David Paul Morris/Bloomberg News
Carl Bass, the chief executive of Autodesk.
Just as important, it is another sign that the cost of computing has fallen so far that it makes more sense to be wasteful with server power, and harvest the results, than it is to plan everything carefully. It is a triumph of brute force and messes over carefully planned design.
“Simulation is the ability to understand something before you build it,” says Carl Bass, the chief executive of Autodesk. “If it doesn’t cost much to do it, why not do 10,000, even 100,000 different simulations of anything, rather than build a model, judge it and keep going back to the drawing board?”
The idea is to try thousands of different conditions, like temperature, humidity, tensile strength or shape, in just a few seconds. Most of the outcomes will be lousy, a couple of them will probably affirm what a designer thought to begin with, and a few might deliver surprising insights no one had considered. The hope, Mr. Bass said, is that “we give an engineer a greater capacity to understand their product, before they make a million of them.”
The brute force computing model is changing a lot of fields, with more to follow. It makes sense, in a world where more data is available than ever before, and even more is coming online, from a connected, sensor-laden world where data storage and server computing cycles cost almost nothing. In a sense, it is becoming a modification of the old “theorize-model-test-conclude” scientific method. Now the condition is to create and collect a lot of data, look for patterns amid the trash, and try to exploit the best ones.
Brute force computing is behind Google’s successful language translation. By comparing thousands of Web pages in different languages to find patterns, in one year Google was able to discern and refine translation better than linguistic theorists had been able to do with their fancy programs for years. There is brute force in genetics research, too; machines plow though the data looking for novel patterns, which researchers then examine to see if they hold valuable insights.
“As a trained mathematician who is supposed to think about finding one elegant answer, it’s kind of offensive, but brute force works,” Mr. Bass says. “Once you start thinking about the process a little differently, it allows the engineer and the designer to do more interesting work.”
Like many of Autodesk’s cloud-based products, the revenue per customer is low, relative to what the company makes on its older packaged software. Still, the simulation product costs anywhere from $3,600 to $10,000 for a year of use, with higher prices buying more functionality and computing power. That is better than its forays into consumer-type design, like a recent $60 million purchase of Socialcam, a company without visible revenue.
As Autodesk is spending a lot to run its cloud, margins at first may be lower than what the company is used to. Mr. Bass says Autodesk is spending “three-quarters of its resources on the future, compared with 5 percent to 10 percent normally.” That has got to hurt, in a company that just missed its revenue expectations, taking the stock down more than 10 percent.
He also says he has no choice, and thinks competitors like Dassault and Siemens will in a few years be blindsided by cloud computing’s impact on industrial design.
“This is a classic disruptive technology,” he said. “They’ll see it, they’ll understand it, they’ll dismiss it as a toy, and they won’t be able to do anything about it when it takes over.”
The new Autodesk service runs on AWS, not the company's own data center servers.