layer onto it, of course, the different rev rec of a subscription model. A better metric for you to track and look at are the ones that we talked about this morning. Andrew's touched on this, but I think it's important. It's not just annualized recurring revenue, it's also the % of revenues, at least for the next couple of years, that are driven by recurring revenue. I'll show you again what we think that looks like out through time. Next point is we've got two different trends going on inside the business, and what we've been reporting is the aggregate of those two, right?
One is what's happening as the new model subscription types scale up, and the other is what's happening in the perpetual license and maintenance business, which is on a different trend through time. We'll start to split those out for you, and I'll show you both for the last six quarters what that looks like, and then we'll talk about what those trends look like going into the future. This is a conceptual slide, but it looks the same if you plotted it on our expectations on either recurring revenue, annualized recurring revenue, or on subscriptions. The top line is the total, and you see it grows nicely out through time. The blue line is what we see happening in maintenance. When we're talking about ARR and subs, this does not include perpetual license, right?
This is what we expect to happen in just the maintenance piece of the business. I'll peel that back a bit more, but we see that actually beginning to level off this year, and then declining because there won't be new perpetual licenses to attach new maintenance agreements to. Declining at the rate of one minus our renewal rate, plus whatever we can attract away to the new business. The green line is what we see happening in the new business models. We've already seen through the statistics that Steve just went through, growing very rapidly. Growing rapidly on both an ARR and on a subscriber base. The same trends in either case, and I think the interesting point is we see both, whether it's ARR or subscription, we see these lines crossing by the end of fiscal 2018.
All right, let me talk about each of those separately. First, I'll talk about the new model, and I'm defining new model as the sum of desktop, cloud, and EDA. When you look at the graph on the bottom, the green bars are subscriptions, and subscriptions over the last six quarters have tripled for our new business model. I'm going to give you these, by the way. You're going to get PDFs of the slides, and in many cases, we've left the axes labels on there to make it easier. Again, this may be going to the point of over-disclosing, but I want to get past the we don't understand you, and onto now let's talk about the things that are going to drive that transition.
Tripling of new model subs, the growth of that, not just that we've seen historically, but what we'll see forward, is three things. It's new sales. Those are new sales from new customers like Steve just showed. It's also new sales to non-compliant or pirating customers, as we take the barrier to entry of a large upfront perpetual license cost out of play and allow them to buy just what they need. That'll drive new sales. Non-subscriber conversions with 2.8 million active customers that are not currently subscribers as they convert to desktop. Beyond that, further out in time, migrating people that are on maintenance over into one of the new model types. That's what'll drive the growth out in time. This space is all recurring revenue, by mid-year next year, all of our product sales will be here.
Virtually everything we sell will be in one of the new model types. Contrast that with perpetual and maintenance. I'll start by talking about what our expectations are on perpetual license sales. We expect it to be lumpy. We've got an end-of-sale of individual products at the end of this fiscal year, an end of sale of perpetual licenses for suite in the middle of next year. Each of those will drive some lumpiness in the billings and in the revenues for perpetual licenses. Beyond that, you need to expect license billings to drop to zero and license revenues to decline sharply. We're no longer selling perpetual licenses at that point.
What we're seeing as we approach this end of sale, we've talked about this a couple of times, is the attach rate of maintenance to that last buy, whether it was an upgrade or that last buy is a perpetual license now or a perpetual suite license in the future, we're seeing a materially higher attach rate of maintenance for those licenses. Customers buying it with the mindset that they're going to leverage that license for quite some time. They know full well we're in the process of stopping to sell perpetual licenses. They're attaching maintenance and renewing maintenance at a very high rate. Over time, as we stop selling perpetual licenses, the day we sell our last perpetual license is the day we sell our last new maintenance agreement. You need to expect this to peak and then slowly decline out through time.
I put both of these on the same axis. This is subscriptions, so that you can get a sense of not just the growth rates of each, but the difference in scale. Maintenance business has been built up on the back of perpetual license sales and maintenance attached for the last decade plus. The new model subs, you see more than tripling, and we've given you the actual data points here over the last six quarters. Same for ARR. Much faster growth on the annualized recurring revenue in the new model space, but again, off a much smaller base than what we see on maintenance. All right. That's looking backward. That kind of gives you a grounding point to start to model as we look ahead. Let's look forward at what our expectations are. Andrew talked this morning about the 2420.
He talked about the growth in ARR and subscribers. Fiscal 2017, I think it's fairly obvious, is going to be an inflection point. We look at Fiscal 2020, because of what we're doing in Fiscal 2017, because we're hitting the end of sale and we are ripping the Band-Aid off at that point. There's no more perpetual license sales. It has a couple of effects. One is, of course, it creates the inflection point in Fiscal 2017 on both billings and revenue. The second is it allows us to build that subscriber base quickly. All of our sales after we've gone to end of sale will be new model type, new model subscribers and ARR. As we build that base, what you see is from the inflection point in Fiscal 2017, a very rapid growth off that bottom.
Just to give you a sense of what the ARR looks like out through time, the green bars are ARR. The yellow line is the % of our total revenue that we think will come from recurring revenue sources. The ARR grows steadily up and to the right. Faster growth, it's hard to see from the bar charts, faster growth in the earlier years and, of course, moderating growth in the later years, but getting to that 24% CAGR that we talked about. Recurring revenue obviously grows very sharply in Fiscal 2017 and will level out somewhere between 90%-95% of our total revenue. All that will be left that's not recurring revenue is consulting and a few legacy products, creative finishing and things like that. Expect the overwhelming majority of our revenues to be recurring revenue. Here's what our expectations are on reported revenues.
We hit the end of sale. The rev rec is different on a perpetual license than it is on a subscription model. Selling a subscription versus selling an entire perpetual license. Our expectations are for that drop in revenues in Fiscal 2017, that's really driven by us moving more quickly, and us deciding to accelerate, to rip the Band-Aid off, to more quickly build that base of subscribers and ARR that drives the acceleration going forward and delivers what we just talked about in terms of Fiscal 2020 results. The one thing you will see as we go through this, even though I expect billings to decline in Fiscal 2017, driven by the end of sale of perpetual, is deferred revenues. We'll continue to grow in Fiscal 2017, and then we'll grow nicely, obviously.
As we're building a recurring revenue stream and subscriber base, deferred revenue will continue to grow nicely throughout that time. That's our expectations on revenue and deferred revenue. Let me talk about spend for a minute. The team has done a really nice job focusing on spend. There's a lot to do. Remember, this is an overall platform transition. There's a lot to do from Amar and Buzz and Chris's team in terms of driving product development, building out native cloud and mobile offerings, everything that we're doing around additive manufacturing, and of course, more recently with IoT. In addition to that, we're building out the core infrastructure to run our cloud on. There's investments that have to be made there to run those 360 product sets. Steve talked about the growth of direct touch and how that's going to look out through time.
We're investing in that capacity as well, both from an inside sales standpoint and building out the eStore to drive that level of direct touch. The business processes and business systems to support this new model also drive cost. We've talked in the past about spending growing in this mid-single-digit range, and I think we've talked about 5%-7%. Our view now is that's about a 5%-6% annual growth rate in spend, that we'll keep at that level by ensuring that we do a really good job balancing and rebalancing exactly where we're spending to get to the highest priorities. If I layer what we just talked about in revenue and spend on the same slide. I see everyone's kind of squinting to try to do the math. Let me do the math for you.
