“我曾经走上歪路”|《财富》专访欧特克CEO

当我给安德鲁·安纳格诺斯特打电话时，他还在旧金山市中心的办公室，刚刚结束了一天的工作。我对未能与他面谈表示抱歉，因为我数天前才从湾区回来。

他笑着说：“还好没有面谈，我正穿着健身服，都湿透了。（有电信软件是真好啊。）”

说真的，软件开发商欧特克（Autodesk）的首席执行官安纳格诺斯特没有理由流汗。他的公司致力于为建筑师、产品设计师和电影制作者开发软件工具，赚的钱比以前任何时候都要多，公司2018财年的收入超过了20亿美元，而且其市值也是再创新高，达到了近300亿美元。

安纳格诺斯特担任公司掌门人一职还只有16个月，但他已经对公司所发生的变化感到兴奋不已。今天早上，他希望在拉斯维加斯举行的欧特克大学会议上向客户详细介绍这一策略。以下是他向《财富》杂志透露的内容，为便于理解略作删减。

《财富》：虽然你在去年春天才出任欧特克的首席执行官，但你已是公司的老人了，已经干了20年。上任后哪些事情让你感到始料未及？

安纳格诺斯特：我在公司已经待了很长的时间，但我实际上是公司亚文化的一部分，分属于产品设计和工程团队。这是一份截然不同的工作。此前团队所在的领域有着异常激烈的竞争。我们的竞争对手与欧特克相当，或者比欧特克的规模更大。因此，我没有必要遵循公司的主流文化。

当我升任自己不甚了解的顶级职位时，我并不确定自己能从中学到什么。但是我开始意识到，要让欧特克实现我们心目中的目标，我们还有很多、很多的工作要做。所有这些科技都能够让我们将这些设计融合在一起，并让流程更加紧凑。而这一切将改变我们的公司。

云技术正在改变所有的软件公司。你可以直接与客户互联，并了解哪些方面做的不错，哪些方面做的不够好。为了应对当前这个世界，并获得繁荣发展，企业得创建一个全新的神经系统、肌肉组织和结缔组织。要实现这一点，我们还有很多工作需要做。

我们随后做出了多个决定，并一度认为我们要做就得做云端产品。公司仍有一些老旧设备[基础设施]需要升级。我们实际上同时改变了业务模式和技术基础。要成功地做到这一点很难。

很多人都听过欧特克，但不知道公司是做什么的。你如今怎么形容公司？

如今，欧特克致力于为那些制造业从业者开发软件，例如桥梁、建筑、产品、道路、电影、游戏。随便说一个行业，欧特克都有一款针对该行业的设计和制造软件，例如土木工程师、机械工程师、电影艺人、结构工程师。这些人使用我们的软件来设计和创造人们周围的世界。

我们通常参与得最多的是其设计流程——例如它的外观，有什么功能等。但由于世界变得越发自动化，而我们又能向所有人交付超级运算能力，因此我们能够越来越多地参与一些深层次的流程，例如他们如何建造建筑以及制造产品等。我们涉猎的内容更深了。如今，我们是一家“设计和制造”软件公司。

目前你打算重点发展的领域都有哪些？以及不再重点关注的领域？

我们正专注于尽全力进入建筑行业，因为建筑行业是经典机遇之一，其中，云、移动技术已经做好准备，而且客户也已经做好了准备。建筑公司说：“我需要通过数字化来保持竞争力。”如今这方面的产品已经出现。因此我们斥巨资投资这一领域，希望把制造领域的方法引入建筑行业。我们发现，建筑市场开始朝着飞机、汽车等行业在数年前所前行的方向迈进。今后，人们建造建筑和制造飞机的模式都是一样的。

每一个建筑项目都是一次性的，但这些项目如今不仅采用了预制构件结构，同时还在这一过程中使用了更多的工业流程。这并不是选择的问题。未来二、三十年需要完成的项目数量正在上升，但能够完成这些项目的人力和物力却很难跟上。我们必须在从事这些事情时实现更高的效率，更低的浪费。建筑业如今是浪费最为严重的行业。

