Clint Crosier, director of the AWS Aerospace and Satellite business. Photo: AWS

Clint Crosier, director of the AWS Aerospace and Satellite business, believes in the power of bringing together cloud computing capabilities and space innovation. His work at AWS has focused on unlocking the future economic potential in space by allowing customers to access data in orbit and harnessing powerful AI tools to speed up the process and transfer of data in space and back to Earth. AWS aims to be the partner in bridging data for space industries, governments and enterprise markets around the world. Crosier believes that AWS’ recognition as a thought leader in cloud innovation will maintain and expand its leadership in the age of AI.

In this interview with Via Satellite, Crosier, a retired U.S. Air Force/Space Force two-star General with 33 years of service, explains the evolution of the space-based cloud, and how AWS worked for nearly a decade to solve one of the most important operational challenges of the modern, connected age – getting a firm grasp on big data.

VIA SATELLITE: The Director of National Intelligence has said that innovation from AWS will be critical in operationalizing intelligence in space. Why is operationalized intelligence so critical to meeting the needs of U.S. government and military missions in space, and how is AWS fulfilling those needs?

Crosier: During my 33 years on active duty in the U.S. Air Force and U.S. Space Force and now five years with AWS, I’ve seen how we’ve evolved the way we use space as part of our military operations. In the early days, we operated space-based intelligence as kind of a stove pipe that was off to the side. We treated it in such as highly classified manner that, in some cases, the left hand and right hand couldn’t even really communicate about what we were doing together in the space domain. And so, we didn’t have very good operational effectiveness or integration. Over time, we’ve seen those barriers erode and there’s now a broader understanding of the importance of space to our operational picture — whether it’s military forces or troops in the field, or strategic intelligence. Today, the United States wouldn’t ever want to go to war or operate through crisis and contingency without our space capabilities. When we think about operationalizing space and space intelligence, it becomes clear that our ability to ensure that we can provide the right data to the right people at the right time has been key to our growth and success.

VIA SATELLITE: The combination of space and cloud capabilities has been essential to the evolution of other critical government services, like weather forecasting and disaster response. AWS Professional Services has partnered with the National Environmental Satellite, Data, and Information Service (NESDIS), part of the National Oceanic and Atmospheric Administration (NOAA), to enhance the processing of weather data. Can you talk about how AWS will approach this task?   

Crosier: For the past decade, AWS has been in the business of moving the cloud closer to where our customers are — edge computing. There’s a lot written about edge computing, And that’s because we have  found that for our customers it’s easier and more cost-effective to bring the cloud to our customers data, rather than making our customers bring their data to the cloud. When people think about pushing the cloud to the edge, they typically think about pushing it out in remote places and markets in parts of the globe that are underserved or underutilized. That’s all great. My team thinks of that as the “Terrestrial Cloud,” because it’s typically on the ground, on the Earth. However, when space customers think about where they need the cloud, in many cases it doesn’t happen to be on the Earth. It needs to be where they operate — and more and more today, it’s in space. Bringing the cloud to space is a big deal because it enables generative AI, which can be a revolutionary tool for processing weather data at the space-edge. One example is how Iceye is using generative AI to enhance the synthetic aperture radar (SAR) data from their satellites. The cloud enables the AI that enables Iceye to couple their SAR data against social media reports and fuse the two together on a platform they call Flood Insights. In other examples, we’re seeing AI being able to process 100 years of historic climate data and pulling out trends and patterns that allow these agencies to make extremely detailed and precise predictions of future weather activity. Think about the value of being able to predict exactly where the next flood will happen or where the next earthquake will happen. You’re talking about saving thousands of lives.

VIA SATELLITE: Are terrestrial and space-based cloud networks providing similar advantages for operations that take place entirely in space, like in the fields of space sciences and exploration?

Crosier: Yes. Right now, NASA’s Mars Perseverance Rover is moving around Mars and taking soil samples and beaming that data all the way back through the AWS network. It’s been a very effective capability. We want to take things a step further. We want to make humans living and working in space a reality. To do that, we need to significantly increase the speed of those types of data transfers. We can think about asteroid mining, which is a mission that our customers are pursuing today, as an example of an industry that won’t be possible without pushing the cloud into space. We will not be able to port all of the data needed to support mining operations back to Earth and process it in the terrestrial cloud and then push it back into space again at the speed required for effective operations. The latency won’t allow it. Now think about on-orbit manufacturing, robotic servicing, robotics, automation and quantum computing all happening at the same time in space. To make the emerging space economy a reality, we’ve got to push the cloud where our customers need it most.

VIA SATELLITE: AWS has already started to demonstrate these space-based edge computing capabilities. What did you learn from those demonstrations?

Crosier: We did a demonstration two years ago with D-Orbit, where we positioned an edge computing software package on a D-Orbit satellite and ran a number of demonstrations pre-processing the data on the satellite instead of pushing it back to the Earth. And what we found is they could meet 100% of their mission requirements with 42% less bandwidth on the downlink. That’s a game changer. That’s almost doubling the capacity of the satellite by processing on orbit and downloading only the imagery that’s actually useful.

