EXPENSIVE SODA STRAWS
The desire for real-time intelligence is as old as human conflict, and there has always been a tension between collecting as much information as possible and getting that information to those who need it as quickly as possible.
Civil War cavalry scouts could gather tactical intelligence on the enemy but could only report on what they learned after a lengthy trip back to friendly territory. Today, there’s a different problem: a modern surveillance drone can relay imagery nearly instantaneously to users but miss key developments in enemy territory if it is not positioned directly above the target. Space-based intelligence assets have grappled with similar constraints of time and scale.
Project CORONA, the CIA’s earliest spy satellite system, could photograph Soviet territory and return film canisters to Earth, but the time from collection to exploitation was measured in days and even weeks. In the 1970s, the advent of school-bus-size satellites that could beam digital images back to users solved the time constraint but were deployed in relatively limited numbers. The handful of satellites the United States could afford to launch into space still left large gaps in coverage.
Today, according to public sources such as space–tracking websites, the Pentagon and the intelligence community may operate roughly a half-dozen satellites that detect visible light/infrared wavelengths and a half-dozen radar satellites that can see at night and through clouds. These overhead systems are “exquisite,” meaning they cost at least hundreds of millions of dollars each.
Although capable of collecting and promptly transmitting high-quality intelligence, there are not enough of them to solve the scale problem. Because not all points of interest can be watched at once, the United States effectively looks down at Earth through a set of high-fidelity soda straws.
CHEAPER EYES IN THE SKY
A convergence of advanced technologies—reusable rockets, ever-smaller semiconductors, and high-powered AI—is quickly making it possible for the United States obtain that constant-stare capability. The cost of rocket launches has fallen dramatically from the days of the NASA space shuttle program, dropping from roughly $27,000 per pound to $1,200 per pound.
Thanks to the miniaturization of semiconductors, the size of what can be launched has also shrunk. Satellites with advanced intelligence-gathering capabilities can be as small as a breadbox or a backpack. Meanwhile, AI has enabled the teaming of humans and machines, with computer algorithms rapidly sifting through data and identifying relevant pieces of information for analysts.
Private satellite-launching companies such as SpaceX and Rocket Lab have leveraged these technologies to build what are termed “megaconstellations,” in which hundreds of satellites work together to provide intelligence to the public, businesses, and nongovernmental organizations. These companies are updating open-source, planet-scale databases multiple times per day.
Some of these companies can deliver fresh intelligence from nearly any point on the globe within 30 minutes of a request. For example, the company Planet orbits about 200 breadbox-size satellites that image the entire Earth once daily. Each satellite can take 10,000 pictures per day, photographing an area equivalent to the size of Mexico.
When a Chinese spy balloon supposedly flew over the United States in early 2023, the artificial intelligence company Synthetaic rapidly searched Planet’s database for images of the balloon’s trip from China. Synthetaic’s machine-learning algorithms were able to identify the balloon’s path, tracing it to a potential launching site on Hainan Island in southern China.
Humans could never have accomplished this task in this amount of time. As the private sector continues to advance, the U.S. government is trying to keep up through its ambitious plans to rapidly expand the number of satellites in space over the next decade.
The Department of Defense envisions a world where it has no gaps in coverage, no matter what the weather is, 24 hours a day. The quadrupling of the National Reconnaissance Office’s satellite fleet would deliver ten times as many images as today. Furthermore, both the Pentagon and National Reconnaissance Office plan to use commercial collection to augment their capabilities. In the aggregate, this architecture will allow the government to watch more things in more places, never turning away.
With the government’s current handful of exquisite satellites, it can take days for a given satellite to pass over the same point again. A megaconstellation of hundreds or thousands of satellites would be able to constantly relay information about a target without ever having to take its eyes off of it. With this capability, the United States would need to make fewer choices about which targets to watch, no longer needing to prioritize certain collections over others.
It could watch objects and sites without losing track, allowing algorithms, intelligence analysts, and officials tasked with selecting targets to identify patterns of everyday life and thus what constitutes normal and abnormal activity.
An unblinking eye could watch everything from the pens in which China’s nuclear ballistic missile submarines are kept to Iranian Revolutionary Guard facilities that can threaten U.S. troops in Iraq and Syria. Today, these kinds of patterns cannot be watched constantly, and critical changes can be missed if a satellite is not overhead. When combined with the next generation of hypersonic weapons the United States is testing and fielding, these new satellites will allow the U.S. military to strike long-range targets along short timelines, and with less risk to its service members.
Commanders would be able to, say, watch the entire western Pacific from above, tracking all ships in the battle space. All told, these satellites would enable unprecedented coordination across the military services as well as with allies in complex combat operations by informing advanced data-driven command and control systems.
Both China and Russia leverage today’s unstable and manipulable information environment to threaten the territory of weaker regional opponents. Constant-stare systems may provide sufficient coverage to anticipate their acts of aggression. Indeed, publicly telegraphing these technological capabilities could serve as a deterrent by making it harder to launch surprise attacks.
