You climb 170 steps up a series of dusty wooden ladders to reach the top of Hangar Two at Moffett Federal Airfield near Mountain View, California. The vast, dimly lit shed was built in 1942 to house airships during a war that saw the U.S. grow into a technological superpower. A perch high in the rafters is the best way to appreciate the strangeness of something in the works at Google—a part of the latest incarnation of American technical dominance.
On the floor far below are Google employees who look tiny as they tend to a pair of balloons, 15 meters across, that resemble giant white pumpkins. Google has launched hundreds of these balloons into the sky, lofted by helium. At this moment, a couple of dozen float over the Southern Hemisphere at an altitude of around 20 kilometers, in the rarely visited stratosphere—nearly twice the height of commercial airplanes. Each balloon supports a boxy gondola stuffed with solar-powered electronics. They make a radio link to a telecommunications network on the ground and beam down high-speed cellular Internet coverage to smartphones and other devices. It’s known as Project Loon, a name chosen for its association with both flight and insanity.
Google says these balloons can deliver widespread economic and social benefits by bringing Internet access to the 60 percent of the world’s people who don’t have it. Many of those 4.3 billion people live in rural places where telecommunications companies haven’t found it worthwhile to build cell towers or other infrastructure. After working for three years and flying balloons for more than three million kilometers, Google says Loon balloons are almost ready to step in.
It is odd for a large public company to build out infrastructure aimed at helping the world’s poorest people. But in addition to Google’s professed desires to help the world, the economics of ad-supported Web businesses give the company other reasons to think big. It’s hard to find new customers in Internet markets such as the United States. Getting billions more people online would provide a valuable new supply of eyeballs and personal data for ad targeting. That’s one reason Project Loon will have competition: in 2014 Facebook bought a company that makes solar-powered drones so it can start its own airborne Internet project.
BreakthroughA reliable and cost-effective way to beam Internet service from the sky to places lacking it.
Why it mattersInternet access could expand educational and economic opportunities for the 4.3 billion people who are offline.
Key playersGoogle; Facebook
Google’s planet-scale social-engineering project is much further along. In tests with major cellular carriers, the balloons have provided high-speed connections to people in isolated parts of Brazil, Australia, and New Zealand. Mike Cassidy, Project Loon’s leader, says the technology is now sufficiently cheap and reliable for Google to start planning how to roll it out. By the end of 2015, he wants to have enough balloons in the air to test nearly continuous service in several parts of the Southern Hemisphere. Commercial deployment would follow: Google expects cellular providers to rent access to the balloons to expand their networks. Then the number of people in the world who still lack Internet access should start to shrink, fast.
“HARMLESS SCIENCE EXPERIMENT.” That’s what was written on the boxes carried by the balloons that the secretive Google X lab began to launch over California’s Central Valley in 2012, along with a phone number and the promise of a reward for safe return. Inside the boxes was a modified office Wi-Fi router. The balloons were made by two seamsters hired from the fashion industry, from supplies bought at hardware stores.
Project Loon is now much less like a science project. In 2013, Google began working with a balloon manufacturer, Raven Aerostar, which expanded a factory and opened another to make the inflatable “envelope” for the balloons. That June, Google revealed the existence of the project and described its first small-scale field trials, in which Loon balloons provided Internet service to people in a rural area of New Zealand. In 2014, Project Loon focused on turning a functional but unwieldy prototype into technology that’s ready to expand the world’s communication networks.
Loon’s leaders planned to buy their own space on the radio spectrum so their balloons could operate independently of existing wireless networks. But Google CEO Larry Page nixed that idea and said the balloons should instead be leased to wireless carriers, who could use the chunks of the airwaves they already own and put up ground antennas to link the balloons into their networks. That saved Google from spending billions on spectrum licenses and turned potential competitors into allies. “Nearly every telco we talk to wants to do it,” says Cassidy.
Google has also made major improvements to its stratospheric craft. One of the most significant was developing a way to accurately pilot balloons across thousands of miles without any form of propulsion. The stratosphere, which typically is used only by weather balloons and spy planes, is safely above clouds, storms, and commercial flights. But it has strong winds, sometimes exceeding 300 kilometers per hour. Providing reliable wireless service means being able to guarantee that there will always be a balloon within 40 kilometers.
Google solved that aviation problem by turning it into a computer problem. Winds blow in different directions and at different speeds in different layers of the stratosphere. Loon balloons exploit that by changing altitude. As a smaller balloon inside the main one inflates or deflates, they can rise or fall to seek out the winds that will send them where Google wants them to go. It’s all directed by software in a Google data center that incorporates wind forecasts from the U.S. National Oceanic and Atmospheric Administration into a simulation of stratospheric airflow. “The idea is to find a way through the maze of the winds,” says Johan Mathe, a software engineer working on Loon’s navigation system. A fleet of balloons can be coördinated that way to ensure there is always one over any particular area.
