Astro Teller, head of the Google X lab, wearing a pair of Google Glass Astro Teller, head of Google X. Flickr/Techcrunch

Not all ambitious projects can materialize.

Google X, a company responsible for "moonshot" projects under Alphabet, works on making pretty outlandish ideas a reality. The most ubiquitous of which is its driverless car project. But as Astro Teller, the director of Google X, wrote in a Medium post: "the moonshot factory is a messy place" and not everything that is dreamed becomes reality.

"We killed over 100 investigations last year alone," Teller wrote in the post. Here are two:

Vertical farming

Google X attempted to tackle the problem of undernourishment by created an automated vertical farm, but the project was killed last year.

The perk of vertical farming, as Teller puts it, is that is uses less water and 100 times less land to grow food. Since it can be set up anywhere, the food doesn't need to be transported long distances.

Google X made progress on automated harvesting and efficient lighting, but couldn't get "staple crops" like grain and rice to grow via vertical farming. As a result, the project was killed.

"If someone comes up with a dwarf species of rice, let us know — because that might crack the puzzle!" Teller wrote in the Medium post.

A Japanese lettuce production company called Spread is actually working to open its vertical farm in 2017, but it's not growing the staple crops Google X mentioned.

Buoyant cargo planes

The moonshot division also attempted to build more buoyant cargo planes that wouldn't need a runway to land. The hope was that making these cargo planes lighter would lower the cost of shipping, reduce their carbon footprint, and increase their speed. Teller referred to the project as "a real moonshot."

Google X actually did make technical breakthroughs (that Teller didn't list) that would make it possible to build them. But they found it would cost around $200 million to pull off.

The problem with that is it would cost $200 million for Google X to learn if they were, in fact, on the right track. That would be a lot of money to spend to figure out if it could even work.

"If a project has an Achilles heel, we want to know up front, not way down the road," Teller wrote. "So we killed this project."

A tweak to driverless cars

google car driverless self-driving Google

Obviously Google X never killed its driverless car project, but they did take it in a different direction after discovering their original idea didn't hold up.

The original plan for driverless cars was to have the car "do almost all" the driving and allow for a person to take over in emergency situations.

"[It] was a really bad plan. It just wasn’t safe," Teller wrote.

That's actually what led Google to invest in making fully autonomous cars. It's for this reason that the Google cars aren't equipped with things like a steering wheel and a break pedal — humans are not meant to ever intervene with what the cars are doing.

The federal government actually issued a statement last week ruling that a human driver doesn't need to be in the front seat for Google's driverless cars because they are fully autonomous.

Read the original Medium post here.