If you want a glimpse of the near-future, go get a box of Cheerios. Open the box and pull out the bag of cereal. Feel the texture of the bag. Note how hard it is to rip apart.

A cereal bag is the closest analog to the material that Google X, the company's high-risk research lab, uses in its Project Loon balloons. These are the balloons that Loon researchers hope will soon be flying in vast armadas in the stratosphere, the layer of the atmosphere 60,000 feet above sea level, far above where commercial air traffic flies. These balloons would beam down Internet service to receivers on the ground, providing connectivity in areas that would otherwise be prohibitively expensive to reach by laying cable or erecting cell towers.

Down on the ground, a single balloon can provide Internet to an area 25 miles in diameter. Put enough of these balloons in a developing country's airspace, and you can blanket rural areas with LTE-level cellular service.

For the past two years, Google has been secretly testing these cereal-bag balloons for months, first in California's Central Valley, then in New Zealand, and eventually at the McKinley Climactic Laboratory, a giant climate-controlled hangar located in the Florida panhandle at the Eglin Air Force Base.

Publicly, Google has dribbled out information about the progress of Project Loon since its 2013 debut. But recently, the company let me behind the curtain, and gave me an exclusive update on their progress in building these balloons. The biggest surprise is that a project that once seemed insane has now moved to the nuts-and-bolts stage. Google isn't worried about the feasibility of beaming Internet down from the stratosphere; they've got that. They're worried about manufacturing better balloons.



When Google first announced Project Loon in June of 2013, it was met with a lot of muffled laughter and deep skepticism. "No, we're not joking," began Engadget's post about the project. "Google is seriously proposing hot air ballon-powered internet access." Bill Gates took a nice potshot at the project, too. "When you're dying of malaria, I suppose you'll look up and see that balloon, and I'm not sure how it'll help you," Gates said. "When a kid gets diarrhea, no, there's no website that relieves that."

Sure, the idea of wifi-producing hot air balloons sounds like a nerd's folly. But there is evidence that connectivity actually does make things better in the developing world. For one, access to real-time price information keeps producers from making bad decisions and/or getting screwed. One canonical study found that increased mobile phone diffusion among fishermen in the Indian state of Kerala "was associated with a dramatic reduction" in seller's price variability (or what economists call price dispersion) as the people selling stuff figured out what the right price should be. Wharton's Richard Jensen, who conducted the study, concluded that "both consumer and producer welfare increased." A later study in Niger found similar results. Other research has found that increased access to connectivity makes families more resistant to shocks from natural disasters, as they can receive support from people outside the affected areas.


Even people who granted that Project Loon's mission was sound had doubts about¬†Google's ability to manufacture¬†the right kind of balloons.¬†Google said it wanted balloons that would stay aloft for 100 days. ‚ÄúAbsolutely impossible‚ÄĒeven three weeks is rare,‚Ä̬†Per Lindstrand¬†told Steven Levy¬†a couple years ago.

But now, after more than 10.6 million miles of test flights and multiple iterations of balloons, Google makes balloons that can stay aloft not just 100 days but at least in one case, 187 days.

This is tremendous progress.

But it's one thing to build a single great balloon, or even a dozen great balloons. If¬†Google really wants to connect billions of people with this technology, they'll need to fill up the stratosphere with thousands and thousands of balloons. And they'll need to do it fast and cheaply. ABI Research released a report this week on Loon's prospects, in which they questioned the economics of the project and concluded that, even if it could be cheap, "It may have a five to ten year project lifetime as the worldwide population continues to urbanize and 4G networks migrate outward.‚ÄĚ


To even contemplate that kind of scaling up, you need someone who understands how to manufacture things.

So Google went to Apple, the best manufacturing and supply chain company in the world, and poached one of their elites.


Mahesh Krishnaswamy did not intend to go into balloon manufacturing.

He'd been a computer scientist and engineer, who loved getting down into the assembly code. That landed him a gig at Motorola, back when the company was still an independent mobile powerhouse and not a mere facet of Google. It was there that he learned the value of manufacturing. "You cannot really manufacture your way out of a poor design but you can screw up a good design if you manufacture it poorly," he told me.


So, he took up the cause of making hardware, which eventually landed him at Apple working on manufacturing of new products. He worked on the first iPad and the Macbook Airs before migrating to procurement and supply chain.

Apple, you should know, has been the number-one ranked supply chain company by the analyst firm Gartner for more than half a decade. While Jony Ive and the design of the products might garner most popular praise, it's really the company's mastery of manufacturing and procurement that make Apple's operation possible.

Krishnaswamy said the job was basically, "How do you take something that is an idea of or a prototype and scale it and build 50 million of those a quarter?"


And then two-and-a-half years ago, he left Apple for Google X, initially with the thought that he'd work on manufacturing across all of the research lab's many products.

