Startup liftoff

See allHide authors and affiliations

Science  10 Apr 2015:
Vol. 348, Issue 6231, pp. 172-177
DOI: 10.1126/science.348.6231.172

Planet Labs co-founders Chris Boshuizen (left), Robbie Schingler (middle), and Will Marshall horse around with a Dove at their office in San Francisco, California.


Vats of homebrewed porter and brown ale ferment under a lunchroom table. In the corner lie a drum kit and guitar, property of Hank and the Doves, the company's pop cover band. Emma the dog roams—and sheds—freely. In some ways, Planet Labs is your typical Silicon Valley startup. But it's not where you'd expect to see the precision assembly of space satellites. “In terms of overall cleanliness, we just don't care,” says co-founder Chris Boshuizen, who wears a droopy Santa hat in preparation for an office holiday party on this rainy December day in San Francisco.

Boshuizen pushes aside strips of clear vinyl sheeting and enters what he calls the “clean enough” room. He stomps on a mat of sticky tape that helps eliminate static charges that could zap satellite electronics—a rare precaution. Beyond another line of tape, no alcohol is allowed. There a shelf is stocked with the company's product: space telescopes no bigger than a loaf of bread.

Two dozen of these telescopes, called Doves, already orbit the Earth, imaging the ground with a resolution good enough to pick out treetops, roads, and buildings. Another 14 are set to ride into orbit next week on a SpaceX cargo rocket. Although heftier spacecraft can spy on Earth with higher resolution, few can match the repeat rate at which one craft in Planet Labs' swarm passes over the same patch of ground. If the company can get between 150 and 200 Doves in orbit, it will be able to take a daily snapshot of the entire planet. This time-lapse flipbook will reveal flooding on rivers, logging in forests, and road building in cities, as they happen.

Change is the name of the game—and the main attraction for researchers and commercial clients. Boshuizen says the company has contracts in place that are worth more than the $135 million in venture capital funding it has so far received.

From the shelf, Boshuizen grabs a Dove. Though it is destined for orbit, he handles it as roughly as he would the phone in his pocket. He doesn't worry about damage, because the inexpensive electronics inside it are similar to the phone's. The nonchalance is all part of the plan. Plenty of Doves have fizzled out in orbit or fallen back to Earth and burned up in the atmosphere. But that's OK, because more Doves are always ready to take wing.

For decades, engineers have been building satellites like bespoke Swiss watches, sparing no expense and spending years to perfect them. More than 1000 people in 17 countries have had a hand in building the $8 billion James Webb Space Telescope, more than a decade in the making. Once launched in 2018, the Webb telescope will never be made again. At Planet Labs, Doves are mass produced to the brink of disposability; in 2015, the company is aiming for a sustained production rate of 25 Doves a month. Boshuizen says the cost of building and launching each Dove is well under $1 million.

Mass production lets engineers experiment with the satellites and launch them in aggressive, iterative cycles. In less than 2 years, Planet Labs has gone from Dove 1.0 to sending version 12 into space. Yet the ratio of total Doves launched (99) to employees (105) is still about one. “It changes because we keep hiring,” Boshuizen says. “But we also keep building satellites.”

Much of Planet Labs' success can be attributed to a decision to squeeze their ambitions into a very small box, a specific form factor called a CubeSat. Named after their characteristic 10-centimeters-a-side size, CubeSats were first launched in 2003 as an educational tool—a way for graduate students to get something in space (see sidebar, p. 176). The little boxes beeped and did little else. But fueled by the massive investment in consumer electronics, the size and cost of most satellite components—radios, computers, solar panels—have plummeted even as their capabilities have exploded. At the same time, CubeSat builders have found new and cheaper ways to get into orbit: packed into spare payload space around larger satellites, or stowed on cargo flights to the International Space Station. In 2014, a record 132 CubeSats were launched—and 93 of them were Doves, according to Jonathan McDowell, an astronomer at the Harvard-Smithsonian Center for Astrophysics in Cambridge, Massachusetts, who tracks satellite launches.


McDowell says CubeSats are gripping the aerospace industry and changing the way business—and science—is done. “Now you're seeing not just student projects, but CubeSats deployed by the military, by space agencies—doing real jobs,” he says. Jordi Puig-Suari, the co-founder of the CubeSat standard, concurs. “These little guys are finally ready to do serious missions,” he says. And one upstart company is leading the charge, he adds: “Planet Labs is the darling of the CubeSat community.”

