Google Loon’s Internet Balloons Come Back To Earth After A Decade In The Stratosphere

After a journey of a decade, what started as Project Loon by Google is no more. Promoted as a way to bring communications to the most remote parts of the globe, it used gigantic, high-altitude balloons equipped with communication hardware for air to ground, as well as air to air communication, between individual balloons. Based around LTE technology, it would bring multiple megabit per second data links to both remote areas and disaster zones.

Seven years into its development, Loon became its own company (Loon LLC), and would provide communications to some areas of Kenya, in addition to Sri Lanka in 2015 and Puerto Rico in 2017 after Hurricane Maria. Three years later, in January of 2021, it was announced that Loon LLC would be shutting down operations. By that point it had become apparent that the technology would not be commercially viable, with alternatives including wired internet access having reduced the target market.

While the idea behind Loon sounds simple in theory, it turns out that it was more complicated than just floating up some weather balloon with LTE base stations strapped to them.

The (Ba)looney Part

A NASA super pressure balloon at altitude.

The balloons that Loon used were manufactured by Raven Aerostar, from their Super Pressure balloon series. These are recommended by Raven for mission profiles such as scientific data collection, reconnaissance, remote communications and surveillance. Such super-pressure balloons (SPB) are aerostatic balloons which keep the volume of the balloon constant even when the external (ambient) pressure changes. This feature allows the balloon to stay at a fixed altitude for an extended duration, which was ideal for Loon’s mission profile in the stratosphere.

Made from 0.076 mm thin polyethylene, each balloon is 15 meters across and 12 meters tall when its two inner sections are inflated with helium and air respectively. To this balloon, a 10 kg payload box is attached which contains the LTE antennae, control system and the rest of the communication system, which appears to be the Ubiquiti Rocket M2. Ground stations used the Rocket M5, with both the ground stations and the balloons using a custom patch antenna to facilitate the long-distance communication.

This package is completed with batteries and an array of solar cells, which allow the system to operate day and night for as long as the balloon retains its altitude. This tends to be a maximum of 100 to 150 days, after which the balloon is either gracefully deflated above a suitable landing zone, or in the case of a more rapid, unscheduled descent, captured by a parachute system that’s attached to the balloon. Loon’s goal was to re-use or recycle as much of each balloon and hardware as possible.

Mission Profile

The Project Loon autolauncher system in action.

While initially Loon balloons were launched manually, this process was eventually automated with an autolauncher system. After the balloon has been inflated and makes its way into the air, it faces the challenge of moving to where it’s needed. While Project Loon could have used dirigibles (‘blimps’) which would have including a method of propulsion, they instead chose to use the air currents in the atmosphere, rather than fight them.

To this end, Loon used NOAA-provided wind data for the stratosphere (starting between 10 and 20 km altitude). Each balloon would be directed to a layer where the air current was headed in the right direction. Over time this would allow the balloon to reach its approximate location. With no propulsion method, the only way for the balloon to stay within the area was to constantly change between wind layers by pumping air in or out of its air compartment to change its overall buoyancy.

Loon balloons during laser-based communication testing in 2016. (Credit: Google)

Each balloon would communicate with ground stations, both for control and for the internet link which subscribers on the ground could then access via an LTE link. Communication between balloons within each reach was also a common feature, with subscriber’s internet link hopping a few balloon links before reaching a ground station. This inter-balloon communication was performed using radio frequency links, though laser-based optical communication was also being attempted over the past years.

After the mission ended, or at the end of the balloon’s lifespan, the balloon would be directed to an easy to reach area where its helium would be released into the atmosphere during which the balloon descended for recovery. On a number of occasions, however, balloons have crashed. Even with the parachute-based emergency system in place, it can still happen that the 10 kg payload plus balloon falls out of the sky.

However, even without those safety concerns, there was something else that was threatening to sink Loon LLC faster than a balloon with a helium leak: their financial balance sheet and increasingly desperate attempts to attract funding. As a commercial entity they would have to put into action a viable business plan that would enable them to somehow make money, if only for maintenance and further development.