We did another demonstration with Axiom Space where we launched a cloud compute device on a mission to the International Space Station. Think about a  series of complex experiments, like genome sequencing in space, with cloud computing capabilities on orbit. They didn’t have to download everything back to Earth, which can take somewhere between 8 and 12 hours pushing all that data through the small pipes that exist on the ISS. Previously, a 20-minute experiment was typically a 10-hour process. With cloud computing on orbit, it takes 20 minutes.

VIA SATELLITE: AWS has been providing cloud capabilities to help commercial satellite operators improve efficiencies for some time. Are those operators also leveraging edge computing for Space Situational Awareness, which is becoming more and more important as space gets more crowded?

Crosier: In this particular area, we’re taking our lead from the customers, and what their needs are. We want to try to be three to five years ahead of those needs so that when they need to execute these missions, they’ve got the infrastructure available to do it. For Space Situational Awareness, we’ve worked with Leolabs, which operates a system of ground-based radars around the globe. They do space traffic management and collision avoidance for satellites in Low-Earth orbit. Several years ago, when Leolabs was starting to really grow big, they looked at their ability to manage their entire workload on their own – we’re talking about tracking 16,000 objects on orbit today, including some 9,000 satellites, and another 7,000 pieces of debris. And part of collision management is you have to look at all the possible perturbations of 16,000 objects and the probability that they can collide with any other 16,000 objects over the next 24 hours. Project that out to space traffic management and collision avoidance and the math gets pretty big, pretty quick.

If they identified a satellite that had a potential collision, they would reach out to one of the companies involved and the company would ask, ‘How should I maneuver my satellite to avoid the collision?’ Leolabs would have had  to run that calculation with on-premises systems and it could have take them up to eight hours to model. Oftentimes in these situations, the probability of collision is time critical.  Taking eight hours, per run, to find the solution kind of defeats the purpose. So instead, they moved all that work to the AWS cloud, where they can spin up hundreds or even thousands of resources simultaneously. Now, that phone call with the satellite operator can include information about a specific maneuver to avoid collision that was produced in seconds vs hours. Instead of eight hours, they can provide that information almost immediately — real-time space traffic management is only possible because of the advanced speed and capabilities of cloud technology.

As I mentioned earlier, the number of satellites in space is going to multiply by five to ten over the next decade. We absolutely cannot do traffic management in that situation without the high-performance compute. Amazon cares about solving this issue because it is a global sustainability issue.

VIA SATELLITE: Considering how critical these capabilities are to government and enterprise operations, and to the larger space economy, how should we approach protecting space-based cloud computing capabilities from cyberattacks?

Crosier: At AWS, we treat the security aspect of the cloud as ‘priority zero.’ It’s not even priority one. It’s priority zero. You can’t have space without cyber and you can’t really have cyber without space at a global level. The two are inextricably intertwined and you have to be able to manage both of them effectively. AWS is built on a foundation that exists today as the most secure platform anywhere in the world. That’s the ground layer. We help customers on the space layer identify the best security practices for securing their satellites and their data in space. To do that, we have some 200 security tools, protocols, capabilities within the AWS cloud that allow you to secure your data. We’re running security systems that encrypt and protect the data, both in rest and in transit as it interfaces from the ground to the space layer, and we’re working with companies that are developing autonomous capability to monitor the satellite itself and identify when the satellite itself has unusual access or unusual behavior and anomalies. We bring all of these things together and I’m very proud to say that we’re ensuring the greatest level of security anywhere in the world, from the ground to the space layer.

VIA SATELLITE: I want to talk about the AWS Ground Station managed ground station service. How has AWS Ground Station evolved as a managed service and what does it include today?

Crosier: The definition of managed services evolves constantly. We seem to have new managed services every week and every month. I could say that we have 221 different managed services today and by the time this is published, it’s actually 242. We’re always bringing new capabilities to the cloud. I would ask customers from companies of any size, ‘Why waste millions and millions of dollars building out your own cloud computing infrastructure, which doesn’t differentiate your unique capability?’ It’s a common capability that everybody needs. So why put millions and millions of dollars into your capital expenditure to provide that basic capability instead of leveraging the AWS Cloud to provide you all of that capability. We’ll also maintain it, sustain it, and continually upgrade it. You focus on the niche of your mission that makes you special, right? The same proposition of the cloud extends into space. We have 12 AWS ground stations today, and we’ve purposefully built those ground stations in such a way that they’re either in close physical proximity or connected electrically to the AWS cloud. As soon as data comes down from your satellite, it’s nearly instantaneously ported into the AWS Cloud, which then allows you some 200-plus capabilities between artificial intelligence, machine learning, advanced analytics, quantum computing, high performance compute, and everything in between. We manage that capability for you in a way that allows you to leverage best-in-the-world technology to analyze the data, manipulate the data, store the data, and push the data to your customers anywhere in the world securely and with reduced latency—without customers having to invest in their own Ground Stations infrastructure.

