The world is changing, says Andrew Anagnost, every time we see him. It's his theme at Autodesk Universities held across the globe, and it’s how he had started when we met up with him at his One Market Street office in San Francisco (see Part 1). The change is coming on fast and we need to keep up. The billions of people that will inhabit the Earth in the years to come will need houses, roads, other infrastructure. It can’t happen using your father’s tools. Construction needs to up its tools and its techniques. Andrew, who came from the manufacturing and aerospace (he's literally the rocket scientist) side of industry, thinks construction has a lot to learn from that side. In Part 1, we heard what construction is doing wrong. We about to learn how this can be remedied and how now is the perfect time for it.
You’ve stated principles used in manufacturing. Is it a stretch to apply those principles to building and construction?
Yeah, this is why we're so excited. I think the biggest manufacturing growth market over the next 10 years is in building and construction.
With all this emphasis on AEC, are you steering Autodesk away from mechanical design and manufacturing?
We believe the fastest growth manufacturing market in the world over the next 10 to 15 years is buildings and roads. We're not steering towards more AEC. What we're seeing is manufacturing and construction coming together. Construction is going to change and manufacturing is going to change. They're just coming from different worlds. The manufacturing process today is highly predictable. It's designed to provide high precision, quality and output. But it’s very rigid. Construction on the other hand is low quality, low precision, and designed usually to do one thing. Construction's going to start getting more precision. It's going to start having more of these prefabricated modular components. And construction is going to start getting more customized. Manufacturing is going to start building smaller lots of things for more targeted audiences. People have been talking about this trend for a long time.
But it has never happened. And why hasn't it happened? Because frankly, the production capabilities to make it real haven't existed. The last 10 years have taught us that the next 10 years are going to be very, very different in terms of construction capability. And it's going to change construction and it's going to change manufacturing. They're going to look more like each other and more similar. Manufacturing's going to become more flexible. Construction's going to become less flexible. That's what we're excited about.
So Autodesk is not giving up on mass customization?
No, no, no.
Construction is ready technologically, demand-wise, and product-wise to go through the digitization that manufacturing went through 25 years ago. Manufacturing is beginning to realize that mass customization is around the corner, but they haven't quite figured out which technology is going to support it yet. 3D printers aren't quite there yet, but they're getting there. They haven't figured out exactly what the economics is for them. Big manufacturer right now, a world where dozens of people can 3D print something that you make in a highly specialized way terrifies them. We believe they're going to have to confront that terror, but it's going to have to take a little bit longer because the underlying technology that's going to help them get there is still maturing. But one of the things you've heard us doing is, again, when we talk about design and make coming together. It's going to come together in manufacturing. And we really have bet on this world where you have rooms full of 3D printers that are printing metal objects. And you have robots moving things from one 3D printer to the next. Or moving it from the printer to a subtractive machine, then assembling it into something.
We believe that is the future, and we're creating our design systems of the future. Not the ones that people are using today, but the ones that we believe are going to dominate in the future to function in that world. And we believe construction is going to embrace things that manufacturing embraced a while ago and come forward. Right now the reason you see us talking about construction a lot is because they're ready to digitize. And they're so far behind.
Pardon our skepticism. We have been hearing AEC will be automated for a long time, that automation’s biggest potential is AEC, because all there is to gain, for decades. We understand your optimism, but the chasm is huge. Why doesn’t construction behave like manufacturing? They're slow. They've made progress towards automation, but it’s very slow.
They are, you're right. We've been talking about this for decades.
What's different now?
Let's go back to Buzzsaw and Internet 1.0. What things like Buzzsaw and the technologies we had then were unable to reliably get to the construction site. They were able to get to the trailer, and that was the end of the Internet pipe. They were not reliably able to get into the field.
