The CEO of Autodesk is in his office. We were invited to visit Andrew Anagnost, for the first time as CEO, in his new digs at One Market Street, San Francisco. You might expect the head of a $2 billion-dollar company to have the corner office, something ostentatious, a show-off expanse with sofas, oriental rugs, fine art and a bar. There’s almost none of that. There’s one sofa, no conference table, or giant mahogany desk. Andrew’s desk is a modest stand up. There’s no monitor showing the company’s stock price (“That’s dangerous,” he says), but there are many pictures of his family.
Andrew Anagnost had been at Autodesk for 20 years when Carl Bass (he of no office at all—he shared a cubicle!) left the company rather suddenly and left the board with a worldwide search for their next CEO. Andrew demanded the position. What would an outsider know about the business? How could they understand the transformations Andrew had seen at Autodesk and project how these changes will affect the future?
In his 20 years at Autodesk, Andrew hasn’t changed. The 53-year-old still looks boyish. No gray hairs from the stress that comes with the top office. There is no trace of wear from the thousands of employees answering to him, the board of directors after the next billion dollars, the millions of users, many of whom he may have pissed off by asking them to pay for software they thought they owned forever. It’s all smiles as he greets us, a lack of formality making it more like seeing an old friend than interviewing a big-shot CEO. But beyond that, we see a can-do attitude. It’s his company now. And he has a lot to do with it—and do to it. The bold initiatives he had as VP, his compelling visions of the future, he can now make happen. He has no VPs with dissenting opinions in the way, no one above him to tell him to wait, and no need to wait for consensus.
It’s Andrews first big interview with the engineering media since he became CEO of Autodesk a year ago. In our conversation, we explored Autodesk’s secret strategy for Fusion (why they’re holding back), why SolidWorks was successful (Autodesk not buying them was a big mistake), why Onshape is not successful and won’t be, and so much more.
Buckle up. It’s Captain Andrew at the helm of the Starship Autodesk and we’re about to go to warp speed.
About your vision for design and building, can you address automation?
We've been having lots of conversations about automation, the impact of automation, the future of learning, and the future of work. We've actually been hosting very diverse and interesting groups of people here at Autodesk to dig into these conversations.
You heard from me at Autodesk University – London and at AU last year. We're not freaking out about automation like the rest of the world is. Take the usual example people give of AI and machine learning destroying jobs: truck drivers. Everybody uses the truck driver story, but there's two sides of the truck driver story.
One side of the truck driver story is: all the truck drivers are going to be out of work and the trucks are going to be driving themselves across countries all over the place. Another side of the story is, we've got an aging truck driver workforce. Nobody wants to go into truck driving. There's no new pipeline of truck drivers and we're worried we're not going to have enough people driving trucks. And by the way, we need more people managing the first and last mile of that route. And not necessarily doing the dangerous, tiring job of getting things between the first mile and the last mile.
There are two sides to every story. We don't wade into that kind of dialogue because when we look at our world, and we see it every day, there's just not enough capacity to do what needs to be done. We constantly see people struggling with the cost of doing individual projects [one-offs]. We constantly see people struggling with their ability to take on new projects in a cost-effective way. We constantly see labor shortages in the built world.
There are shortages of talented machinists. There are shortages of construction workers right now. And when you're closing the borders of the country, you exacerbate some of those.
So that was a story about jobs no one wants, and truck drivers is one of them. 911 call operator is one of them too.
Like AECOM facility? We heard they are short 30% of the people they need on every job site.
Yes. And closer to home, in Sonoma [north of San Francisco], after the fires last year, how many houses do you actually see being rebuilt in some of those neighborhoods? Not that many. Houses are not being rebuilt quickly. Some of it may have to do with getting funding, you also hear about shortages of wood, or that wood is now too expensive. In reality, construction companies are struggling to get work crews almost 9 months later. You also can't get someone to design the home because they're too busy. Everywhere you go, you see this problem in the built world: people cannot respond to the demand of what we need built.
We're moving towards a world with nine billion people. We're going to have to build buildings and infrastructure at a pace never seen before. And we struggle to just prepare the infrastructure we have today. How are we going to do this? Where is all the concrete, the steel, the aluminum going to come from? How do you make sure you have enough of these things? And if buildings and materials are in high demand, how do you keep the cost from skyrocketing? We shouldn’t be afraid of losing jobs. We should be afraid of not being able to accomplish what we need while ensuring we live a high-quality life. So, while we at Autodesk aren’t afraid, we also fundamentally understand that the jobs of the future are going to be different.
Different how?
People get freaked out when you say jobs will be different. But one thing I can say for sure will change will be the construction industry. If you look at the first 50 years of manufacturing and the kind of skills that have evolved in that industry, it used to be hand craft skill. In the early days of manufacturing, you had people operating looms or hand-filing and assembling. That's how the construction industry is evolving today. We're moving towards a work flow where construction looks more like the modern factory assembly process and all the things that are related to it. We're actually moving away from a skill base that was used to cutting and fitting things onsite by interpreting drawings.
Take construction drawings, for example.