IN THE LATE 2000s, small satellites were still a curiosity, but a buzz was in the air at NASA's Ames Research Center near Mountain View, California. Boshuizen and another physicist, Will Marshall, recall a senior Ames engineer waving around a government-issued smart phone, declaring that it had more computing power than the average satellite. Why not just launch the smart phone? he asked. “We eventually took him seriously,” Boshuizen says. “We got a smart phone, stuck it in a vacuum chamber, and it still worked.” It turned out that costly “rad-hard” parts, built to withstand the vacuum of space and its radiation environment, weren't so important.

“Perhaps you've been lied to with this whole notion that things need to be space-qualified,” Boshuizen says. After all, he notes, astronauts on the barely shielded space station use iPads. (The more intense radiation of deep space, however, might pose a threat to future CubeSat missions to other planets, which would venture outside Earth's protective magnetic field.)

In 2010, Boshuizen and Marshall assembled a PhoneSat team at Ames. They took an HTC Nexus One smart phone out of its case, reprogrammed its Android operating system, and added extra batteries and a radio that would downlink pictures to Earth. Total cost: about $3500. In 2013, the first three PhoneSats were launched. Two of them lacked solar panels, but they took and sent back pictures in the week before their batteries ran down. “We bet the farm on this idea that we could launch a phone into space and that it would work,” Boshuizen says. “And it did.”

Even before the PhoneSats were launched, Boshuizen and Marshall began talking about taking the idea further with a third friend, Robbie Schingler, who was working for the chief technologist at NASA headquarters in Washington, D.C. The fast-talking trio knew each other from long before: They had met at a 2002 Space Generation Summit in Houston, a workshop for young space evangelists. “The day I showed up [there], I decided they were my people,” Boshuizen says.

In December 2010, they registered their own company, initially called Cosmogia. They talked about many possible business models, including a constellation of Cube-Sats that would provide global Internet. They ultimately decided that an Earth-imaging mission carried fewer risks and was a service for which there is growing business demand. “Frankly, we chose to do the remote-sensing mission because we had a higher probability of success,” Schingler says.

Even though the team had no funding and no way of paying anyone, Ames engineers like Ben Howard were eager to join. “They had a big idea for what to do with these CubeSats,” he says. “I also didn't think they were crazy.”

A fire in a Brazilian field is captured by a Dove (bottom) a day after the same spot was imaged by Landsat (top).


For most of 2011, they worked out of the “Rainbow Mansion,” a six-bedroom house in Cupertino, California, that Marshall and Schingler had rented in 2006 and packed with like-minded people. With house policies achieved by consensus and weekly salons given by the likes of physicist Roger Penrose and Internet activist Lawrence Lessig, the house is like a “hippie commune except it's mostly filled with tech geeks,” Howard says.

Like so many Silicon Valley startups, the group began working out of the garage. They pushed camping equipment to the side and hung a stop sign from the ceiling to keep random Rainbow residents out. The team worked on couches, hunched over laptops loaded with computer-aided design software. One person might design a part and get it 3D-printed, while another would work on orbital simulations.

In designing the Doves, the group chose a frame the size of three stacked classic CubeSats: 10 centimeters square and 34 centimeters long. They found that Questar telescopes—used by hobbyists for more than half a century—fit the space perfectly, so they ordered a couple of custom ones fashioned out of Invar, a thermally stable nickel-iron alloy. They pointed Questars out the garage door to test settings on megapixel cameras attached to the rear of the scopes. Occasionally, they would cross the bay to Lick Observatory, set up a receiver antenna, and test the strength of their radio. “We were not naive,” Howard says. “We had come from NASA and had seen the so-called right way to design a satellite. We were very aware that we were doing something different and risky.”

Soon, they had assembled mostly off-the-shelf parts into a working satellite that they were willing to send to space. With their personal savings, the three co-founders booked a $250,000 spot on the maiden flight of Orbital Sciences' Antares rocket, in April 2013.

Michael Safyan, another early employee, says he didn't sleep much during the 6 days that Dove-1 lived. The team was ecstatic when a 4.5-meter radio dish in Chilbolton, U.K., captured the first image: a patch of forest in the Pacific Northwest, in such good focus that they could count the trees. “We had no idea if the thing would even turn on,” Safyan says. “It was such a huge validation that we were on the right track.”