That does the difference between those two lines. This is our expectation on operating margin. You're probably focused on the dip in fiscal 2017. You shouldn't focus on the dip. That's arithmetic and revenue accounting applied to the very trends that we've talked about. The trends we've talked about stopping selling perpetual license and what that means for revenue, and the trends we've talked about the demand to drive this platform transition on spend. I think the more encouraging place, and I've talked about what our expectations are for fiscal 2020. The place to look on this is because we're taking that approach and because we are going to end the sale of perpetual and allowing the sales team and our partners to focus on building that subscriber base.
You look at the rate of growth from the bottom, that's what our expectations are, and that's what leads to the fiscal 2020 results that we've talked about. Beyond fiscal 2020, you can see it's not the end game. Operating margin continues to accelerate beyond fiscal 2020, and gets above the 30% range. This is the money slide. This is the slide I think everyone's been trying to figure out and see. This consolidates what we talked about on the two different trends of the maintenance model and the new model. It consolidates our view of what the rev rec will look like as we transition our way through that, and it gives you the roadmap of what we think the growth in margin, and the growth, obviously, cash production will follow that out through time. All right, you're going to get these slides.
Let's talk about cash flow and capital allocation. I know it's a topic that's on everyone's mind. It's certainly a topic that I read a lot about. You've seen what our expectations are on the P&L. Let me talk about our expectations on cash flows as we go through that. Billings come down, and you saw what we think the expectations are for operating margin. Free cash flows come down at the same time. We're expecting free cash flows to follow a trend like this, accelerating rapidly from that point of inflection to $1 billion in free cash flow by fiscal 2020, which is north of $4 per share of free cash flow per share. Layer in where we spend cash. We spend cash predominantly for two things.
We spend cash to buy back shares and offset the dilution of our equity plans, and we spend cash on M&A. This layers in, and it's intentionally a broad line because the reality is it's going to be a range out through that time. As we see what we expect to see in terms of billings and cash flows in 2017 and 2018, and continue to prioritize spending to buy back shares and materially lower level of M&A, materially lower level of spend on M&A, we still will consume cash in 2017 and to a certain extent in fiscal 2018. I think everyone's looked at our cash balance slide. This is what they look like in total, and I'll just make one quick point. Of that total cash balance, the domestic cash balance is a small fraction of that, a small portion of that.
Let me lay in what's going to happen to net cash as we go through that, because you've seen back in June, we took on another $750 million of debt to replenish the U.S. cash position, again, to drive what we need to do on share buybacks. You see the net cash position has been coming down for the last couple of years as we've taken on debt. Based on what our expectations are on both 2017 and 2018, you see our net cash position, we think at the bottom, is in that $500 million range. Before it begins, as we come back out and we see the acceleration on the top line and on billings, not only does cash flow come back out, but the net cash That's our expectation.
I think that's different than a lot of people had built into their models, and I wanted to make sure we had a chance to walk through that in detail. Now as we think about capital allocation, we have to balance being opportunistic with share repurchase to take advantage of what's a very low share price today versus what our expectations are in the next couple of years with being good stewards of the balance sheet and protecting the company at a point where we see the dip in net cash into that $500 million range. That's what we see. You now have the facts that we have as we look at where we're headed. We'll continue to offset dilution with our share repurchase programs. We will materially reduce what we spend on M&A, and in particular, materially reduce what we spend on M&A in the U.S.
As cash becomes available by tax reform, which is not something I'm willing to bet on, if cash becomes available via tax reform, or we move much more quickly or there's some other way cash becomes available, we're committed to returning excess cash to shareholders. All right. Let me wrap. Again, you're going to get copies of these slides. I see everyone furiously taking notes. To summarize, you heard from Carl and Andrew and Amar about the once-in-a-generation platform shift we're taking. You know, we've said several times, we're moving quickly to make this shift.
We are stopping the sales of perpetual licenses to drive the focus on the new business model types, to build the subscriber base, and to build up the position that we think we can be in and the very valuable model that that creates in terms of not just revenue growth, but recurring revenue growth that is more visible and more predictable than our previous stream and than many of our competitors have. You heard from Steve the proof points. Our customers and our partners are adopting the new models. That's what gives us confidence that moving fast is the right thing to do. The quicker we move, the deeper fiscal 2017 gets, but the faster the growth rate is coming out as we build up that subscriber base. That's where we're headed.
I want to just close by saying it's an exciting time to be part of Autodesk. I'll have my one-year anniversary in about 30 days. I think what we're doing is leading the industry in the next generation of engineering and design software. You saw what that can mean in terms of the users we can get to and the opportunities we can address as we do that. I think it's building a very valuable business model out in time. With that, thank you. What I'd like to do now is bring up Carl to say a few words in summary, then we'll go to Q&A. Okay, well, I'll bring up Carl and the rest of the executive team to jump right into Q&A. All right.
Great. Okay, we have time for one or two more questions. Well, Dave, you can be in charge.
All right, Carl, two questions. Actually, could you talk about the operational and transactional capacity that you expect to have in the future relative to where you are today? In other words, if through today you've had X amount of capacity based on new license sales, upgrades, classical maintenance, DTS, et cetera, when you get to fiscal 2020, what multiple of X would you need to support in terms of growing subscribers and so forth? Secondly, for Scott and perhaps for Steve as well, would it be fair to say that in terms of the margin expansion you're looking for, it would be sales and marketing coming down as a percent of revenue where you might have the single greatest degree of leverage in terms of expanding the margin. Gross margin might not increase that much. G&A is not significant. R&D can't pull back, obviously.
The most leverage you would get would be bringing down significantly the sales and marketing line.
Well, Dave, rather than thinking of it as a difference in capacity, there's actually this quantum step that has to happen in terms of us bringing on these new systems to be able to support, it's almost an order of magnitude more transactions. Just number 1, by reaching out to the new customers, also number 2, because of the options for people both to do things like monthly subscriptions as well as on-demand use of the cloud. People will be able to do pay-as-you-go models in consumption models. I'd say it's almost an order of magnitude more. The way we're solving it is not trying to take our current back office and scale it by 10X. We designed another system that we've been ramping up and slowly bringing online to deal with all that.
Yeah, I would say the same. I think it's an order of magnitude. Then to your second question on where's the leverage on the margin expansion. When you look at the scale of the revenue growth, Jay, from fiscal 2017 at the bottom of the inflection up through, I think, fiscal 2020, every line is going to benefit from that. G&A will benefit from that kind of scaling in the top line versus what we see in spend. R&D will benefit from that. Probably the biggest opportunity, and I'll let Steve weigh in because he's shown you some of the trends that we see on direct touch, I think probably the biggest opportunity is in sales and marketing.
I think the R&D, you'll see, as we move to a more singular model instead of supporting a desktop model and a cloud model, I think you'll be able to see some leverage there as well.
Hey, guys. Two quick questions. You guys talked about direct growing faster and indirect, but both growing. Is the margin target, does that fully contemplate the higher incremental margin that you get from people coming direct? It would seem just as I do the back of the envelope, that seems almost like a plug model. Second question is, Scott, you I think generally addressed the balance sheet, and obviously, you have a divot in onshore cash that'll accelerate back up. My question is really what in your view does the end state look like? As you guys have discussed today, you're going to be 90% recurring revenue, essentially 100% recurring profit business. Why should, in the future, pick your year, 2020, 2022, 2050, whatever you want, why should that be a net cash business at all?