The first version of this system sent new commands to Loon balloons once a day. It could find a way for a balloon launched over New Zealand, for example, to dawdle over land until prevailing winds pushed it east and over the Pacific Ocean. Then it would have the balloon ride the fastest winds possible for the 9,000-kilometer trip east to Chile. But that system could only get balloons within hundreds of kilometers of their intended target. For tests of Internet service in New Zealand and elsewhere, the company had to cheat, launching Loon balloons nearby to make sure they would be overhead. In late 2014, Google upgraded its balloon navigation system to give balloons fresh orders as frequently as every 15 minutes. They can now be steered with impressive accuracy over intercontinental distances. In early 2015, a balloon traveled 10,000 kilometers and got within 500 meters of its desired cell tower.
Google has also had to figure out how to make the balloons sturdier, so they can spend more time in the stratosphere. The longer they stay up, the lower the cost of operating the network. However, weight considerations mean a balloon’s envelope must be delicate. Made from polyethylene plastic with the feel of a heavy-weight trash bag, the material is easily pierced with a fingertip, and a stray grain of grit in the factory can make a pinprick-size hole that will bring a balloon back to earth after less than two weeks.
Preventing those leaks is the work of a squad inside Project Loon that has doggedly chased down every possible cause and come up with preventive measures. These researchers have studied balloons retrieved from the stratosphere, pored over video footage of others inflated to bursting on the ground, and developed a “leak sniffer” to find tiny holes by detecting helium. The leak squad’s findings have led to changes in the design of the balloon envelope, fluffier socks for factory workers who must step on the envelopes during production, and new machines to automate some manufacturing steps. Altogether, Google has introduced the first major changes the balloon industry has seen in decades, says Mahesh Krishnaswamy, who oversees manufacturing for Project Loon and previously worked on Apple’s manufacturing operations. Those changes have paid off. In the summer of 2013, Loon balloons lasted only eight days before having to be brought down, says Krishnaswamy. Today balloons last on average over 100 days, with most exceeding that time in flight; a handful last as long as 130 days.
Google has also made many improvements to the design of the Loon balloons’ payloads and electronics. But it still has problems left to solve. For example, Google needs to perfect a way of making radio or laser connections between balloons, so that they can pass data along in an aerial chain to connect areas far from any ground station.
But Cassidy says Project Loon’s technology is already at a point where stratospheric Internet service can be tested at a global scale. In 2015 he aims to evaluate “quasi-continuous” service along a thin ribbon around the Southern Hemisphere. That ribbon is mostly ocean, but it will require a fleet of more than 100 Loon balloons circling the globe, says Cassidy. “Maybe 90 percent of the time,” he says, “people in that ring will have at least one balloon overhead and be able to use it.”
“It was just for some minutes, but it was wonderful,” says Silvana Pereira, a school principal in a rural area of northeastern Brazil. She’s thinking back to an unusual geography class last summer in which pupils at Linoca Gayoso Castelo Branco School could use the Internet thanks to a Loon balloon drifting, invisibly, high overhead. Internet service is nonexistent in the area, but that day’s lesson on Portugal was enhanced by Wikipedia and online maps. “They were so involved that the 45 minutes of a regular class wouldn’t be enough to satisfy their demand for knowledge,” says Pereira.
Her school is only around 100 kilometers from a metro area of more than one million people, but its location is too poor and sparsely populated for Brazil’s wireless carriers to invest in Internet infrastructure. Google’s goal is for Project Loon to change those economics. It should be possible to operate one Loon balloon for just hundreds of dollars per day, Cassidy says, and each one should be able to serve a few thousand connections at any time. The company won’t reveal how much it is spending to set all this up, or even how many people work on the project.
Cassidy is also confident that his balloons will be able to hold their own against Internet delivered by drones (both Google and Facebook are working on that) or satellites (an idea being pursued by SpaceX CEO Elon Musk). Those projects are less far along than Loon, and it’s expensive to build and power drones or launch satellites. “For quite some time, balloons will have a big cost advantage,” Cassidy says. Nevertheless, Google might be hedging its bets with more than just drones: in January it invested $900 million in SpaceX.
Technology is not the only thing keeping 4.3 billion people offline, though. For example, policies in India mandate that telecom companies provide coverage to poor as well as rich areas, but the government hasn’t enforced the rules, says Sunil Abraham, executive director of the Centre for Internet and Society, a think tank in Bangalore. He is also wary of Project Loon because of the way Google and other Western Internet companies have operated in developing countries in recent years. They have cut deals with telecoms in India and other countries to make it free to access their websites, disadvantaging local competitors. “Anyone coming with deep pockets and new technology I would welcome,” he says, but he adds that governments should fix up their patchy regulatory regimes first to ensure that everyone—not just Google and its partners—really does benefit.
Those working on Project Loon are confident the public good will be served. They seem as motivated by a desire to make people’s lives better as by Loon’s outlandish technology. Cassidy’s voice wavers with emotion when he thinks back to seeing the delight of Pereira’s pupils during their Internet-enabled geography lesson. “This is a way of changing the world,” he says.
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
Data analytics reveal real business value
Sophisticated analytics tools mine insights from data, optimizing operational processes across the enterprise.
Driving companywide efficiencies with AI
Advanced AI and ML capabilities revolutionize how administrative and operations tasks are done.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.