Skip to main content

Alphabet no longer wants to blanket the earth with internet balloons

Alphabet no longer wants to blanket the earth with internet balloons


Project Loon gets a new direction thanks to a machine learning breakthrough

Share this story

project loon in hangar

Alphabet is changing the course of Project Loon, its plan to cover the Earth with a “continuous stream” of balloons that beam internet to the ground. Astro Teller, the head of the company’s experimental wing, announced today that the Loon team has learned how to control the navigation of these internet balloons so well that they can now concentrate them in areas of need.

The original plan for Loon looked like a reversal of the way our phones connect to cellular towers as we move around in our daily lives. This constant stream of balloons would always be moving over us, 20 kilometers overhead, allowing devices on the ground to connect to each passing “tower.”

It could let them work with dozens of balloons, instead of hundreds

But Teller writes in a blog post that the Loon team has used machine learning to wrangle more control over where the balloons go, and where they stay. The team “exceeded even their own expectations for how well their smart software algorithms can help their balloons navigate the globe,” he writes. The result of that breakthrough, he says, is that the team can now cluster balloons in certain areas for months at a time. Teller says this should shorten the timeline for using Project Loon to provide balloon-powered internet for remote areas around the world.

Taking a more targeted approach will reduce the total number of balloons needed for the project (from hundreds to just dozens) as well as the time required to move those balloons to specific coverage areas, according to Teller. He also says this will help the company get greater value out of each balloon, and increases the chance of the project becoming profitable. If that winds up being the case, it might be the most important takeaway from this change.

Alphabet has lost billions of dollars on its moonshot projects, and profit (and cost) have become enormously important under Ruth Porat, the company’s chief financial officer. One of her tasks when she was brought on board last year was to reign in spending on experimental projects that weren’t making money. Since that time the company has killed off its modular phone project (Ara) and its solar-powered drone project (Titan), while its drone delivery and fiber internet projects have struggled.

Porat isn’t just a project executioner, though. Google’s former self-driving car project was spun off into its own standalone company called Waymo under her tenure. And while it’s certainly not a business yet, it’s at least still alive. With that in mind, Loon’s new direction should be able to keep the idea afloat for a while longer.