You want me to get this? In the model, we did actually bake in channel mix changes and things and our efficiencies that will come from having more business in direct, especially as more purchases come through the low-touch, self-service type of market. eStore growing, as an example, gives us great efficiency. We have modeled that out, which was part of the reason why we're sharing with you how both are changing over time. To your question on net cash, you've seen the trends, right? You've seen what our expectations are in terms of cash consumption over the next couple of years as we kind of rip the Band-Aid off in fiscal 2017 and what effect that has on free cash flow.
I think the overriding consideration that we have right now in trying to balance out being opportunistic, because I think our shares are a great buy at this price, versus how do we take more cash on to go off and make a significant buy, is being good stewards of the balance sheet given what we see the low point being in fiscal 2018. That's without any shocks. That's without any unexpected another 2008 could happen. There's a balance point that we need to maintain as we go into that.
I would say, though, I tried to touch on this, to the extent that tax reform happens or something like that, we can repatriate a large portion of that offshore cash that we've got, obviously one of our priorities beyond just offsetting dilution and maintaining a real reduced focus on M&A would be to try to find a way to return that cash to shareholders.
Phil Winslow, Credit Suisse. I'm back here. A couple quick questions for Andrew. Andrew, you said in your slides the 20% CAGR in subscribers was predicated on a 30% conversion of the 2.8 million in non-subs right now. Given that you've talked about a three to five-year life cycle in the past for some of your products or buying cycle, does that feel relatively conservative? Help us maybe get to why 30%. Just one quick follow-up to that.
Well, remember, there's an assumption built in there that we're going to get better at moving that base forward as time goes on. The idea that you're suddenly converting them at those rates every single year is not the way we model it. What we're looking at is next year we're going to still be in the transition. There was a lot of upgrade buying and things associated with that. As we move into FY 2018 and 2019, we start converting more and more of that non-subscriber base, and we get better and better at it. As we exit the model timeframe in FY 2020, we're actually at a higher rate of conversion than we are when we start in FY 2017 and 2018.
We would love to get higher. We just thought this was kind of a prudent way to look at it. At 30%, it felt like it was very attainable, there's definitely upside built into that.
Got it. Also just a follow-up to that. I wonder if you could give a sort of analogy. One of the things I've heard is that, hey, if there's a lot of forward buying, let's say this year, could you see a big drop-off in new subscriber adds as next year? Is there any sort of analogy that you could give and maybe sort of end-of-life programs that you've seen in other geographies of how we should think about sort of the run-up to an end of sales and what happens sort of the following year?
Yeah. You heard a little bit from Steve about what happens. We've done this both on channels and in geographies. We did a bunch of this on eStore channels to see what happens. We've done it in a geography like Australia and New Zealand. We see a pretty similar pattern every time. As you run up to the end of sale, you do get some accelerated buying of the old model. People buy up the old model. As you exit the end of sale, you're selling at higher volumes than you were before. That's what we saw in Australia and New Zealand. We're sustaining it into this quarter. It's what we saw in the eStore when we did similar experiments. You're selling at higher volumes when you exit, and that's because more people are buying.
We expect to see every time we've done this, we see the same kind of phenomena. We expect to see that moving out as well. It's a lot of data.
Thank you. Steve Ashley, Robert W. Baird. I have two questions. One first one for Carl. Carl, today, one of the big messages is this is not just a financial transition of the way people buy their software. We are actually changing how it's architected, how, and where it's going to be delivered. Can you help us just get a picture in our mind how that transitions over time? We're talking about long term, we're talking about transitioning from a desktop offering to something that's ultimately augmented with cloud and then maybe becomes cloud. Can you help us just walk through how you see that transition?
Sure. I think there's two things to think about. One is, you can see it today, I think the way to think about it is our new offerings are all entirely cloud-based, as if there was no desktop business. If you look at products like A360, Fusion 360, PLM 360, those ones are just as if they were being done by a small company, a startup company coming to compete. There are ones that I put in the middle category, wherein they naturally have always belonged on the cloud. There just wasn't a cloud there for them to be on. I would put some things like simulation in that category. Simulation has always been something that's compute intensive.
Some of the visualization and rendering we do, heavily compute intensive, way better done on an on-demand kind of model and a consumption pay model, those will move relatively quickly. The existing desktop business, what I see those transitioning is by increasingly connecting them to services on the cloud, data storage, the access to simulation and visualization, so compute-intensive services that are there, access to some of the collaboration services. If you were to look out three years, your desktop products are connected, but still have a large installed component that sits on the desktop. You have things like simulation that have moved in based on customer choice. They decide to run them locally or on the cloud, then products all the way from A360 to Fusion that are pure cloud-based products.
Question was for Scott. After the midpoint of next year, there's going to be license remaining. You have a meaningful EBA business that continues, I think that's a number in our models that we all struggle for. Any color you could give us as to what kind of license is sustained beyond that end of life would be helpful. Thank you.
Sure. What's going to remain is not from license billing. What's going to remain is our new desktops are our new model types. EBAs is an example, not cloud. Cloud is all subscription, the billing and the revenue comes back into the subscription line. Both desktop and EBAs accrete out of deferred revenue partially through license, 70%-80% to the license line, and about 20%-30% into the subscriber line. The way to think about what that looks like in the future is to model out each separately. To model a perpetual license line going to zero, to model what we talked about and showed in the 24% CAGR, with a slow decline on the maintenance side and a very rapid growth rate on the new model side, then allocate 70%-80% of that over to the license line.
Unfortunately, it's an artifact of accounting more than it is mapping to the economic transaction. Economically, these are subscription sales.
The way the accounting works in the term license in particular, some of that comes back into the license line.
Kirk Adams, Rosenblatt. I apologize for only having one question. How does the model react in an economic downturn?
About as well as it does in an upturn.
I'll start and then-
I think I did already.
Yeah. Okay. Done. Do you have another question?
Look, over the long term, it has less volatility in the model. There's no doubt about it. Any way you map it, you can go through downturns and upturns as you get to a steady state, you would much rather be in this model than the old model. You'd certainly rather even be there than in a hybrid model. If you're trying to forecast what happens next year if we have another 2009, my guess, it's ugly, but even next year, it doesn't have the volatility that a 2009 would've had last year.
Right. A perpetual license model, when you get to a 2009 type event, you shut down capital spending, and there's no new perpetual license sold. In a subscription model, you may have a smaller workforce, but they still have to get their jobs done. It is going to be slightly less volatile. There may be fewer workers than you had before, but you still have to get your job done.
Thanks, Scott. This is Brent Thill with UBS. Steve, I know you said the direct to indirect was approximate, but the thing that caught my eye is that number you gave on direct is well below many of your peers that are sitting on stage what they've indicated to us in terms of the direct business. I'm just curious, is there a change that as you've gone into this, that you're seeing with the indirect or the indirect coming through and falling onto this a little bit smoother than perhaps six to nine months ago? Again, I know you said it was approximate.
It was approximate. No change. I actually even showed I talked a bit about that mix shift even last year overall. Nothing has changed actually overall. The same places that we're getting our most growth from direct continue to be the places that will get the most direct. It's through our enterprise business agreements, our cloud offerings. A reminder, our cloud offerings, we transact direct, even if a partner is involved and we pay the partner an agency fee, a service fee for that. Our eStore is growing overall and things like that. If anything, it's actually a strong projection of what we anticipate is going to happen in the growth of our cloud offerings and the growth of our EBAs and the continued growth as customers decide they want to go transact business on the eStore.