Drawings are an incredibly primitive way to communicate 3D concepts on a flat surface. If you don't understand the drawings, you have no idea what they are trying to tell you. It's arcane, a second language, if you will. But, when you look at the construction site now, the crew does not operate like those in a manufacturing environment, they are leveraging new technologies. However, the current skillset of the construction worker today is lacking. We need a re-skilling.
Does that mean there will be less people employed in the construction industry?
Probably not in the next 50 years because things don't change that quickly. We're going to move to a world where people are doing more projects and yes, there's going to be fewer people per project. But if you have fewer people per project and your projects are much more optimized, you're using only as much material as you have to. You've come up with design solutions that are actually highly material- optimized. You're going to be able to afford to do more projects. And you're going to have people employed on more projects. This is the world we see.
How much waste is there on a construction site?
30 percent of the material on a construction site is wasted. And that's a conservative estimate. [Autodesk cites a McKinsey construction report.]
Do you think we are heading for a time when those at a construction site won't know how to read drawings?
The people that come into construction are people that are going to look more like manufacturing workers looked 20 or 30 years ago. And they're going to have to have those kinds of skills, such as knowledge of machinery and how to use it. They're going to have to be able to use a 3D model.
What about permits?
There's a little permit challenge that gets in the way of this vision. Most permits still require 2D drawings. They are still flat. Most of the processes are still flat based.
This is why construction will change slower than people want it to because the ecosystem construction operates in has drawing-driven permitting. In some cases, they're permitting on the wrong things. They don't mandate BIM [building information modeling] in any form for the processes. There's all sorts of things that are going to slow down the pace of change in construction.
Are software vendors too far ahead of the construction industry?
This is why I don't worry about jobs going away. How long do you think it would take for that capability to actually percolate through the processes that build a building? Even if we did it tomorrow. Even if we had some super brilliant artificial intelligence that could automate everything.
So just build it and they come. What about AI?
They will eventually come. But we have to ask what problems do we solve initially? I think the problems that we need to focus on in terms of automation for our customers are ‘should you be building geometry or should geometry be built for you based on what you're trying to accomplish.’ With Revit and Inventor, you are mostly using tools to make little pieces of geometry. You're not actually thinking about the design. You have to create this piece of geometry, attach it to that piece of geometry. And when you’re done, you will have something that looks like a building or a phone or a machine. We need to automate the creation of all that geometry. We're moving to a world with disaggregated data in the cloud. We’re able to apply machine learning algorithms to it. Then we can automate the creation of more and more geometry. That leaves people more time to explore the design decisions they're making, optimize the material usage, and plan out the construction process more deliberately. Then we understand the implication of a design decision in terms of how it's going to get made very early in the process, we will reduce the amount of waste in the processes.
We generate the geometry based on the kind of machine they have at their disposal. We generate a geometry for an additive machine if they have one. We generate a geometry that can be created on a subtractive machine. When they change the design objective, the geometry adapts to not only the design objective, but the method of making.
Is this software available?
We've done some projects with it. It's our early explorations in generative design. So we've done some examples with General Motors, a swing arm for a motorcycle, a seat belt anchor. Some things we haven't shown publicly yet.
So, the automation of design itself?
When you automate those things, when you are able to make design and make decisions simultaneously, this is the convergence of design and making. That's when you get this real-time feedback about the implications on the making side, for all the decisions made on the design side. Sometimes you might not like the feedback.
But right now design and making are very separate.
It's still a waterfall process. Take building construction. The architect builds the initial design, then they hand it off to the MEP firm or just the MEP engineer. When the engineer is done, he hands it back and says, ‘hey, I modified that.’ Then they hand it off to the construction firm. The construction firm goes through a pre-construction process and a pre-construction planning process. All of these things are for the most part disconnected and linear. What if you're able to connect the data flow across all those things and actually provide architects instant feedback? Like how to prefabricate 40 percent of a building and leave more budget and time to be creative, to make a more swoopy design on the outside? What if we were able to give you information in real time that allowed you to have more latitude to do something with this design than you ever had before?
Because the basic, individual processes are now happening in an automated fashion? So, architects can be more creative.
Yes. We believe that this kind of augmented interaction with automation is going to allow people to explore more creative options. We also believe it's going to allow them to create less resource intensive designs. That's another thing we're trying to build into the software moving forward; it’s this idea that it's not just about automating the design and make processes, it's about automating the intelligence that really allows them to come up with things that have a much less negative impact on the world than what they've been doing previously.
By using less material?
Less material and less energy consumption. It may also be new forms of making that reduce the amount of waste.
Or putting the right material where it's needed – and nowhere else. Perhaps even fiber-reinforced material?
They're experimenting with all types of materials here. It could be a much more durable material.
You're saying we're just bound by bricks because that's what we're used to making. What about energy?
Energy consumption, energy usage, a design would be more energy-efficient because you'd be able to design in the insulation.
Also think about how much stuff you have to bring to the construction site to make something happen. When you look at the whole process this way, all of a sudden you start saying, “wow, you know what, I can not only reduce the energy of the building over time, I can reduce the energy I deploy to actually build the building.”
(Our interview continues in Part 2)