By that time, the fledgling company had moved out of the garage and into its San Francisco office. More Doves entered orbit, and the images they took began to stitch together the patchwork quilt of Earth. The Planet Labs team started to notice changes. The boundaries of pit mines expanded. In Rio de Janeiro, they saw that favelas had been bulldozed in the weeks leading up to the 2014 World Cup. They even noticed plumes of smoke over a patch of ground in California—an incipient forest fire. “It was just 10 minutes old,” Howard says.

Investors began to line up, and customers began to sign contracts. Planet Labs has publicly announced only a few, but Boshuizen talks about several areas of commercial interest. A primary one is agricultural: the ability to monitor the productivity of fields. Environmental compliance is another—for instance, mining companies wishing to show that they have restored an area to the correct standard. A third area is in commercial mapping: By monitoring the growth of roads and homes, Planet Labs can identify areas where Internet mapping companies need to concentrate their data-gathering efforts.

Two Doves are expelled from the International Space Station on 27 February 2015.


Plenty of scientists would also like to get their hands on Planet Labs data, says Curtis Woodcock, a geographer at Boston University and co-leader of the science team for Landsat, the venerable series of Earth-monitoring satellites operated by the U.S. Geological Survey (USGS). Woodcock says officials routinely ask Landsat scientists what the satellites could do better, and the answer is always the same: Make more frequent passes over each patch of Earth. Daily snapshots could help home in on the moments when forests green up in the spring. “It's becoming more volatile and has shifted earlier in many places as a result of changing climate,” Woodcock says. Another scientific application, he says, is nailing the timing of snowmelt, which is hard to predict and has big implications for water management.

But with an entire earth to survey and just two working satellites, Landsat 7 and 8, an 8-day repeat rate is the best Landsat can manage, Woodcock says. To be sure, the big orbiters have 11 spectral bands to the Doves' three basic color bands and a heritage that ensures that images are precisely calibrated from one mission to the next. “The scientific community functions in the measurement domain,” Woodcock says. “Whether Planet Labs is going to make it from the picture world to the measurement world is still up in the air.”

But Doves, with 3- to 5-meter resolution, already outperform the 15-meter resolution of Landsat 8—although other Earth-imaging services can do even better. DigitalGlobe, for example, provides satellite imagery that has a resolution of better than a meter, but tasking the company's fleet of six truck-sized satellites to get a new image for a specific area can take a week or two and is expensive. Another company, SkyBox, plans to launch a constellation of two dozen satellites the size of mini-refrigerators—still small by typical aerospace standards, but much bigger than the Doves—that are also capable of reaching submeter resolution. (Google paid $500 million to buy Skybox last June.)

Still other imaging companies are developing drones for jobs such as high-resolution monitoring of oil and gas pipelines. But Boshuizen says the Doves can complement those efforts—for example, by flagging changes that drones or more capable satellites can then examine in more detail. “I view them as our customers, not competitors,” he says.

Although Planet Labs is trying to make money, the three co-founders exude humanitarian idealism. For example, they say they hope someday to make Planet Labs data free—perhaps, Boshuizen says, by charging only for the newest images or for access to large numbers of images, as Google Maps does. They say their tiny telescopes could allow watchdog groups to monitor environmental degradation, or let human rights groups keep tabs on the size of refugee camps and the movements of marauding militias. Schingler says the Doves will empower people who have never before had access to daily geospatial data. “What happens when anyone on the planet can understand the state of the world?” he asks. Power dynamics will shift markedly, Marshall says: “Not everyone's going to like it. We're very cognizant of that.”

Planet Labs' way of doing business could be equally transformative, McDowell says. The whole notion of doing Earth observation via constellations of small satellites poses a threat to the old order, in which large aerospace companies build expensive, large satellites for agencies like NASA. For example, Landsat 8, launched in 2013, was built by Orbital Sciences at a total cost of $855 million.

USGS is planning a replacement Landsat mission, and budget realities mean it will have to be cheaper than its predecessors. Boshuizen says Planet Labs does not intend to formally compete to replace Landsat but would be happy to sell its data to USGS. The company has also discussed other missions: CubeSats stuffed with sensors rather than telescopes. “You could do radar, lidar; you could do GPS augmentation or replacement,” Boshuizen says. “We could move on to do things like earth science, atmospheric science, science in low-Earth orbit; we could do astrophysics and heliophysics, in part.”