It's just a quick follow-up for Carl and Scott. Intuit and Adobe had something in common, and they both missed their targets. They gave all of us out of the gate on the move. Why do you believe the targets you've given all of us are conservative enough? You've had 100% hit rate of failure on the initial guide to the street.
Yeah, I was going to say, I thought we were in good company already. I don't know why we had to overachieve and do it again. I thought we'd already missed along with Adobe and Intuit in guiding on the model. I would say, one of the things we told you several years ago when we were doing this is that it was very difficult to actually predict the rate at which it moves. There were things that were in our control that we could move one way or another, such as the dates. We can play with levers like pricing. We also had longer term, but probably most effective of all was changing the offerings. We try to signal as clearly as possible that the error bars were big in the beginning.
I think if you go back and you watch or listen, what you see is we say, "This was our best guess at the time. We will tell you more as we know more about it." I think that's exactly what we're doing. We have learned a lot more about it, some from watching others like Intuit and Adobe. Certainly watching Microsoft, although they're a little harder to dissect the numbers out of there. Even you can see the customer behavior. We've certainly learned more from us doing it broadly as well as the experiments that Andrew described. I think we've been through it all. We've been going through it. This isn't like we're showing up and going, "Hey, here's what we think." This is part of the thing we told you that we would be constantly readjusting as we saw how the market reacted.
Hi, great. This is Heather Bellini with a question for Carl.
Where are you, Heather?
Right here. Sorry.
Oh, there you are.
I'm short. You can't see me. What changes are being made to the executive team's comp plan as a result of these new targets that you're laying out for us today? I guess specifically, you can share with us what are going to be the key drivers of the executive comp plan as a result of this.
As I will remind you, even though you guys will hate this, our exec comp plans, which used to be highly aligned with many of the interests in the room, at the behest of many in the room, moved off being highly aligned to less aligned. I know you guys would all like to forget your huge dissatisfaction with stock options and great fondness for RSUs. We can go back and find everyone kind of complaining about the problems with stock options. That was, in my mind, the greatest alignment we possibly had with our shareholders was that our stock moved together. Instead, you can see, you can go read through the K, and you'll see the disclosures. There are a number of things that we do.
We try to do is we try to tie the short-term compensation to kind of annual goals around some of the numbers you saw today. I guess they'll come out in the spring, and you'll see how the short-term ones and the long-term ones, which are now RSUs instead of stock options, will get tied to longer ones that including some of the metrics about return to shareholders. You'll see the two things. We're going to line them up pretty closely. There's also, as you know, a strong desire in the advocacy community out there to have them be different. If you wanted to have a long drink about this, I could tell you how this indirect manipulation through kind of advisory firms back to companies is a terrible way to align things.
I can certainly explain how we try to use the situations we see to do as good a job as we can. It really is the kind of things that you're seeing that Andrew and Scott talked about will end up in everyone's short-term compensation, and longer-term RSUs will be aligned with shareholder return averaged out over a period of time.
Hi, guys. Ken Wong from Citi. On the 3% average revenue per subscriber growth that you guys have talked about, it sounds like most of that is just a mix shift. Just wondering, have you guys embedded any element of pricing changes into there? You guys in the past have kind of talked about convergence of pricing between maintenance and desktop subs.
At this point, no. Mostly the phenomenon of 3% is all around the balance. It's all about the mix that makes the big difference. Like I said, we picked a point in time in which we're going to project out to the future. We've talked about some of the more flexible offerings that we go forward. One of the big variables in the model that's still out there is as we have more flexible offerings for the majority of people who are currently on maintenance. That's where we see an opportunity to move people more quickly to subscription as well as change the price. It's not fully contemplated in anything that we've told you because those offerings aren't in the market yet.
One quick follow-up on the maintenance uptick. I think you guys mentioned roughly the fiscal 2018 timeframe. Is that largely just due to, I guess, the last couple of years, you guys have had this really aggressive push to maintenance, and you have to kind of wait for that date, maybe end of life, not end of life, but for their contract to come up for renewal?
There's a bunch of that, yeah. Like you said, let's just back up for a second. If you have a customer who bought a perpetual license and has been paying maintenance for two years, five years, whatever it is, it is hard to move them to a subscription model, unless you offer something different. We think that flexible offering, just as when people shake their head and go, "Why would anyone move to this thing where you'll end up paying more?" Hopefully, you saw through the stuff that Steve showed and Andrew showed that our enterprise customers are willing to pay a lot more, if they get more. We think there's a huge opportunity. There's barely an account that I ever visit where I walk in, I go, "We've done such a great job. We've saturated this entire account.
They don't need any more software." That's what's being borne out when you're seeing these enterprise accounts use a lot more. The same thing is true in smaller businesses. We think the flexible licensing will offer the same benefits, at a reduced scale to these midsize companies, and that's what's going to move them over. There's some amount of just the natural timing that comes from when they got on maintenance, and some of it comes from having a more compelling offering going forward.
Back here. Matthew Hedberg from RBC. Scott, I had a question on the cost side again. You talked about the margin, the leverage in the model.
Yeah.
5%-6% of costs on an annual basis growing. I'm wondering, to what extent could those actually flatten out further, similar to what we've seen with Adobe? I guess a follow-up to that would be, you guys have been planning this for 2 years, and I assume you have a lot of duplicate systems running right now. To what extent can you help us with the duplicate costs that come out next summer in terms of running these two different models?
Yeah, sure. I'll take the first one, then we can tag team on the second. Just the cost of living increase, Matt, that we see each year. Nominal merit increases for the team and things get a little more expensive, probably consumes around half of that 5%-6%. As we look at that and start to dissect what are the investments that we have to make, because this is a fundamental change in the platform, and it changes Steve's team, it changes Buzz and Amar's, and Chris Bradshaw's team quite a bit. It has an effect on the way we market and the way we sell. It really is touching every part of the company, and the path we've been on is ensuring that where we are spending within that envelope, that we're getting the biggest bang for the buck.
We're putting as much wood as we can behind the most important arrows on that. Could it flatten out in time? I don't see it within the window that we laid out between fiscal 2016 and fiscal 2020. Could it flatten out after that? It's a little bit hard to say.
Yeah, in some ways, I think it's the flip side to the other question there about what happens in a downturn. From what I can tell, we are in one of the hottest job markets we can remember. I think probably all of you have stories that are starting to rival 2001. We're certainly starting to see that phenomena, at least in probably half the markets in which we compete for talent. We're seeing kind of crazy offers out there. Particularly, I was talking to someone just before the event started, and I'm not sure about the irrational exuberance in the public markets, but certainly in the private markets. You're seeing the billion-dollar unicorns are as crazy as anything we ever saw in 2001. Unfortunately, we're competing for talent with many of those companies. That's one part of the equation that really drives the cost of living.
It's a combination of salaries and just overall benefits, particularly healthcare that drives a small part of that. I think there is some opportunity for synergy in removing systems. The big trick is the degree in which we're meticulous about making sure that the systems, we get rid of that last 5%.
Right.
There is just a tendency within companies, even when they're moving to something new and they've moved most of the people off, out of some kind of bizarre sense of contingency, they keep the old systems around for beyond when they're really needed. As an experiment, how many of you still have a BlackBerry? No? Okay.
Everyone else is too embarrassed.