Mobility has fundamentally changed the conversation around what technology you can get to the field. They buy Android tablets and iPads. They put shock shields around them and they carry them around. This is the big difference between now and 2000. In this age, mobility is making all the difference. So right now, the problem that construction companies and field construction operations are solving isn't the big fancy problem I'm thinking about. They want us to think about the big fancy problem. The problem they're solving is getting the 3D model on the construction site. I want to connect that 3D model to any drawings that I need. I want to be able to capture information, add it back to the in-process project, and communicate to my subcontractors and everyone in real time on the site.
The investment community is now recognizing the potential of automating AEC?
That’s the other things that's different now. There are hundreds of millions of venture capital dollars being poured into this space. If you add it all up, we're already seeing over half a billion dollars being spent on site automation software. Half a billion dollars! It was never like that in the Buzzsaw days. VCs are dumping money in. They were also dumping money in during 2000, but it was a smaller world. And nobody was buying at the level that people are buying now. You walk into an AEC firm now and find their IT budgets being used for either internally building these mobile solutions or they're building off the shelf mobile solutions.
Of course, cellular data networks have become more robust, too.
Yes, that iPad on the construction site is robust enough to bring down megabytes, and sometimes gigabytes, of data.
The idea is to get that capability across the board to all of your customers in the construction industry, right?
Yes. And then move from digitization of the site to making the model, to where people have a conversation. Luckily, we have manufacturing as an example. Manufacturing was a very early adopter of CAD. And especially CAD for not just design processes but for manufacturing processes. They were the first adopters of 3D models, and they basically moved their entire processes to 3D models and started building data systems, PLM, to try to connect model data from design to manufacturing and create digital pipelines. We have the capacity now to build digital pipelines like this [snaps his fingers]. Now we can build a digital pipeline, a digital flow. From a model to a site very quickly, very easily, and move data seamlessly.
What else do you attribute this to?
It's the cloud as a platform. It's the rise of AWS [Amazon Web Services]. It's the rise of Azure. It's the rise of distributed computing. To take a large model and break it apart into small pieces in the cloud costs nothing.
There’s been a platform transformation. The cloud platform transformation is as profound as the rise of the PC. So, when you start looking at the rise of the mobile devices attached to the cloud, we're able to do things we were simply not able to do before. For instance, let's talk about Inventor and Revit. Fifteen years ago when I was running Inventor, we had a customer, Permasteelisa, that built curtain walls.
Curtain walls are a manufactured component of a building. And 15 years ago, Permasteelisa was requesting to have their Revit model communicate with their Inventor model so they could fit my curtain wall onto the building model. They needed a high-precision curtain wall model because it has to be manufactured, ye, they were working with a low-precision building model. They wanted to take the Revit model, feed it into their manufacturing model and go straight to production. There's a lot of challenges to that approach.
The difference in scale. The level of detail required to match the mechanical model?
Bingo! We weren't able to do it 15 years ago. The problem was that, fundamentally, the representation you use for a building information model is very different from the representation you use for a manufacturing model. One is built for high-precision operations and high fidelity. The other one is built for lower precision. An inch is high precision in the AEC world. An inch is a disaster in the manufacturing world. So we had a hard time bridging these two worlds together into a single workflow for this customer. We cobbled something together and we made it work. We built a one-off for Permasteelisa. But today, we have the capability to build a meaningful workflow between Revit and Inventor for a manufactured building and manufactured components that actually works for everyone—because of the cloud. Because we’re moving away from file-based information flow to just database information flow. Everything is different now.
How does the database help with the problem of dissimilar scales, that which you say is a problem with file-based design?
With a database, we are able to move the information with a fidelity required for each environment. You don't have to try to suddenly shoehorn it in. You can easily move a representation of what's necessary with the connection points. You can also deliver the connection points seamlessly to a building information model at a fidelity that makes sense.
You’re freed from this whole notion of file translation and things like that.
It’s like exploding the design information into a database and allowing some high-speed flow between the various types of data that makes it all happen.
We don't have to build one-offs customer by customer by customer.
(Our interview continues in Part 3)