Who Needs Floating LTE Towers?

As it turned out, large, fragile floating structures that are constantly subjected to the harsh conditions in the Earth’s stratosphere are quite expensive to maintain. As pointed out by Slate’s Future Tense on Loon’s demise, each balloon needed to be replaced approximately every five months at a cost of tens of thousands of dollars. With the target areas featuring mostly potential clients on the lower end of the income scale in that nation, subscriptions to the service were not going to cover the expenses.

There was also the issue that as a result of being solar-powered some parts of the globe were off-limits, reducing the potential markets. Finally, by being dependent on the right air currents in the stratosphere to navigate, any mistake or incorrect prediction there could lead to a balloon being taken many kilometers away from its intended location, cutting service for people in that area.

Despite having multiple ongoing trial projects by 2019, Loon as an independent entity was still reliant on its originating company Alphabet (formerly Google, now the parent company of Google). Having already burned through the money from an external investor, Loon was by that time trying to sustain a yearly loss of $100 million.

Between the advent of ground-based 4G and 5G towers in more and more areas and encroachment of wired internet, the possible market for Loon was rapidly shrinking. In addition, although space-based internet access had been an option for as long as Project Loon had been around, it had to that point been a remote threat, especially with the high cost and latency from primarily geosynchronous satellites. When SpaceX announced Starlink in 2015 and launched the first batch of Starlink satellites in 2018, Loon’s prospects began to look quite dire.

Lasers in Space

Dozens of new StarLink satellites ready to be deployed in 2019.

Unlike Loon’s balloons, Starlink satellites can be navigated to an exact orbital position in low Earth orbit (LEO), unaffected by weather and the atmosphere. Due to their close proximity to Earth, round-trip latency is minimal and bandwidth is expected to reach tens to hundreds of megabits using its Ku and Ka-band transceivers. SpaceX also has full vertical integration of Starlink satellite development and launch capability using Falcon 9 rockets today and conceivably hundreds of satellites being launched at once by Starship.

It’s not inconceivable that Starlink will be able to provide service to every square meter of the Earth’s surface within the next few years, at increasingly lower cost per satellite. With the lifespan of a single satellite being approximately 3-5 years based on what we know at this point, it’s reasonable to assume that they would be cheaper than Loon’s balloons over this time period.

As of writing, StarLink is being used by Beta testers primarily in the northern USA, providing people who are living in areas that are poorly serviced by existing wired and wireless (LTE) internet options suddenly with access to proper broadband internet connectivity. At the beginning of this year, SpaceX’s Transporter-1 mission also launched the first 10 StarLink satellites into a polar orbit which contain active laser-based inter-sat communication, allowing for multiple hops between satellites before reaching a ground station.

It’s the Business Plan, Silly

Perhaps more important than the costs was the altruistic angle of Loon: by focusing on providing LTE to impoverished areas first of all, it seems to have completely ignored the issue of a business needing income to finance itself. Their approach was to contract services to existing telecoms who would in turn sell to subscribers, but this never turned into a solid and reliable model. Much is yet unknown about how Starlink subscriptions will operate, but SpaceX decided to focus on first providing poorly connected North American citizens with the service, betting that they will be able to generate sufficient subscriber fees in this market.

As for the wind-down of Loon, a few assets from the project will find themselves absorbed into another Google X project: Taara. Taara involves none of the balloons or other flashy parts of Loon, instead taking mostly the newest technology that Loon was working on: the optical data links that would have increased the bandwidth between individual balloons. By using line of sight transceivers, this should theoretically enable remote areas to be connected with the internet and more without having to dig trenches to put in fiber.

While Loon’s story is at an end, it seems that with Taara the spirit of the project will live on. Here a major and so far unanswered question will remain whether Project Taara makes sense when Starlink exists. I’m however quite certain that within a matter of years we’ll have our answer here.