All right. There's at least four then. By the way, that was the quickest hand raise I ever saw. That's like this. Yes. There's examples of things that are no longer the appropriate technology. Companies keep them around. It's really the burden's on us to make sure that when we move over, we really kill the old stuff. In that case, there's a lot that can be done.
Hi, it's Daud Khan from Berenberg. I was just wondering if you could talk about your plan and what is the assumed attrition rate in that, especially the trend between what you have today in the subscription base and what will be going forward.
Yeah. When you say attrition rate, you mean the churn rate-
Yes
The renewal rate. It's pretty consistent. What's assumed in there is consistent with what we've seen, and what we've seen, I talked about briefly, is a significant step up in attach rate. Along with that, a high renewal rate at the same time. You can take the two pieces separately for a minute. If you think of the maintenance business, the people that are buying that last license are going to attach maintenance to it and are going to renew that maintenance to get the full value out of that last buy. On the new business model types, in EBAs, they renew or they no longer have access to the software. On desktop and cloud, it's the same. You renew or you move on.
In that space, the renewal rate boils down to the rate and pace that we can be more competitive than the other players in the marketplace that are going to offer similar technology. To date, we haven't seen anything that gives us concern that's going to take us that down.
You're just, like, in the high, kind of mid-90s or something like that?
Yeah, we don't actually quote that number, but it's a very high renewal rate.
Hi. Back bench mechanical engineer here. I don't know if you can see me behind the TV screen. Kash, I'm going to be with Meredith. Question for you on the longer-term part. If you look at the updates implied by fiscal 2021 guidance, it feels like you're expecting a significant decline in the perpetual license space, and therefore a significant pickup consequently in cloud EBA and other varieties, which is where I would think that your ARPS per customer. Is it loud enough? Sorry. Should I repeat-
Yeah. We can hear you.
Okay, great. The non-maintenance types is where you should see increased value for subscribers. It feels like the model is skewed heavily towards that inflecting. Although we've been hearing from Carl that the offerings are going to be sufficiently different to incentivize the licensed customer to move to subscription, we've not heard more, I think, on that. Could you talk about what exactly is the nature of the product that is going to look different 2-3 years from now, and how would you incentivize the customer and the channel partner to make the move? It occurs to me that one part of the equation is the sub base, and the other part of the equation is how that sub base actually mixes itself. Is it going to be more maintenance than you expect or less than what we expect?
Can you just talk about that a little bit?
Yeah. Let me be clear, and then let's have Andrew jump in. What we're talking about is we didn't have a model that made big presumptions because the offering's not out there. The biggest thing that we can do is take the same kind of offering around flexible use of our products. We've talked about it with various names, but basically a complete suite of products that you want across your company for a single price per user. That's what moves people. If you look at the old days, people would license single seats. We then prepackage things up into suites that we hope met the needs of the customer. The next step in that is letting people configure and choose the things they want. That's exactly what the EBAs are, that's exactly the path we followed with them, and that's the same thing.
It's not wild speculation that the customers will want this. We've seen that customers want it. We started with a handful of enterprise customers, worked our way down, and now with what we've learned from that, we're making a broader offering that goes to all the people in medium-sized businesses, primarily served by our channel partners.
In a follow-up, I guess customers that have licensed, but for some reason, if they choose to not go to desktop, will they get a license update as they've been getting in the past?
If they continue to pay maintenance, yeah, they're going to continue to get it.
A license update. There will be a license update.
Okay, just to be clear, the people who are on maintenance today or maintenance subscription, if you continue to stay on subscription, you get the update. If you never got on subscription or you choose to not renew your subscription, you don't get any updates.
The assumption is that there's going to be significant difference between the license update functionality in the future and the desktop and the EBA and the cloud, that people will see the value and therefore will.
Sure.
Okay. Got it.
Yeah. If I can add, one of the things that we touched on earlier is cloud services, right? You can start seeing intermixed cloud services will play a greater role in the new subscription than on the license update for maintenance. We're going to start intermixing many more cloud capabilities, we start to see some of the capabilities are the same, but the subscription offerings will have more capabilities because of the intermixing with the cloud.
Got it. Final one for you, Scott, then we'll.
Also, one more bit. The other part about this is that customer environments also are changing rapidly. It's hard for them to do a 21st century job with desktop software. Things like additive manufacturing, you need new cloud capabilities, you need generative design, new techniques that you just can't get the old way. Products that we have today just wouldn't be possible without the cloud. They're real different now.
Hi. Neil Doradla from William Blair. Carl, you talked about from product conceptualization to realization, this whole ecosystem developing, simulation is playing an increasingly important role in this. Can you talk about when you look at Autodesk and the multiphysics simulation capabilities that Autodesk has, how do you view yourself in the competitive landscape? Also going forward, one of the issues in simulation is if you have an incumbency factor, there's a lot of stickiness, the willingness to change is not there. You talked about this being one of the earlier markets that's going to embrace the change. Help us understand how should we be looking at this whole space.
I didn't say necessarily it's the earlier market to embrace. I said it's the most natural to embrace. Many of the incumbents are very reluctant, or the models that they want to use in this are kind of old world. To the extent that any of the incumbents are embracing it, they're doing it fairly half-heartedly. That would be a more accurate thing. The thing about simulation, I'm not talking anymore about simple linear static stress, the stuff you run on your desktop in four minutes. I'm talking about the kind of simulations that take hours or days or weeks. Those have always been natural to run in an environment in which you can scale your computing to the need, but the next day you may not have a job to run.
It's always been incredibly expensive for people to build that capability behind the firewall. Now some have done it, and some will continue to do it, but it's not a natural way to do it. When I look at our capability, I think we compete really effectively in some places. I think our computational fluid dynamics is good. I think our plastic simulation is world-class. I think with the acquisition of our Nastran stuff, we've helped answer many of the questions about what keeps something sticky. Being able to put the Nastran sticker on it really helps in lots of customer accounts. I think we're trying to go about this differently, and many of the people who cater to the simulation market would say they've tried for years unsuccessfully to get designers and engineers to use simulation. That's really the part of the market we're focused on.
I have very little hope of really winning in some of the places, either kind of the extreme kinds of simulation, or the ones where it's an analyst who does very specialized work. There continue to be a whole bunch of niche companies, public and private, out there that serve those markets. That's not what we're looking at. We're looking at bringing this general capability to the mainstream designer and engineer. Buzz, you want to.
Yeah, I'll just add that we have a big focus on advanced materials, composites, plastics, and the cloud. Those are the two things. Advanced materials, cloud. Also sort of design for printing is a big thing where we're going to differentiate ourselves and I think have really unique offerings already.
This is Deepak Shah from Morgan Stanley. Thank you guys for the presentation today. We got a lot of clarity on the Business Model Transition, and Andrew did a great job of sort of giving us that clarity again into how to think about the model on a going forward basis. What should we expect going forward from you guys in terms of metrics, in terms of guidance? Are you going to be guiding to subscriptions? Are you going to be guiding to?
I think we go mum for five years. We'll get back to you guys then.
Well, let me ask it. Will we hear from Andrew on a quarterly basis versus an annual basis?
Some of you already do.
Yes, no. I mean, Scott, jump in. The idea was to try to recalibrate everybody on the metrics that we thought were important. Again, this was part of what we had told you before, that as we understood better what were the things that really drove the model as maybe as opposed to those things that were maybe highly correlated with the model, that those were the ones we would talk about and we want to update those on a regular basis.