欧特克的服务对象涵盖各大规模的项目，但公司业务的重中之重还是大型项目——垂直建造的建筑。但技术创新将逐渐向供应链后端延伸，也就是向小项目延伸。

建筑业领域的技术和客户目前还没有完全到位，其背后的支撑是制造业。3D打印技术和微工厂的崛起让建筑行业发生了改变。这需要为不同的制造方法采用全新的设计方式。因此，我们正在开发适用于小规模设计流程的全新设计工具，以及高度定制化的产品。大多数3D造型工具被用于支持特定类型的制造方法；当前的范例是基于3D参数化特征的造型。借助新工具，人们便可以获得新的可选方案。

公司的AEC（建筑、工程和建造）业务约占业务总量的40%-50%。另外30%-40%为制造。其余的是媒体和娱乐。

能介绍一下公司从许可软件销售到订购的业务模式转型吗。

有人认为，同时做这两项业务的转型可能已经给公司带来了不必要的伤害。然而，我们已经做出了决定，而且了解要做到这一点需要时间。我们深知，我们必须做出改变，要么整个转型就无法实现。如果我们真的希望向未来迈进，那么所有的业务都必须以未来为目标。否则，我们将变成一无是处的大怪物。

令人欣慰的是，对于我们这个行业来说，产品的采纳需要很长一段时间。即使Adobe改变了其业务模式，它也于随后开始打造云产品。Creative Cloud的云产品并不多，而且要达到Office 365的那种程度还有很多工作要做。

公司股价在2016年一路高歌。

人们终于意识到我们有能力做这件事情。

他们以前不这么认为吗？

这一点我不确定。我们的业务有90%都是间接业务。投资者曾经对我们是否有能力完成转型持很大的怀疑。如今他们不再怀疑了。

让我们先说说Adobe。我情不自禁地想到了那家公司，它成立于1982年，而且也是一家已经向云端转型的硅谷公司。但Adobe的市值是欧特克的5倍。

这是时机的问题。我们正朝着“设计与制造”转型。Adobe已经是一家“创造和营销”公司。他们对Omniture的收购标志着其营销云的开端。如今，他们又收购了Marketo，借此扩大了营销云业务。在创意业务和营销业务中，我们把创意业务拿出来，并面向世人发布。随着我们公司业务的发展，公司创意套件的价值会不断显现。Adobe目前在同时运营着这两套业务。

欧特克去年的营收达到了20亿美元，对于很多公司来说是个不小的数目，但在硅谷软件企业中依然不算高。是什么让你做出了退步，还是说以这种方式来衡量目标市场是错误的？

建造业今后将成为公司的一项数十亿美元的业务，这是毋庸置疑的。这个行业是巨大的，而且正在迅速向前发展。就制造行业业务而言，我们在未来五年内可能能够实现其规模的翻番。我们在这一领域花了很多钱。在建造行业市场，我们可能已经投了15亿美元，而且市场中小公司收入总和已经超过了5亿美元。这是个实实在在的机遇。

公司的服务对象遍布全球。会受到什么宏观趋势的影响吗？

公司整体业务目前在各个地域都做的不错。这有点不同寻常。建造技术最大的市场依然在美国。下一个最大的市场将是欧洲，因为这一地区拥有一些相当先进的总承包商。在我们看来，当前的全球市场十分稳健。

让我们来探讨一下人工智能，它已经进入了科技行业的各个方面。它是如何与欧特克的业务进行结合的？

从根本上来看，我们将其称为自动化。我们的首要任务是将“设计”与“制造”更为紧密地结合起来。

以建造流程为例。建一栋楼有着非常特别的流程。一名建筑师先提出设计构想，然后进行设计，例如螺旋形大楼，然而交由承包商来建造：“也就是弄清楚具体结构是什么样的。”我们正在实现建筑师与总承包商之间对话的自动化，并利用机器学习洞见来了解建造这一建筑都需要什么。如果在这一过程中能够尽早地获取这一信息，承包商就可以降低无法在预算内完成建造工作的风险，并提升可选择性。位于后台的计算机会筛选和选择信息，为选项分析提供助力，也就是如何解决特定的问题。目前人们并没有采取这种方式，但这正是我们努力的方向。