NOWHERE TO HIDE
But with new opportunities comes a new danger: overreliance on constant-stare technology could create bad habits. Hamas’s October 7 attack on Israel has demonstrated the pitfalls of depending on advanced technology. For years, Israel used a high-tech surveillance system to watch its border with the Gaza Strip, but the technology gave Israeli leaders false confidence and contributed to a decision to redeploy Israeli soldiers who were monitoring Gaza to the West Bank.
Overreliance on intelligence collected in space could lead the U.S. government to make similar errors. Plus, the AI-enabled systems that will churn through all the new data can make mistakes. What if an object perceived to be a mobile missile launcher is actually a realistic-looking decoy or, worse, a civilian bus? Conversely, what if an AI system somehow fails to flag the deployment of a mobile nuclear missile launcher for human analysis?
Having access to this kind of information could also shape the way the U.S. government functions and makes important national security decisions. Access to a nearly bottomless supply of intelligence could lead to micromanagement of day-to-day national security affairs as well as acute crises. The NSC could become plagued by do-somethingism, the compulsion to take action in response to perceived problems or threats for fear of being seen as sluggish and ineffectual.
The Department of Defense will also need to come to terms with the risk that China and Russia will acquire these technologies. Sensitive U.S. military and intelligence operations could be exposed to public scrutiny, disruption, and failure. Warfighting concepts that emphasize the dispersion of forces, which are gaining currency across the U.S. military, may be rendered less effective if it becomes harder to hide the movement of U.S. troops.
The Marine Corps’ Expeditionary Advanced Base Operations concept, for example, calls for spreading Marine units across remote islands to attack Chinese ships with long-range missiles in the event of war. If China could employ constant-stare surveillance of the Western Pacific and its islands, it could track the movement of U.S. forces and the assets that resupply them.
China is already poised to roll out such capabilities. The company Changguang Satellite Technology is orbiting a megaconstellation of its own that, by 2030, should be able to image any point on Earth every 10 minutes.
Given the blurred lines between the public and private sectors in China, it is reasonable to expect that the Chinese military will have reliable access to Changuang’s collection. Perhaps most dangerous, constant-stare systems could pose a risk of nuclear escalation.
China and Russia already chafe at U.S. advantages in intelligence collection, regularly harassing American surveillance flights and even approaching and aiming lasers at U.S. intelligence satellites. Constant-stare systems could increase their perception of intrusion as their geographic and temporal sanctuaries are rendered less secure.
These perceptions will be especially sensitive in regard to mobile nuclear missile launchers. Mobile missiles are a particularly effective second-strike capability, meaning that they give a country a chance to respond to a nuclear strike with a nuclear counterattack because they are so hard to find and destroy beforehand. Should U.S. satellites achieve 24/7 coverage, they could track mobile nuclear missiles from the moment they leave a garrison to when they arrive at a launching site.
In the event of a nuclear crisis, such a surveillance capability could cause strategic instability: China and Russia could even launch missiles in response to a perceived threat to their second-strike deterrent. At the very least, the United States should expect China and Russia to develop sophisticated countermeasures and tactics to protect these mobile assets.
CAMOUFLAGE AND COUNTERMEASURES
To prepare for this unblinking future, the Biden administration should develop guidelines for the kinds of data sets that will be used to train algorithms for satellite-image recognition. Training should be conducted with both real and computer-generated collections built from credible and accurate image data; algorithms must be trustworthy and understandable.
The administration should also consider whether to limit the users with access to near-real-time intelligence—for example, granting access solely to frontline analysts with a need to know—and allow wider dissemination for specific operations and situations only.
The NSC should review and adapt its processes before all this new intelligence overwhelms it. The national security adviser should direct an internal review of processes and staffing to assess the council’s capacity and necessity for ingesting real-time information. Such a review may conclude, for example, that NSC staff members do not need regular access to this type of information and that this collection should remain in the hands of the downstream intelligence analysts who prepare products for the council’s consumption. The findings should inform the government’s National Intelligence Strategy, which directs the tasking, collection, processing, exploitation, and dissemination of information for the intelligence community.
The U.S. military, for its part, must also grapple with how to operate under constant-stare conditions. The luxury of overwhelming military superiority in post–Cold War conflicts has caused the long-term atrophy of U.S. capabilities in camouflage, concealment, denial, and deception.
The Department of Defense should consider nontraditional options for policy, operational, and technical countermeasures that leverage advances in fields such as AI, human behavioral science, and materials science (including nanomaterials and metamaterials) that directly address a future in which adversaries deploy constant-stare technologies. Such technology would help the U.S. military to blend its troops, weapons, and other assets into the physical environment and confuse the adversary’s image-recognition algorithms and intelligence analysts.
A new strategy should inform investments as soon as possible, spurring the U.S. military to spend more on developing next-generation camouflage systems and tactics and less on items that will become outdated in this new environment. With advances in space-based intelligence collection moving faster than ever, there is no time to waste.