Yeah.
That's what today was about. To get everyone on the same page, take a little bit of a longer session to do it. We also thought it's much more difficult to do on one of the calls. We thought with a little bit of background, you've seen the whole model, we get there, and then we can go through this and report more regularly.
Will ARR be something that you guys are reporting on a quarterly basis and guiding on a going forward basis?
We'll definitely report on it on a quarterly. We haven't crossed the bridge on guidance on either ARR or subscriptions yet.
Okay. In terms of subscriptions versus subscribers, it's nuanced, but there's a difference there in terms of what you're counting.
Yeah. We're being careful because we have a strong ability to count one and a lesser ability to count the other accurately enough to keep Scott out of jail.
Legal.
Which is pretty important to him.
We would all sacrifice on the fire.
If it was up to us, this entire side. Him and Pascal have this insistence on keeping the SEC happy. It really comes down to, in many cases, you could map them identically one to one. A subscriber and a subscription's the same. The problem we have is that we might have a Revit user who is also a member of an A360 project, or a Fusion 360 user who is not the primary owner, but somewhere tucked into one of the subscribers that's paid for individually. For example, you could be a PLM 360 customer. You're paying us for 500 seats of PLM 360, and the person happens to be a Fusion 360 user, and they're paying that thing independently. Our ability to ensure they're the same is not there. One, we have a financial relationship with them, and we can track the other one.
It's an email address that's put into the system by the administrator of the PLM 360 account. In order to be as safe as possible, you know what? As we go through this, we will probably do some work ourselves to understand what that overlap is. Yeah, we are being really careful in the language, and it really is in a financial reporting sense. We're not trying to be too cute about it ourselves. It's just that when we realized that our ability to count this perfectly was not there, we recognized we had to. Does that make sense, Keith?
Am I correct in thinking, because one of the metrics that was probably lower than people expected was a 3% growth in average revenue per subscription. That doesn't account for if one user starts using multiple products.
That does not. If you've got user A, subscriber A, and you've subscribed to three different products, obviously the value of that subscriber is significantly higher.
Right. It's not an ARP count at all.
That is not kept.
It's not an ARPS number. That's what we worked hard to get rid of because we couldn't do it.
Okay.
Right now, I would go with the assumption that the vast majority, actually, of subscriptions and subscribers line up. We'll try to put a finer lens on this as we go through it. The correct language for now is actually subscriptions, which we can count.
It does have the effect you just laid out, Steve, which is if one person, so it would be one user, has three subscriptions, we're looking at the average revenue per subscription, not the average revenue per user.
Per subscriber.
Subscriber, correct.
Thank you.
Saket Kalia from Barclays. Thanks for all the detail. Just two quick clarifications also for you, Scott. This first kind of housekeeping. Should we use the 2015 or fiscal 2015 or fiscal 2016 ending subscriber or subscription count and ARPS number, if you will, as the starting point for that 24%? Or should we use the 2016 number as sort of the starting point for 2020? Does that make sense?
Yeah, it does. Those targets are ending fiscal 2016 to ending fiscal 2020.
Got it. Secondly, I can appreciate the difference between billings and sort of ARR in a hybrid model versus a pure subscription model. Is there a way to map ARR to billings, especially thinking about $1 billion in free cash flow in 2020? Billings have to be an important number here, how do you map the two together?
Once you get out that far in the future. Well, the only kind of fine point that you need to bear in mind is these subscriptions have different terms. We'll have some subscribers that are monthly, some are quarterly, some are annual, and some are multi-year. Billings will have that perturbance all the way out through time. That's why we talked about ARR instead of billings. Ultimately, ARR and revenue is going to be very close.
They all converge.
When we get through this transition.
Billings. They all converge.
Exactly. Remember, our objective is to give you a clear, unambiguous set of metrics that if they're going up like we said they're going up, you know that we're managing the transition correctly. Once we're through the whole thing, all of these things start to converge to classical metrics. It's when you're in it that it matters.
Sterling Auty with JPMorgan, here in the back. On a question earlier around the macro, you kind of jumped to the extreme in kind of pulling in 2009. I think what's still on a lot of investors' minds is the current environment that we're in now, which is really uncertain when we look at some of the geographic regions. I think what some investors want to understand better is, looking at what you laid out today, did you give yourselves any cushion for, let's say, a little bit more squishiness on the macro side than what we're seeing? Put another way, looking at the levers that go through the transition, where is the room where maybe you could absorb some additional squishiness from where the macro sits today?
We don't have a squishy line. I would say there's a couple places where we think the assumptions are conservative. If I was to be an accurate student of history. Let me maybe baseline here, because we're not assuming that the economy gets better than it is now. We're assuming kind of the same kind of uneven performance with kind of a backdrop of dark clouds out there, but it's not pouring yet. We also are not making big assumptions about what happens with currency. Those are variables that could perturb the model. Now back to where we are. I would say my experience is that if the economy goes south, not 2008 south, but south, a lot will depend on where it happens. Right now, we continue to see strength in Europe and the U.S.
If those got weak, along with China weakening, Japan, I would get worried that we're not conservative enough. If they all turn sideways at the same time, I think there's some risk there. On the other hand, the way we got the squishiness in the model is there are a number of assumptions that I think are fairly conservative.
Just a quick follow-up on the construction end. It seems like one of the biggest TAMs, the biggest opportunities. I think it's been one of the highest net inflows of $, new $ into the company, correct me if I'm wrong, but can you just walk through where we're at and what end you think we're at in construction and seems like you might want to be there?
I'll try. I think we're in very early stages still of construction adopting technology. There's really two technology stacks that they're adopting. One is Building Information Modeling. I'd put that as we're at the top of the third. The large general contractors are the ones who are adopting Building Information Modeling to sort of figure out how to build the thing that they're being asked to build. Some of the EBAs that both Andrew and Steve touched upon, they're large construction companies. There are large general contractors who are adopting BIM technology. I would say we're in the first or second inning of trying to get the BIM 360 stack broadly adopted. It's really only North America and Northern Europe right now. There's a whole worldwide construction adoption opportunity.
You think about how much construction's going to happen in emerging economies, the large opportunity there that's still out ahead of us. We're still really The technology adoption is really only happening at the top of the pyramid. The mid-market of medium-sized to small-sized construction companies, that's wide open. I think we are still in very early days, and the growth we are seeing is the general contractors adopting technology.
One thing I'd add is, both Amar and I met the other day with two
Yes
relatively good-sized contractors. They both came into the office, what was interesting to me was their insistence that the way they made their business better was through technology. Construction worldwide is a $4 trillion market. It's up there though with restaurant businesses as being infamous for being low single-digit margins, in bad years, it's negative. The way they see to change this is more widespread than I've ever seen, is a belief that they need to adopt technology. Both those companies were in here for exactly that purpose. As a matter of fact, some of our most advanced general contractors, one of the services that they're selling to building owners is how to use technology not only for the construction of their property, but moving into the ongoing maintenance of their building.
They're expanding their markets, they're mostly doing it through technology, not thinking that if I hire twice as many crew members, my business is going to get better.