例如，我们进行了大量的运算来弄清楚某个建筑是否能够站立，或者建筑师可以让大楼最大化地接受自然光照。我们非常希望能够为建筑师提供这些功能。

我们最近与通用汽车共同开展了一个项目，利用3D打印来重新设计座椅支架。这是人类所无法想象的，但它直接满足了设计要求。我们能参与的领域非常广泛。

你是否对媒体报道的人工智能的负面影响感到担忧。

我们对自动化持有乐观看法。当我们审视周围的建成世界时，我们看到的是产能问题。我们没有足够的人手和时间来建造所有所需建筑。全球中产阶级仍在继续崛起。我们不能以整个地球为代价，建造所有所需的建筑。我们发现，自动化这种方式既能够让建筑得到更好的建造，同时也大大减小了对世界的负面影响。我们相信，越来越多的人将进入建造行业，因为更多的资本将进入这一行业。人类的创造力是宝贵的。自动化将接手重复性的任务。

你常驻旧金山。很多湾区科技公司今年因为越过道德界限或伤害其客户和社会而受到抨击。你如何看待最近的事件？我们需要相关法规吗？

在我看来，不能放任技术渗透而不加监管。一个比较严重的问题是，美国目前没有全国性的隐私立法。我们需要这种立法，也需要对如何以一种可接受的方式使用个人数据以及如何获得这方面的许可进行定义，并创造公平的环境。目前，大家采用的都是基于广告的业务模式，而我们在其中歪曲了这些科技的某些效益。欧洲通过GDPR的实施在一定程度解决了这一问题。加州也有类似立法。但如果每个州的隐私保护法律都各不相同，这对于公司来说无异于噩梦。我们需要全国性的立法，但我们要的并不是摆设，而是行之有效的立法。

我在不经意间注意到，你的成长道路十分有意思。你之前曾从高中辍学，但如今从加州州立大学北岭分校拿到了机械工程学位，然后从斯坦福大学拿到了太空工程硕士和博士学位。你为什么又重新接受正规教育，而且恕我冒昧，为什么还要自虐一把，参加十分耗时的博士项目。

尽管年轻时我没少惹麻烦，而且还辍了学，但我一直痴迷于做某些事情。我希望参与太空项目，因此走上了歪路。但后来我又回归正轨，而且再次开始努力工作，我开始觉得学习很有意思。我是否能在这方面有所建树？能否成为影响力人物？之所以拿博士学位，是因为我想赶上别人，并获取大量的知识，从而成为一个有影响力的人。当时我对于并行计算非常感兴趣。此外，我真诚地感受到我对父母和家庭的那种责任，因为我在年轻时给他们带来了不幸，但他们的痛苦如今也得到了回报。

你还称自己为“太空怪人”。

我的职业生涯始于洛克希德，然后供职于美国国家航天局埃姆斯研究中心。那是我儿时的梦想，我对此感到非常幸运。但我意识到，我很难在大型机构中完成某些事情。我看到很多人一直都在为一个实际上从未完成的项目而工作。这对于我来说是一个十分残酷的现实——将自己的一生花在那些无法发生的事情上。这一点也促使我进入了软件行业。

如今，你的工作就是定期推出云软件。

我们推出了众多的产品。（财富中文网）

译者：冯丰

审校：夏林

When I call Andrew Anagnost at his office in downtown San Francisco at the end of a long work day, I apologize for being unable to speak with him in person, having traveled through the Bay Area mere days before.

“It’s a good thing we aren’t talking in person,” he says, laughing. “I’m wearing my sweaty gym clothes! ” (Thank God for telecommunications software.)

The truth is Anagnost, CEO of software maker Autodesk, has little reason to perspire. His company, which makes tools for architects, product designers, and filmmakers, is making more money than ever before—more than $2 billion in fiscal 2018—and is more valuable than it ever has been, worth nearly $30 billion.

Anagnost has held Autodesk’s top job a mere 16 months, but he’s already excited about the changes underway at the company—a path he plans to articulate to his customers this morning at his Autodesk University conference in Las Vegas. Here’s what he had to say to Fortune, lightly edited and condensed for clarity.

Fortune: You took the top job at Autodesk last spring, but you were a 20-year company veteran. What did you find that you didn’t expect?

Anagnost: I’ve been at the company a long time, but I’m actually part of a subculture at the company—the product design and engineering team. It’s very different. The space that team competes in is highly competitive. Our competitors were as big or bigger than Autodesk. So I was not necessarily of the mainstream culture of the company.