Soon, though, he found himself deep inside Project Loon, having the same thought that many Google-observers would have when Loon was announced: "Having done high-volume consumer electronics manufacturing, you get handed down balloons, and it was like, 'What?'"

But "manufacturing is manufacturing," Krishnaswamy maintains. They just had to find the right materials and suppliers. So, they cast about for who made the best high density poly-ethylene (HDPE) bags.


The food industry, as it turns out, has developed most of the best plastic film technology for these purposes. That's one reason that Krishnaswamy ended up exploring the Cheerios bags that he said are the closest in form to the bags Loon calls balloons. Then they found the right manufacturing partner in Raven Aerostar, makers of all kinds of inflatables for research and the military. And they were off.


The Project Loon balloons themselves are 65 feet in diameter‚ÄĒand they are really two balloons, one inside the other. The balloons are inflated with helium, but air can be pumped into and out of the inner chamber, which changes the buoyancy of the contraption and allows it to move up or down. Those altitude changes make it possible to move into different winds, allowing for a modicum of steering.


The biggest problem with balloons, of course, is leaking. And Google's researchers initially struggled with how to prevent leaks. Even a tiny hole a tenth of an inch across could reduce the duration of a balloon flight by 10 days. And just try to find a leak that small on a balloon that's 65 feet in diameter.

So, they now take great pains to avoid creating leaks. They even have special padded socks for use when they walk on the plastic films. They've perfected the manufacturing process for the seals that attach each panel of plastic film to the next.

Google's new balloon manufacturing line.


They've also created more sophisticated tests for the balloons. Generally speaking, "You don‚Äôt know if they‚Äôre going to work until you fly them,"¬†Krishnaswamy said. That's because you can test the materials themselves, but the work you can do with a few square feet of plastic might not tell you enough about a giant balloon made of the same material. There are also ropes that sit atop the balloons that Google calls "tendons" ‚ÄĒ and the interactions between these tendons and the plastic films are difficult to predict, especially in the frigid conditions of the stratosphere.

And that's where the Google's new Florida research hangar has allowed the company to test these balloons in practice. The lab is 250 feet wide, 200 feet deep, and 70 feet tall. It's a massive space where Google's scientists can simulate the negative-60 degrees Celsius temperature of the stratosphere. Researchers dress in arctic parkas and head out into the cavernous hangar to do studies, or watch them from adjacent facilities on high-speed, high-resolution cameras.

"It was like taking a magic carpet and flying up 60-thousand feet and being to see it all," Krishnaswamy said.


They tested the balloons' weakest spots, studied what happened when their balloons burst, and the diurnal temperature cycles the balloons endure.

For the last test, for example, Krishnaswamy said Google manufactured balloons with a known pattern that they could track with high-speed cameras to see the distortions in the materials that the rising and falling temperatures caused. Then, they used computer vision to convert the images into data, which they fed into their models of how the balloons worked. That showed how they could slightly rework the dimensions of the design to relieve the tension caused by the expansions and contractions. Less tension means a bit longer lifespan up there in the air.

That's a lot of work for a little fine-tuning, but the very level of effort they have to expend for small optimizations of the design and manufacturing are a sign of how far they've come from the early days when their balloons could barely stay aloft for a week.



Ever better lab performance is one thing. Next, Google will have to prove that Project Loon can actually be put into practice in the developing world.

Google has diversified its bets on airborne data service, as Steven Levy, who has chronicled Loon since the beginning, notes. They purchased solar dronemaker Titan Aerospace and Skybox Imaging, a satellite-imaging company with intriguing potential.


Facebook, too, has a drone company and a whole Connectivity Lab, which is dedicated to bringing some kind of Internet service to the developing world. Meanwhile, carriers on the ground will continue building out their cell networks to the most profitable places. And all those rural people will continue their mass migration into the cities, as they have been for decades, leaving fewer customers in the hinterlands.

It all makes for a fascinating, competitive environment in which to deploy Project Loon. And that's to say nothing of the interesting regulatory implications of stratospheric projects. Who has legal rights over the stratosphere? And even if a country wanted to enforce stratospheric law, how could they practically do it?

Despite the questions, Project Loon floats onward. In a sign that Google may be ready to flip the switch on manufacturing large numbers of balloons, the company has promised that hundreds of balloons will fly over Indonesia in a test with local cell service providers Indosat, Telkomsel and XL Axiata. They'll act as floating mobile-phone towers dropping LTE from the edge of space. The Indonesia project begins next year.


Loon has transformed from the exclamation point on Google's craziest ambitions into an operational unit in just a few years. Maybe it's not insane to imagine the stratosphere filled with objects beaming down Vine stars to our handhelds on the ground.

"The stratosphere is a new dimension that is unlocked and everyone is looking at this new dimension to use it for communication and delivery," Krishnaswamy said. "I absolutely think the stratosphere is going to play a big role in changing how people interact with each other."