Hi, Steve Koenig from Wedbush, way back here. Greetings. I wanted to ask about how you're going to capture the current demand for Suites, which are, those perpetual license will go away in about three and a half quarters. You guys gave yourself, I believe it was over four quarters before the end of the standalone perpetual sales to get traction with Desktop Subscriptions. I'm wondering, what will be the vehicle there? Will that be this new Flex model that you talked a little bit about? Maybe the second part of the question is, could you give us any more specificity on how you might convert existing maintenance customers to a higher value subscription? Is that also the new Flex offering, or am I confusing things? Are there other new cloud-based offerings that might come to help convert those customers to higher value subscriptions?
All right. When we look at the run rate, you asked two questions there. When we look at the runway to Suites, I made this allusion to this notion of collections of products. They're essentially a going-forward buying model for a Suites customer. No, the economics will be attractive for that type of buyer. You also have to remember that the Suites are our highest Maintenance attached products out there. They have super high attach rates, and they have super high renewal rates. We're going to have a very stable base there. The new run rates and new customer acquisition is going to come in on these new collection offerings, both as the multi-user and single-user license. That's something you're going to see happening as we move past the middle of next year.
When it looks to moving Maintenance customers, one of the reasons Carl alluded to, you got to have the right offering, and you also have to have the right understanding from the customer. One of the reasons why we look out into FY 2018 and beyond in terms of how we're going to start moving and accelerating and moving the Maintenance customers, they have to understand, one, that access to more products is better for them. Two, that the experience we're delivering, bundled with these new offerings delivered through the cloud is superior, so they have to see the evidence. You know what? This isn't all that different than the journey we had when we introduced Maintenance. Carl, how long ago did we introduce Maintenance?
Eight, nine years ago.
Jay, how long ago?
11 years ago.
It's over a decade. I'll tell you, during the first three years of that, customers were not exactly embracing Maintenance across the board. You're going to see some of the same kind of dynamics, some of those same kind of transitions as people start to see, hey, this is a better way to manage my relationship with Autodesk. It's going to be a lot faster than Maintenance because I think the benefits we're putting on the other side are a lot better, it's going to follow a similar path.
Yeah. Just one thing I'd add to what Andrew is saying is that one of the assumptions that people make is that we're offering the same thing in a different way. I actually think we are going to differentiate the subscription offering by intermixing it with better delivery experiences, better management experiences, and also these intermingled cloud services. The offering is going to look higher value to the customer. They're not just going to look at it and say, I can rent the car or buy the car. The cars are actually going to be fundamentally different.
Well, it's just what's happening with the enterprise customers. They see this key Flex model is something quite different than what they had with perpetual and Maintenance. This is exactly what we want to go and replicate with the new offerings in the SMB space. Just a quick follow-up on that, I remember the Suites also took about four quarters until you had a chance to tweak the packaging and pricing before some of the traction was evident in the field and among the resellers. Do you feel like you're giving yourself enough time with moving this Flex model to the mid-market, given that you don't really have an offering for the mid-market?
Oh, yeah. Like I said, we don't really have any migration models into our plans until FY 2018 and beyond. We've learned from all of the adoption things. One, just listen to what Amar said. We have to start showing the market that these are different cars. That's going to take time in and of itself. The whole process of getting it out through the various channels is going to take time. Yeah, we look out, and we see this accelerating and
FY 2018 and beyond, not immediately.
Yeah. I think it's important to note that the model that we're working on is going to be a simpler model. It's less complex. There's actually fewer things. One of the things I'm really pleased with is I actually think it'd be easier and more straightforward to take it to market to our partners than what we have with the many different suites. We're adding value for customers, but we're also simplifying the offering. It should be a bit more straightforward to get to customers and provide them with access to that incremental value.
Thanks.
Hi, it's Gregg Moskowitz from Cowen. Question for Steve. The 90% ARR CAGR that you quoted for your Multi-Flex customers, very impressive. It is only about two dozen accounts today. From your conversations with your named accounts, I know you talk to a lot of these guys. Is there a sense that there's an early adoption spend dynamic, if you will, where maybe you sort of cherry-pick the biggest and best opportunities initially? Or do you think there are big opportunities for increased spend with those other named accounts as they move to Multi-Flex? Secondly, wondering because there were a lot of fiscal 2020 type objectives, and I know you want to get all these named accounts to move over time, but I guess what percentage would you be satisfied with by fiscal 2020 on the Multi-Flex front?
A couple clarifying points. We're moving them to the Token Flex model, not the Multi-Flex model. The 90% ARR was across our enterprise business, not just specific to those accounts. Your premise is right. Our ability to move customers over is basically limited by how much time I need to get my team in front of the customers and work through them to show them the incremental value that they can access through having this new model, this new approach that helps them become more competitive in their market or win more business or increase their margins and ultimately reinvest. It takes some time because they want to analyze how their current use paradigm is compared to what they actually can have. It's just a matter of getting there and talking to those folks. The Multi-Flex customers had already made a step in that direction.
They moved from maintenance to something that was not consumption but had a different structure. They were the ones that were easier to get to because they'd already been making a switch, and they actually saw limitations in that model that the Token Flex model solves. The Token Flex offers a lot of incremental value. Right now, we're engaged with lots of our named accounts on helping them understand the value of making that move. I anticipate that what we're going to see is a lot of customers adding to the maintenance and perpetual current model moving to the Token Flex model. That's where I actually see the growth coming. We don't have modeled out a specific set target for the number that have moved by the end of FY 2020.
I have actually mandated to my team, our goal is to get in front of every single one of our customers and show them why the very best option for them is to move to Token Flex.
Can I add just one thing, Scott? The reason our customers are interested in the Token Flex model is the portfolio we have is actually one of our biggest competitive differentiators, right? You look at these worlds, they are sort of converging. People are using storytelling from the world of movies. They're using mechanical fabrication technologies in construction. This range that we have from mechanical applications to M&E to AEC, it actually lets the customers do the kinds of things they want. They're moving to Token Flex because it makes sense for their businesses. That's why I think that we're just still scratching the tip of the opportunity there.
That's helpful. Just a follow-up for Scott. One of your charts was around recurring revenue as a percentage of total. I think by fiscal 2022, which I believe was around kind of 95% or 95%-plus. Fiscal 2020 looked like it was around mid-70s, if I saw that correctly. Just kind of wondering why that isn't higher by fiscal 2020. If it actually is higher, does that impact your revenue target of at least $3.5 billion by fiscal 2020?
Yeah, no. This year, we see it in the mid-50s as we just reported in Q2. It'll take a big step up in 2017, then it'll hit that between 90% and 95% rate by fiscal 2018. You may have had the column slightly misaligned when you looked at it. It'll hit that by 2018 and kind of steady state its way out.
Okay. Makes sense. Thanks.
I have a question for Carl. The gaining 30% of the users that are not subscribers today seems somewhat conservative, yet financially, that 30, 60, or 70 could really drive the model. When you look out three or four or five years, what are the plans in place to exceed that 30, to incentivize those people to come over to be subscribers? Because the leverage of those guys since they're higher value add customers is tremendous.
I would say the thing that's true here has always been true, is the lifetime value of these customers is enormous. The thing that we can do is during these next two to three years, is get as many over as possible. That may be by the offering, it may be by pricing. There are a whole number of things we may do. That's why we've been more conservative about looking at the mix, because one of the things is we would like to way exceed that number, and the way to do it is to make sure that while the people are contemplating it for the first time, the offer is as attractive as possible. Glenn, you want to add something?
I said to you earlier, there's going to be learning associated with this, and we anticipate that we'll convert at lower rates initially and start accelerating away from that. How we approach these customers, how we communicate with them, what we offer them to move them forward is different than what we've done in the past. That means we're going to have to learn. What those methods are, and that's kind of what we've built into there. Is there upside? Sure there's upside.