I’m not sure I learned anything when I ascended to the top job that I didn’t already know, but what I came to appreciate was how much work there was to do to become the company we wanted to be. All of these technologies are allowing us to bring together the design and make processes closer together. And that’s going to change our company.

The cloud is changing every software company. You have instantaneous connectivity to your customer—what you do right, and what you do wrong. You have to build a whole new set of nervous systems, musculature, connective tissue to respond to and thrive in that world. That still requires a lot of work to get there.

We made several consequential decisions at one point that we weren’t going to make any products that weren’t cloud products. There’s still a legacy [infrastructure] that needs to be undone. We’ve actually changed our business model and our technological underpinnings at the same time. It’s really hard to do successfully.

A lot of people have heard of Autodesk but may not know what it does. How do you describe it today?

Today, Autodesk makes software for people who make things—bridges, buildings, products, roads, movies, games. You name it, Autodesk has a piece of software for how somebody designs and makes it: civil engineers, mechanical engineers, film artists, structural engineers. These people use our software to design and make the world around you.

Classically we were most involved in their design process—what it looks like, what functionality it has. But more and more, because the world is becoming more automated and we can deliver supercomputing power to anyone, we’re actually able to get deeper into the process of how they build the building, or manufacture the product. We’ve moved deeper. Now we’re a “design and make” software company.

Which areas of the business are you trying to emphasize? Deemphasize?

We’re focused on driving greatest penetration in construction. It’s one of these classic opportunities where the technology—the cloud, mobility—is ready. The customers are ready; construction companies say, “I need to digitize to stay competitive.” And the products are ready. So we’re doubling down in a big way. We want to bring manufacturing methods to construction. We see the construction market starting to move in the same direction as airplanes and cars moved years ago. You’re gonna make a building the same way you make an airplane.

Every building project is a one-off, but they’ll now not only be prefabricated but use more industrial processes to do things. This isn’t optional. The amount of projects that need to be done in the next 20, 30 years is going up but the capacity to do them—humans, materials—is not there. We have to move to a higher efficiency, lower waste way of doing these things. It’s the most wasteful industry today.

Autodesk serves the smallest projects to the largest, but our biggest emphasis is for the biggest projects—vertical construction buildings. But technological innovation will all move down the supply chain to smaller projects.

Coming up behind construction, where the technology and customer isn’t quite ready yet, is manufacturing. It’s going through change with the rise of 3D printing and microfactories? That’s going to require a whole new approach to designing things for different types of manufacturing methods. So we’re working on completely new design tools made for small design runs and highly customizable products. Most 3D modeling tools were built up to support certain types of manufacturing methods; the current paradigm is 3D parametric feature-based modeling. With new tools, you can unlock new optionality.

Our AEC business—architecture, engineering, and construction—is about 40% to 50% of our overall business. Another 30% to 40% is manufacturing. And the rest is media and entertainment.

Tell me about your business model transition from licensed software sales to subscriptions.

One could argue that doing both at the same time may have caused us unnecessary pain. But we made a decision and knew it was going to take time. We knew we had to change or this whole transformation wouldn’t work. If we really wanted to move into the future, everything had to move into the future. Otherwise we’d be some chimeric monster that didn’t do anything well.

Thankfully, it takes a long time for products to be adopted in our industry. Even Adobe changed its business model, then started building cloud products afterward. There wasn’t a lot of cloud in Creative Cloud, and it’s a long way away from looking like Office 365.

Your stock has been on a tear since mid-2016.

People finally believed that we could do it.

They didn’t before?

I’m not sure they did. Our business was 90% indirect. Investors were highly skeptical that we could get through the transition to the other side. Now they’re not.

Let’s go back to Adobe for a second. I can’t help but think of that company, which was also founded in 1982 and also a Silicon Valley company that’s made the cloud transition. But Adobe is worth five times as much as Autodesk.

It’s a timing thing. We’re on our way to be a design-and-make company. Adobe is already a create-and-market company. When they bought Omniture, it was the beginning of their marketing cloud. And now they’ve bought Marketo, which adds to it. The creative business and the marketing business, which takes the creative business and puts it out there. As we grow our businesses, you will see our creative suite become as valuable. Adobe has both sides of the business up and running right now.