Pete Worden, director of NASA Ames, is proud that the people and ideas behind Planet Labs were incubated there—even if their visionary approach comes back to haunt NASA centers. “Applying Silicon Valley to aerospace is the most revolutionary thing that's happened probably since Goddard built his rocket,” he says. “It's not surprising that it started here. But it's spreading.”

After this Antares cargo rocket exploded on 28 October 2014, Planet Labs scrambled to get new Doves into orbit.


THE FINAL MONTHS OF 2014 brought major changes to Planet Labs, and one big challenge. It started on 28 October, when 26 Doves—known as Flock 1d—were scheduled to ride into space aboard an Orbital Sciences Antares cargo rocket bound for the space station. The launch, from the Wallops Flight Facility in Virginia, was the company's seventh.

At the Planet Labs office in San Francisco, Schingler made pancakes in the lunchroom—a launch-day tradition at the company—and served them to 70 or so employees and guests who had gathered to watch a NASA TV webcast. Also in keeping with tradition, Marshall delivered a prelaunch briefing. The Doves, he explained, were locked inside special deployers, wrapped in bubble wrap, and strapped down in the capsule, alongside food, water, and experiments destined for the space station. He reminded his listeners that launches were uncertain: Delays could occur at the last second, failure was always possible, and everything was out of their hands.

At 3:22 p.m. Pacific Time, the rocket lumbered off the launch pad. Fifteen seconds later, one of the engines exploded; then a larger fireball engulfed the entire rocket. “There were gasps of shock,” Marshall recalls. In a single moment, the company had lost a huge chunk of its assets.

For an old-school space mission, it would have been a shattering blow. In 2009, for example, another Orbital Sciences rocket fell into the ocean while attempting to deliver into space the Orbiting Carbon Observatory (OCO), an important climate-monitoring satellite. It took 5 years to return a copy of OCO to orbit.

Planet Labs, however, still had Doves in space and a production line to make more. Boshuizen took stock: There were enough parts on hand to make 10 Doves immediately. Schingler began calling officials at NASA and at NanoRacks, the company that built the deployers, to see how quickly he could get on a subsequent launch. He worked to transfer regulatory licenses to the replacement satellites. He secured room for two Doves on the next SpaceX resupply mission, scheduled for December. That meant building and boxing up the satellites in record time. The team finished in 9 days. “The team did a major sprint,” Schingler says. “It was a herculean effort.”

Two Doves, covered in personalized graffiti (“NBD, all sats burn up someday”), were locked in padded Pelican cases and shipped to NanoRacks' facility in Webster, Texas. Boshuizen sneaked special features onto each of the spacecraft: One would gather three times as many pixels over a larger field of view, and the other had an experimental infrared filter on top of the three needed for color pictures.

The SpaceX launch was delayed twice. When it finally took place—on 10 January 2015 at Cape Canaveral in Florida—Planet Labs was already in the middle of its next big project: boxing up to move. The new headquarters, a few blocks away, boasted exposed brick walls and wood beams, breathing room for the growing staff, and space for brand-name diagnostic equipment such as a ThermoStream, a machine for pumping hot air into chambers to test the resilience of circuits and other equipment. The dog-friendly policy was being rescinded (primarily because of a new employee's allergy, Boshuizen says).

The launch itself signaled another transformation for the company: In contrast to October's drama of loss and recovery, it was routine. Schingler watched from a laptop at another of his communal compounds, in the jungle of Costa Rica. Marshall was in Florida, watching the predawn launch of the SpaceX Falcon 9 rocket in person for the first time. In San Francisco, Boshuizen went out for dinner and returned to the soon-to-be-old office around midnight. With only a handful of employees hanging around so late, there was no need for a prelaunch briefing. Or pancakes. Boshuizen, lead guitarist for Hank and the Doves, picked up his guitar and started to noodle around. “What else do you do on a Friday night?” he asks. “We were just sitting around chatting and drinking.” A few more employees trickled in on their way home from San Francisco's bars and clubs.

Finally, the countdown came at 1:47 a.m. Pacific Time. Boshuizen, remembering how hard his team had worked to get the two Doves on board, watched nervously as the Falcon 9 roared off the launch pad, a streak of orange in the humid Florida night. He kept his eyes fixed on the video screen for another 10 minutes, until both the first and second stages had cut out—until the big bird had carried his two little birds to space. “Once it got out of the atmosphere, I could relax,” he says. “I knew it would be fine.”

  • * in San Francisco, California

View Abstract

Stay Connected to Science

Navigate This Article