47 thoughts on “Google Loon’s Internet Balloons Come Back To Earth After A Decade In The Stratosphere

  1. Seems a shame to me, this is a neat and effective system, so easy to deploy where needed rapidly.

    Starlink and the like is clearly superior in many ways, but something goes wrong with a few of their satellites its serious downtime as prepping a rocket, plotting all the orbital manoeuvrers for each satellite in the rocket takes time. And as space gets more congested that last bit is only going to get harder, while also increasing the chances that a failure will cascade and wipe out larger numbers of satellite, the failures need not all be electrical/software gremlins and the odds of failures to debris keep going up..

      1. Yeah if/when such a system is fully deployed a few electrical/software failures won’t matter much, but the more you add to similar orbits the more you risk a debris chain reaction taking large numbers of them out, and that will matter, might even make replacing them impossible for quite some time – having to wait for the worst of the debris to de-orbit before you can hope to keep a satellite alive..
        Not to mention they are still vulnerable I would assume to high altitude nuclear blasts EMP for instance, lots of ways things can fail, and things really in orbit are much harder to just replace or even redeploy.

        1. The Starlink satellites aren’t actually intended to be super-long lived – they’re only intended to be up for about 5 years or so. The whole thing is only viable because Starlink launches are functionally free through SpaceX – in a very real sense, Starlink is them trying to monetize excess launch capacity.

          1. You still actually have to have a launch window, space in the right orbits, orbital transfer plans for all the stream of satellites one of those things launches.

            I’m not saying its too expensive to replace, the way rocket development is going that isn’t going to be much of an issue into any Earth orbit soon, and LEO is already pretty damn affordable. But it does take time! You don’t want your customers to be without for weeks/months while you do all the building/planning/waiting for a launch window/transfer orbit phases, but if something serious goes wrong that’s how a satellite system will be. If the orbits are too full of crap after an accident that won’t clear up in a hurry either, you’ll have to wait for all the junk to de-orbit or smash into something else before its worth a new launch (I hope it never happens, but with so much stuff up there…).

            The balloon on the other hand can pretty much just be inflated and released, anything that goes wrong the replacement is up faster than the ‘Service Upgrades’ downtime from my ISP, and so far those have all been pretty quick…

        2. Not to mention they are still vulnerable I would assume to high altitude nuclear blasts EMP for instance, lots of ways things can fail, and things really in orbit are much harder to just replace or even redeploy.

          I… what. You think people are going to have internet access in their top million things to do if a nuke goes off.

          I mean, EVERYTHING is vulnerable to a nuke. For all you know there are SpaceX ships sitting on the launch paid already full of starlink satelite just waiting to replace them if necessary.

    1. Where i live, starlink would have to provide unlimited +150mbps bandwith at less than 25 $ per month to be competitive. Sure, large swaths of the planet doesnt necessarily have the infrastructure, but a not insignificant percent of that area cant afford even 20$ subscription fee monthly. I do not see how the economics can work out in the long run.

      1. Where I live, I’d happily pay $100/mo or more for unlimited 150Mbps, because right now the other options are either ludicrously expensive, capped at anything from 40-160GB, painfully slow, or totally unreliable – Pick any two.

      2. The economics are really simple for Starlink:
        ~$250-300m OPEX / month (launch + satellite costs) over 5 years for 12,000 satellites
        More than 10m households without broadband in the US alone represent potentially $1b / month in subscription fees.
        Add FCC funding, military/government contracts, Canada, ships, airlines, and wealthier residents of infrastructure poor countries?

        1. Excellent reply you’re definitely talking 10 billion a year just from internet now we had a hundred billion a year for TV cable service turned into an internet satellite TV service. Comcast etc are in the bathroom right now throwing up

    2. 90-95% of global communication is sent over transatlantic cables -FACT- and these high pressurized balloons are low orbital relay telemetry transmitters and they get deployed mainly from the south antarctic pole, hence; that is why the populus has never witnessed these deployments. There is no such thing as a orbital satellites launched from rockets.Sorry to burst anyone’s bubble, wakey wakey!

