SAN FRANCISCO — In October, the Chinese province of Guangdong — the manufacturing center on the southern coast that drives 12 percent of the country’s economy — stopped publishing a monthly report on the health of its local factories.
For five consecutive months, this key economic index had shown a drop in factory production as the United States applied billions of dollars in tariffs on Chinese exports. Then, amid an increasingly bitter trade war between the United States and China, the government authorities in Beijing shut the index down.
A small start-up in San Francisco began rebuilding the index, lifting information from photos and infrared images of Guangdong’s factories captured by satellites orbiting overhead. The company, SpaceKnow, is now selling this information to hedge funds, banks and other market traders looking for an edge.
High-altitude surveillance was once the domain of global superpowers. Now, a growing number of start-ups are turning it into a business, aiming to sell insights gleaned from cameras and other sensors installed on small and inexpensive “cube satellites.”
The companies and governments that spent decades using internet services, cameras and other devices to collect data on regular people may soon get a taste of their own information technology.
“Businesses will not be able to hide from competitors or regulators or watchdogs,” said Mark Johnson, chief executive and co-founder of Descartes Labs, another satellite information start-up. “They need to realize that their traditional competitive advantage — information — will be available to everyone.”
Nearly 730 Earth observation satellites were launched over the last decade, according to Euroconsult, a research firm that tracks the space market. In the next 10 years, 2,220 more will follow them into orbit, training an increasingly wide array of sensors on the planet.
Orbital Insight, in Palo Alto, Calif., is one of the first companies to build a business around cube satellite data. Sitting in Orbital’s offices on a recent afternoon, James Crawford, the company’s founder and chief executive, who goes by Jimi, opened his laptop and pulled up a report on three big-name retailers: J. C. Penney, Macy’s and Sears.
Based on the company’s satellite data, a color-coded line graph showed a steady drop in the number of cars parked outside the thousands of stores operated by the three retailers. The drop was particularly steep for Sears, which had filed for bankruptcy just days earlier. “This is one of the reasons they’ve been under so much pressure,” Mr. Crawford said.
You don’t need satellite photos to know that Sears is failing. Companies like Orbital Insight are typically tight-lipped when it comes to more important data — as are their customers — mainly because they see this information as a competitive advantage.
But the line graph showed how Mr. Crawford and his start-up can target the performance of individual businesses. Orbital Insight tracks activity in more than 260,000 retail parking lots across the country, and it monitors the levels of more than 25,000 oil tanks around the world.
Not surprisingly, Orbital Insight and SpaceKnow said, some of their customers use this satellite data to track the progress of their direct competitors, though those customers and their competitors are very reluctant to talk about it.
Mr. Crawford believes the satellite analysis will ultimately lead to more efficient markets and a better understanding of the global economy. Fred Abrahams, a researcher with the advocacy group Human Rights Watch, sees it as a check on the world’s companies and governments.
Mr. Abrahams and his team use satellite imagery to track everything from illegal mining and logging operations to large-scale home demolitions. “This is why we are so committed to these technologies,” he said. “They make it that much harder to hide large-scale abuses.”
All of this is being driven by a drop in the cost of building, launching and operating satellites. Today, a $3 million satellite that weighs less than 10 pounds can capture significantly sharper images than a $300 million, 900-pound satellite built in the late 1990s. That allows companies to put up dozens of devices, each of which can focus on a particular area of the globe or on a particular kind of data collection. As a result, more companies are sending more satellites into orbit, and these satellites are generating more data.
And recent advances in artificial intelligence allow machines to analyze this data with greater speed and accuracy. “The future is automation, with humans only looking at the very interesting stuff,” Mr. Crawford said.
Orbital Insight does not operate its own satellites. Nor does SpaceKnow or Descartes Labs. The start-ups buy their data from a growing number of satellite operators, and they build the automated systems that analyze the data, pinpointing objects like cars, buildings, mines and oil tankers in high-resolution photos and other images.
Now, satellite operators are building similar systems, selling analysis as well as the raw data. The market topped $4.6 billion in 2017. By 2027, it will reach $11.4 billion, according to Euroconsult.
What began with satellite cameras is rapidly expanding to infrared sensors that detect heat; “hyperspectral” sensors that identify minerals, vegetation and other materials; and radar scanners that can build three-dimensional images of the landscape below. As it reconstructs the Guangdong economic index, SpaceKnow uses infrared imagery, which can help show activity around roughly 600 factories and other industrial sites in the province.
After a new satellite went up in December, a Virginia start-up called HawkEye 360 will soon track wireless signals — a way of understanding the behavior of everything from cellphone networks to cargo ships. This could provide new insight in the progress of cellular companies like AT&T and Verizon, including how many cellular towers are in operation, how active they are, and what technologies and wireless bands are being used.
But as technology improves and costs drop, some still warn satellite intelligence gathering has its limits. Finding useful information in satellite imagery can be expensive, said Shawana Johnson, a veteran of satellite intelligence work who is now the president of Global Marketing Insights, a consulting firm dedicated to this area.
“You have to want to look at a variety of activities across the Earth and look at them daily — or weekly — for the cost to makes sense,” she said.
Others question how useful this data is, particularly for market traders. John Alberg, a data and machine learning specialist who co-founded the investment firm Euclidean Technologies, said traders could find more important and detailed information about companies here on Earth.
“Satellite imagery is a narrow niche,” he said.
Mr. Johnson, of Descartes Labs, acknowledged that satellite data is not always as useful as it seems to be. Descartes and other companies are still searching for the best ways to use this new trove of information.
The data collected by SpaceKnow provides a rough guide to factory production in Guangdong, said Jeremy Fand, the SpaceKnow chief executive. “We can see when activity is growing — and slowing,” he said. But it is by no means as detailed or as reliable as the information Guangdong officials were collecting for their original index.
In January, one satellite also provided a reminder that this is far more complicated technology than smartphones and street cameras. The WorldView-4 satellite operated by Digital Globe suddenly went dark.
“Space is cool, but it takes a long time to get things right,” Mr. Johnson said. “You can’t just send a technician into space.”
Orignially published in NYT.