The other thing I'd say is-
I think we're being-
the differentiated offering will be very apparent to our customers, because they'll have, in most cases, a suite of purchasing all perpetual licenses. As they need to add users, they won't have an option but to add anything but one of the new model types. They'll have them running side by side, and they'll get the value add that we can build into it will be very apparent.
Thanks. Two things, one for Steve and then for the group. Steve, first, what's your view of the long-term structure of the channel? That is to say, are you adhering to the consolidation expectation that you've talked about in the past? What, for instance, do you foresee for the role for what is nominally your largest customer, which is a distribution company that you sell to, is typically about a quarter of your business, and has in fact grown disproportionately. Then for the group as a whole, what have you seen your classical competitors in CAD, PLM, et cetera, do, if anything, thus far, either the pricing or what might they do to possibly forestall the attractiveness of your offerings? For example, SolidWorks has become more promotional-oriented than they've been in the past. PTC is about to go through a pretty significant repricing of subs versus perpetuals.
STLM has become more price competitive, et cetera. Would that be enough of a headwind via pricing to perhaps forestall what you might be doing on the new product front?
Yeah. I can take the first one, if you want to please say one. We do see partners still working through a consolidation process. They're doing it on their own, by the way, so it's not our strategy. It's actually how they're going about going to market in a way that actually benefits their businesses, but it actually is benefiting ours. We've seen over this past year a handful of partners that have consolidated or acquired other partners. What typically happens, by the way, is that we have fewer partners to go and manage and support, but the good people from the companies that were acquired move to the companies that are stronger, healthier in investing in their businesses.
If this continues the way it is, and we hesitate to say that it will, we actually see it as a positive because it's fewer partners for us to manage. It's not taking away capacity. It's actually increasing capacity and making the stronger partners even more capable of investing in the future. Distribution serves different roles in different parts of the world. It's more important to us outside of the largest markets, where they have to manage a lot of the elements of the partner relationship. We manage the strategic elements, but there's lots of smaller partners. Will that actually change over time? Yeah. As fewer partners, smaller partners exist in the marketplace, some of the value add for distribution actually diminishes. Our distribution partners are still adding a lot of value in many of the markets around the world, where we need some local presence.
We need to be able to transact business in those local currencies that we're not ready to go do ourselves. The role will evolve over time, but they will continue to be part of our ecosystem.
I would just say one thing. I think, Jay, if you think about it, when an incumbent is on the wrong technology, lowering prices can forestall something for a while. It's not that effective. On the margin, certainly, there are some customers who can go, "I'll wait another year." Fundamentally, they have to do something differently. I think it's hard at that point because it becomes somewhat of a downward spiral, is to just try to do with price. The only situation I can imagine in which it's worth doing that is if you have to forestall something, because in 12 months you'll have something that competes more effectively. Otherwise, all you're doing is taking your product near the lead and walking it down to zero.
Just a little clarification for Scott or for anyone. Do your long-term margin assumptions embed a complete conversion to an agency model succeeding a VAR economic model?
No.
What would the mix be?
Well, as Steve said, the slide that Steve showed was a conceptual slide of the mix between direct and indirect. We are doing some agency models. Steve, I'll let you weigh in on exactly where we are on that today. Don't think of that. If the reseller is adding value in the loop, whether they're paid via agency model or they're paid via the way we pay the channel today, discounts and back-end incentives, the economics aren't going to change. The line item that it shows up in the P&L will change, but the economics aren't, if they're adding value.
Just to add, some of that growth in that direct line was agency fee programs that are predominantly associated with either our cloud offering scaling and growing, because those are direct transactions. We're paying the partners that way. Or if a partner's helping us close an EBA somewhere, and we're paying them in that capacity. The growth associated with the agency was in that direct growth that I showed at the start.
This is Keith Alfaro again. Maybe a question for Carl as a board member around capital allocation. If I listen to you guys from a broader perspective and listen to Scott talk about how the stock is underpriced, obviously the stock has pulled back a lot on people's lack of understanding or concerns around the business model transition. You guys have the utmost confidence in this business model transition being very positive for the stock on a going forward basis, interest rates are at an all-time low and about to go higher. To me, that seems like the perfect storm for taking down debt and buying back your stock. From a board of directors' perspective, what are you guys waiting for?
What has to go further for you guys to really step in there and show confidence in this business model transition and buy back your own stock?
One, we've been buying back more aggressively during the last quarter. How much would you prudently say we should borrow to do that?
Where they are and take down $1.5 billion.
In addition to what we have?
Yes. You guys have been offsetting dilution a little bit, but you guys haven't been taking out shares.
No, dilution, it's pretty flat depending on what period you look at. One of the things that we've discussed is buying back stock more aggressively, opportunistically, we've been doing that. We have been contemplating some other things which I'm not prepared to talk about now. It all comes about because of the offshore cash. With that, we've been contemplating other alternatives that we're not ready to say anything about right now. Thanks for the free banking advice.
Over the last two years, could you talk about some of the positives and some of the negatives of your anti-piracy approach and how that's changing going forward in the next year or two? What opportunities are there? Talk a little bit about geographies.
I'd say the negative one is anti-piracy doesn't do any good. It has hardly made a dent in desktop products. I would say, I don't think it will. With the abundance of sites at which you can download software, I don't think piracy rates on desktop products will change. The biggest plus is it is much more difficult, I wouldn't say impossible, but much more difficult to steal cloud-based products. You don't hear the peer companies, you don't hear NetSuite or Workday or Salesforce talking about their piracy rates. The biggest advantage we get with trying to fight piracy is moving to the cloud. The one in the short term that helps is that we do think there are really two kinds of pirates. There are serious people who will steal no matter what. It's a business.
There are casual pirates who, because of the inconvenience of the way we offer licenses, that they decide not to buy. Some of the subscription offerings are geared towards them. As a matter of fact, I'd say one of the high points is in the M&E market. I suspect a lot of the short-term subscriptions we see are people who otherwise wouldn't have paid for the product.
Okay.
More questions?
Let me follow up on the share repurchase question. There really hasn't been a big material pickup. There was a pickup in Q1, Q2, you have 11 million shares authorized. Why not go to a ready offense and not huddle and get in the game a little more aggressive? Can you just help us walk through the mechanics of what is causing you guys to stall? We get this question every day, and I don't know if we've heard a real clear answer.
I'll say one thing, Scott, feel free to jump in. I wouldn't look too much at the authorization. That's just a mechanical process. You need one, when you have one, I think people read too much into it. You can do a lot with it or you can do nothing with it. I think it's interesting. I don't think it's a big deal. Like I said, I continue to want to buy opportunistically. I don't think necessarily the approach that everybody favors, certainly throughout the tech community. I like that. I still think if I can buy something for $45, I like buying it for $45 instead of for $60. I think we should continue to buy opportunistically.
We are being cautious, as you saw the dip in free cash flow worries us some, and we're being very aggressive in trying to pursue some other ways to unbind us from the situation where the next couple of years of free cash flow would not worry me if we could free up our offshore cash. It's just really the price at which you do that. I would say, we just finished the board meeting last week. We had a very long conversation about it. It continues to be a serious topic with the board.
Nothing to add to that. That's spot on.
Q&A session. Please stick around if you have time. Join us for lunch. Thank you.