  2. So to be clear, the NASA SPBs (pictured aloft) and the Loon SPBs (pictured at ground) are, uh, just a *tad* bit different in scale. As in, the Loon SPBs can lift a 10 kg box… and the NASA SPBs could lift a car.

      1. I mean, not really? Helium’s nearly as bad, although the additional reactivity of hydrogen’s a problem in other cases. But I know of at least 1 company (World View) that uses hydrogen for ground testing for similar thin film balloons – with famously flamey results in one case.

        Really the reason is that these companies use the same production material to make *super big* balloons too, and no one’s filling a 28M cubic foot balloon with freaking hydrogen. That’s 4x the volume of the Hindenburg!

        1. But an important distinction is that these balloons do not have living beings on board!

          And if they did catch fire the hydrogen would burn up long before it hit the ground.

          Helium is used because it is safer for small ground crews, but for a long term service an unmanned launch tower that fills the balloon and sends it up could make hydrogen use safe, easy and cost effective.

          Note the Hindenburg was a sad note in history but with 35 fatalities and 62 survivors, was no worse than any other commercial aircraft coming down. Look at the actual damage it caused

          It really is a poor argument against using a particular technology, considering the giant aluminum clad diesel bubbles we fling through the air at hundreds of miles an hour on a daily basis. Somehow we made that safer than driving to the local store.

          1. I think you didn’t understand my point: again, they use the same material for balloons that are ~10 times larger: as in, football-stadium scale when inflated.

            You can’t launch those with a tower! And yeah, they don’t have humans on board – they’ve got multiyear multimillion dollar payloads that weigh the same as a car. Bad Things happen if those payloads fall uncontrollably to the ground (which I know from experience).

            Even launching them *at all* is sometimes dangerous. You can find YouTube video of the NCT crash in Alice Springs, Australia where they misjudged launch conditions and the payload ended up trashing an SUV as it ran across the ground. Now imagine that, except the balloon *above* it is a massive bomb.

            “considering the giant aluminum clad diesel bubbles”

            Again, these are around 30M cubic feet. Thirty. Million. Those “giant aluminum clad diesel bubbles” are *pebbles* compared to these things.

            Go look up the World View balloon explosion in December 2017. That was a smaller-scale balloon that they were testing that detonated, causing $500K in building damages and a shockwave that caused effects miles away.

            I know it’s a common meme/trope/idea that “hey, the Hindenburg wasn’t that bad, hydrogen is still safe!” but no, in fact, the people who do ballooning for a living use helium for a reason. If there were other good options, they’d use it. Helium’s effing expensive. It’s a good portion of the cost of the launch for the big guys.

          2. I understand your objections.

            I know a giant balloon full of flammable gas can be dangerous. However the examples you provided are really not nearly as scary as you are making them out to be.

            Calling an airliner a “pebble” is disingenuous. And the fact that the part of the balloons that are large is lighter than air changes the equation quite a bit.

            A car falling from the sky is a “pebble” compared to a 747.

            I agree that launching a flammable balloon in the middle of a populated town is a very bad idea.

            And maybe my off hand example of a tower was not the best.

            As an alternative how about a helium filed drone ship that would ascend then inflate and deploy the payload ship at high altitude. That way neither humans nor infrastructure would be endangered.

            I’m not just spouting off drivel from a media outlet. I’m actually trying to think about the cost benefit analysis of a technology that I think has untapped potential.

          3. I’m not spouting off stuff from a media outlet, either. I’m speaking as someone who knows exactly how much NASA’s balloon program has to go through when there’s any incident, and how often balloons like that have issues

            Large scale balloons *do not* have the launch success record that planes have. It’s probably like 1 in 100 (probably higher, unless I’m an outlier: I know personally of at least 2 launches with leaks, and I don’t personally know of 200 launches) that end up with a leak: they’re tiny thin plastic that has to get rolled out. They get torn.