Autodesk generated $2 billion in revenue last year, impressive for many companies but small in the world of Silicon Valley software companies. What’s holding you back, or is that the wrong way to think about your addressable market?

Construction’s gonna be our next billion-dollar business, there’s no doubt about it. It’s huge and moving quickly. Manufacturing, we’ll probably be able to double it in size in the next five years. Lots of money is being spent. In construction, there’s probably a billion and a half of investment going into that market, and it’s already generated over half a billion dollars in revenue across all these small companies. The opportunity is real.

The businesses you serve span the globe. Are there any macro trends affecting it?

Our whole business is doing well across all geographies right now. It’s kind of unusual. The largest market for construction tech is still the U.S. The next largest will be Europe because it has some fairly advanced general contractors. We’re seeing a fairly robust global market right now.

Let’s talk about artificial intelligence, which is seeping into every facet of the tech industry. How is it working its way into Autodesk?

We talk about it fundamentally as automation. The big thing for us is to bring “design” and “make” closer together.

Let’s look at the building process. Building a building is a very serial process. An architect comes up with a design, they’ll make a design—say it’s spiral—and dare a contractor to build it: “Figure out how the concrete structure is.” We’re automating the conversation between the architect and the general contractor, using machine learning insights to figure out what it will take to make that building. Get them early in the process, you decrease the risk that you can’t build it within budget, and increase optionality. The computer sits behind the scenes, picking and choosing information, facilitating option analysis—how they might solve a particular problem. It doesn’t work this way now, but it’s what we’re working on.

For instance, we do a lot of crunching to figure out if a building is going to stand, or if an architect can maximize the light exposure of a building. What if we’re able to provide architects with options like that?

We recently did a project with General Motors and redesigned a seat bracket as one 3D-printed part. It doesn’t look like anything a human would come up with, but it instantly matched the design requirements. We have a pretty broad space here to play.

Are you concerned about some of the negative aspects of A.I. that have been reported?

We’re automation optimists. When we look at our built world, all we see is a capacity problem—there’s not enough people or time for all the things that need to be built. The middle class is still rising in the world. We can’t build everything that needs to be built without destroying the planet. We see automation as a way to build all these things, better, and with a lot less negative impact on the world. We believe more people will be employed building things because more money will go to building things. Human creativity is valuable. Automation will take away repetitive tasks.

You’re based in San Francisco. A lot of Bay Area tech companies have taken a hit this year for overstepping ethical boundaries or otherwise doing harm to their customers or society. How do you feel about this recent narrative? Do we need regulation?

I don’t believe in unfettered technological proliferation without some kind of rational oversight. One serious problem is we do not have national privacy legislation right now. We need it. We need to define and level the playing field for what is acceptable for how we use people’s data and how we ask for permission to use it. Our ad-based business models right now, we’ve distorted some of the benefits of these technologies. Europe got it partly right with GDPR. And we have California. But it’s every company’s nightmare to have every state set a different version. We need real national legislation. Not some toothless thing; something real.

On a lighter note, you’ve had an interesting path to the CEO job. You dropped out of high school but now have a mechanical engineering degree from Cal State Northridge and then a Masters and Ph.D in Aerospace Engineering from Stanford. What made you return to formal education and, if I may, engage in the self-flagellation of a lengthy doctoral program?

When I got into trouble as a young man and dropped out of things, I always had a passion for doing certain things. I wanted to be part of the space program and just fell off the rails. When I got back on the rails, though, and started working hard again, I started having fun learning. Can I be good at this? Can I make an impact? The Ph.D thing—you want to catch up and gobble up knowledge and make an impact. I was very excited by parallel computing then. And I honestly felt an obligation to my parents and family that the grief I gave as a young man, their suffering was not for naught.

You’re also a self-proclaimed “space geek.”

I started my career at Lockheed and went to work at NASA Ames. It was my childhood dream; I was very fortunate. But I realized it’s really hard to get things done in large bureaucracies. I saw a lot of people who worked on projects that never actually got done. That was a sad realization for me—spending my whole career working on something that might not happen. And that launched my career into software.

And now you launch cloud software for a living, on a regular basis.

We launch a lot of stuff.