FarmedHere_190bJennifer KathrynAn indoor vertical farm.

The world's first robot-run farm is opening in 2017, courtesy of a company called Spread. But in an alternate universe, Alphabet's X unit (formerly Google X) might have been behind it.

At the TED2016 conference in Vancouver, Astro Teller — the head of X — revealed that the secretive moonshot lab once tried to create an automated vertical farming system. 

When X looks for "moonshot" projects to pursue, it follows three steps: find a huge problem that affects millions of people, propose a radical solution for solving the problem, and come up with a reason to believe the technology to create that radical solution could be built.

 "1 in 9 people suffer from undernourishment, and vertical farming uses 10 times less water, 100 times less land than conventional farming," says Teller. Vertical farming, the practice of growing crops indoors on racks that stack atop one another, is a potential radical solution to the problem of global hunger. Take human labor out of the equation and it becomes even more attractive.

X made progress in automated harvesting and efficient lighting technology, and managed to grow some greens. But the Google lab couldn't grow staple crops like grains and rice using the technique. 

As a result, X ultimately killed the project.

During his TED talk, Teller also discussed another X project that was killed before it got off the ground: a lighter than air cargo ship that didn't need a runway to land. That project got the axe because it would have cost $200 million to make the first prototype.

"It's just way too expensive because X is structured with a tight feedback loop of making mistakes, learning, and creating new designs. You can’t spend $200 million on the first data point of a project," says Teller.

Failure is an important part of X's work, according to Teller. "The Silicon Valley hype machine has created this myth of visionaries who effortlessly create the future. Don't believe the hype," he says. "We spend most of our time breaking things, and trying to prove we’re wrong."