            You keep comparing this to planes, which is nuts: planes are *powered*. It’d be like you saying “why is skiing without poles more dangerous than driving? Cars are bigger!” No power, no control.

            I’m not saying you couldn’t do it: Yahoo at one point I guess was looking into it.

            What I’m saying is that if Google goes to Aerostar and says “balloon?” they’ll set things up and order helium. Ballooning’s enough of a crapshoot you don’t add flammable gasses to it.

    1. Interesting point. I wondered whether hydrogen was too ‘leaky’, but apparently it leaks less than helium. Can’t see a reason not to use it. Methane is quite a lot less buoyant, so would need a larger balloon, but the material can be cheaper… and the methane itself is cheaper. It might come down to the public idea of burning wreckage coming from the sky.

      1. Yup, helium diffuses fastest because it’s monatomic, and smaller than a hydrogen molecule.

        I don’t know if any other companies make smaller SPBs, but the Loon (and NASA) ones are both made by Aerostar, so they obviously would want to have the same design/setup for both the smaller and larger balloons. I do know that other balloon launch companies (World View) have used hydrogen for ground testing, but I’m not sure if that’s the same material. Even if it was, though, hydrogen’s dangerous enough that I know World View doesn’t use it for actual launches.

        SPBs are tricky to deploy, though – it’s apparently a fairly complicated math problem to ensure they don’t tangle (and one that NASA’s been fighting for decades). So honestly I’d imagine that Aerostar (and Google) had no intentions whatsoever of adding any risk to the launches.

      1. I believe WWI saw semaphore and winky blinky morse sent between balloons, relayed to ground with a field telephone. I think the French were using them in that fashion for a couple of decades before that.

        1. The US Army experimented with hot air balloon telegraph stations during our Civil War.
          One could get a good view of a battlefield and send messages to ground units for direction.
          (Disney made a movie about that)

  3. I think the project’s fate was doomed when SpaceX announced StarLink. I’m not too sure about the viability of StarLink in the long-term either, but in terms of providing connectivity in remote areas even in times of emergencies and crisis, I can imagine it’s a hard sell deploying balloons over installing a few (portable) antennas.

    I’m curious about this, though, “…being solar-powered some parts of the globe were off-limits…” Which parts of the globe and why is it off-limits? IOW, what parts of something being solar-powered make it off-limits?

    1. In order to operate entirely on solar power, you need to be able to collect your 24-hour energy “budget” during daylight hours. In this particular case, there’s always going to be a latitude above (or below…) which you’re inevitably going to have issues doing that, especially during the winter, and winter-adjacent times of the year.

      I’m not sure how hungry the payloads on these things get, but given that most data communication uses some degree of Tx/Rx, regardless of where the bulk of the data is coming from, and transmission to ground requires jamming a metric fuckton of angry pixies through an antenna, I can’t imagine that they’re particularly low-power method.

    2. I assume the high/low latitudes during their winter.

      At the poles, during the dead of winter, there is very little sunlight, and I assume that being high up wouldn’t help enough.

    3. Sorry, I should’ve been clearer. I read that statement (i.e., the off-limits bit) as the balloons were forbidden from being deployed (by local authorities) simply because they’re solar-powered. That’s why I asked what the reasons are (e.g., political, environmental/pollution, scientific, religious(?), military, etc.). Yes, I’m aware of the natural reasons, I just thought there were some man-made ones. (c:

  4. I’d think the biggest problem with long term balloon based communications is the things wouldn’t stay over the service area for long enough. Even if they went “Sherwin Williams” (Covers the Earth) with them, they’d still end up clustering into swirls and streams depending on air currents. If the wind blows away from where you need service, you can’t have service.

  5. “Are we rising again?”
    “No. On the contrary.”
    “Are we descending?”
    “Worse than that, captain! we are falling!”
    said Jules Verne
    …and also the Loon CFO & Marketing Strategy Team.

Leave a Reply to Foldi-OneCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.