Recently I sat down with Dmitry Grishin, the billionaire co-founder of mail.ru and head of the world’s largest venture capital fund devoted solely to robots, Grishin Robotics. As you might expect from someone who has put $25 million into robotics startups, he’s bullish on the potential of robots (as he defines them) to become an area in which fortunes will be made in just the next few years.
Here are Grishin’s arguments for why now is the time for the consumer robot industry to take off (followed by the reasons he might be wrong).
We live in an era of self-driving cars, robotic vacuum cleaners, and ever more autonomous military drones. That’s because innovation comes relatively quickly and cheaply nowadays. “Let’s assume you wanted to build a robot 20 years ago,” Grishin says. “You would need to invest several million dollars to build one robot, and it might take three years.” Costs and timescales like that just don’t fly for modern VCs accustomed to putting their money in software, says Grishin.
The biggest improvements in robotics have come from companies outside the field, says Grishin. “Right now because of smartphones, the price of components [useful for robots] is 1% what it was,” says Grishin. “Most of the components in smartphones are same ones you need in robots—sensors, cameras, batteries, processors. The biggest difference between now and 20 years ago is that the components have become cheap.”
So much of the basic infrastructure of an app or web service has already been built and is available for free or at a low cost—from hosting on Amazon to the collaborative software development at the Apache web server—that launching startups is now more about ideas than the technology itself. The same thing is happening in hardware, says Grishin, as basic components (like the open-source microprocessors Arduino and Raspberry Pi) and libraries of code (like the Robot Operating System) begin to reach millions of hobbyists and professionals.
“It’s very important you have a lot of blocks to easily combine, and you don’t spend too much time to build any of them from scratch,” says Grishin.
“In the past, prototyping was really hard,” says Grishin. Hardware companies would build a prototype, show it off, then spend another year or two creating the next prototype. “But now, because of 3D printers and good 3D software, you can do prototyping much quicker,” says Grishin.
In software, it’s easy for a handful of engineers to build something with massive reach—just look at Instagram, which had but 16 employees when it sold to Facebook for $1 billion. Thanks to better components and easier prototyping, Grishin is now seeing teams consisting of just a handful of engineers launching successful robotics startups, he says.
“I give speeches and say ‘guys, it’s possible to start hardware companies,’” says Grishin. “For most it’s not obvious—people still think that hardware is initially a multi-million-dollar investment. It’s changing, but still there’s this mindset that hardware has to be very big.”
Presently, there are hundreds if not thousands of incubators and accelerators designed to help launch software startups, but only a handful aimed at hardware startups. The first, HAXLR8R, launched in 2012. Since then, others have followed, including Highway1 in San Francisco and the R/GA Connected Devices Accelerator. Even big companies like Nike and Foxconn are launching hardware accelerators.
Two accelerators that focus specifically on robotics are AlphaLab Gear and Grishin’s own Bolt. Grishin says he doesn’t see other accelerators as competition—instead, he’d like to see more, so that several startups would compete to address a need with robotics, rather than just one or two.
The robots that we already rely on—automatic doors, cash machines, even the Roomba vacuum cleaner—are good at what they do precisely because the task they are designed for is relatively straightforward. This, says Grishin, is key to turning robots into a real business.
“What I’ve found is in the robotics niche, you have a lot of people who overcomplicate the situation,” says Grishin. “They start to build complicated universal robots. I think Japanese companies are heroes at this; they just build a mass of technology. But there’s no real way to use that in real life. I’m the opposite. I remember from the early internet that it’s important to take technology that already exists and focus on an area where you can solve real problems.”
It’s not just prototyping that’s become easier for hardware startups—it’s also possible to manufacture small runs of goods quickly and cheaply in China in a way that simply wasn’t possible 10 or 20 years ago, says Grishin.
Plenty of smart things have been said about how cloud computing can grant robots access to huge amounts of computing power. This is good for accomplishing difficult tasks like processing what a robot is “seeing” without having to put all that computing power into the robot itself. But Grishin has other notions about how cloud computing could make robots better. For example, robots can learn from each other’s mistakes.
“If a robot can learn something, you have a network effect,” says Grishin. “The more robots you have the on the field, the more they can teach each other.” For example, while iRobot’s Roomba vacuum cleaner isn’t very smart on its own, Grishin argues that if all those little bots were sharing data to a centralized database, iRobot would be able to examine all the situations Roombas have collectively encountered. It could then use that data to re-program them to be more effective.
It used to be that every gadget in our lives needed its own control panel and user interface. But as these devices become internet-connected, it makes more sense to simply control them through our smartphones—a system that most fitness monitors and other wearable technologies rely upon. Plus, it saves manufacturers from having to build this functionality into the device itself. This, says Grishin, saves “a big part of the cost” of robots.
Grishin says that since he started his robotics investment fund about a year and a half ago, he has received more than 600 applications from startups. Many told him that they had despaired of receiving funding, with so many VCs focusing on software rather than hardware.
One of Grishin’s investments is in a company called Double Robotics, which sells a telepresence robot allowing workers to maintain a physical presence in an office. It costs $2,500, whereas competitors’ models go for up to $70,000. Grishin believes that “about as much as a Macbook” is the maximum anyone will pay for a robot if it’s going to reach a mass audience. Because of all the factors listed above, that’s now possible.
“I definitely believe this will be next trillion-dollar market,” says Grishin. “But I don’t know when.” Many investors have been wrong before about robots—open-source robotics pioneer Willow Garage shut down in January, for example.
And the results don’t always come quickly enough to fulfill investor’s expectations. “People are usually right about what new technologies will become businesses, and the one thing they’re wrong about is the timing,” adds Grishin.
The simple fact is that hardware, which must contend with the unpredictable physical world, is much more of a challenge than software, which operates in the knowable environment of a computer.
Take, for example, the problem of energy and storage. The human body, and indeed pretty much every organism, is remarkably efficient about how it uses energy. Robots aren’t, especially when you start looking at things like their “weight to power ratio,” or the efficiency with which they turn stored electricity into movement.
“In memory, processors, and sensors you have Moore’s law,” says Grishin, referring to the trend first described by Intel founder Gordon Moore, in which silicon-based microprocessors double in processing power on average every 18 months. But batteries and motors aren’t on anything like the same trajectory for improvement.
Both batteries and motors are critical in another big industry, however—electric cars. So it’s possible that, just as cell phones have laid the groundwork for some aspects of the imminent robot takeover, innovation in electric cars could bring us even closer to the point where there are few jobs left for humans to do.