Starting August 2025, we have a very exciting change with AWS Ground Station. AWS is expanding its AWS Ground Station as a Service (GSaaS) Partner Program, increasing the options for space customers to bring space data to the AWS Cloud and realize its benefits. As the inaugural partner in the GSaaS Partner Program, AWS is excited to announce an enhanced collaboration with Kongsberg Satellite Services (KSAT), a world-leading provider of communication services for spacecraft and launch vehicles. Satellite operators can now seamlessly access KSAT’s global satellite ground station network and operation services as well as the services and tools of AWS — the most comprehensive and broadly adopted cloud.  By adding AWS Ground Station to their networks, service providers, such as KSAT, can offer additional antenna locations and the advanced data delivery capabilities of the AWS network to satellite operators, without having to make upfront infrastructure investments. A robust network of partners also results in greater service plan flexibility, more operational support, simplified onboarding, and solutions tailored for satellite operator’s mission-specific requirements.

VIA SATELLITE: You’ve recently added digital twin testing capabilities to the AWS Ground Station managed service. What has been the response to this addition?

Crosier: The digital twin is one of my favorite concepts that also spins into artificial intelligence. A digital twin essentially uses cloud computing capability and services to create a digital replica of the system. It could be a digital replica of every building in Washington, D.C., or it could be a digital replica of a 200-satellite constellation. I experienced the value in digital twins firsthand when I used to command the GPS satellite constellation. Anytime we wanted to run an upload to the GPS satellite constellation, we would test it rigorously on the ground. But we had no exact representation of exactly what the constellation looked like with the exact number of satellites. Anytime we’d run an update to the constellation or change to the constellation, there was a little bit of apprehension that it worked in the way we anticipated. Digital twin takes the assumption and hope out of the equation, because we now have an exact replica all the way down to the details. We can replicate all specifics in a situation and run the entire test process in a digital twin, with zero hypothesis about what will happen when you actually run the upload. Another thing digital twins allow us to do is save hundreds and hundreds of hours in spacecraft and satellite design, testing, modeling, simulation, and save millions and millions of dollars.

For example, Lunar Outpost designed their MAPP rover in the cloud, which sped the design process. They built the prototype, took it out to the desert , and actually ran it around in a simulated terrain, and then later, the lunar surface. It actually met the performance specs with only a few modifications. They locked in the design saving millions of dollars and months and months of design and manufacturing costs. And then they built their mission operations center in an AWS environment they call Stargate which allowed the integration of the command and control, design, testing, modeling, and simulation. They did this in such a way that they were using AWS to do all the commanding  of the rover, which gives you operational resilience. When you do cloud-based command and control, you can replicate an operation center in a nearly infinite number of places. As long as someone can open a laptop and connect to the internet, we can replicate the operational environment command and control the rover. Lunar Outpost did all the command and control on the cloud and linked it with digital modeling, simulation, design, and test. Now, whenever they want to send a command to the rover, they can go back and look at all the original test data, modeling and simulation data. They can simulate what it would look like if they run it, all in a single environment. That’s amazing.

VIA SATELLITE: AWS Ground Station has been serving operators working in S/X-band, but how is the company adapting to the multi-orbit, multi-altitude, multi-constellation model?

Crosier: Good question. Watch this space and stay tuned. AWS Ground Station supports a lot of LEO activity in the X- and S-bands, but we know that other customers operate in other bands — from GEO to MEO to LEO, and also VLEO, Very Low-Earth Orbit. We can handle all of these with AWS Ground Station. But as you try to integrate these different orbits, you need some different capabilities. So, we’re looking into that. Stay tuned.

VIA SATELLITE: What is the most exciting thing you’re hearing about or learning about here regarding the future of AI development for space?

Crosier: We’re at the early stages of defining what the future of space operations will look like, leveraging cloud technology, and then we think about artificial intelligence. I grew up in the age when a geospatial analyst would look over data with a little magnifying glass and draw conclusions with limited data. I recently saw a statistic that the amount of data that is coming down from Earth observation satellites today, globally, is close to some 50 petabytes a day — approximately 500 times the entire Library of Congress digital library per day. We are beyond the ability to do any manual assessment of a data set that size. It all has to be processed in the cloud, right? How do you do it securely in real time? We’re seeing companies use artificial intelligence to program change detection algorithms and capabilities so that as soon as the data comes down and hits the AWS cloud, the system can flag for you using artificial intelligence.

Then, there’s generative AI … we’re moving towards this new concept of agentic AI, or using AI agents to go out and perform work. Adam Maher, CEO of Ursa Space, recently pulled me aside, opened up his laptop, and started showing me real world geospatial data against unclassified, real-world military intelligence requirements. He said, “Clint, look, I set up three agents. I sent one agent to advise me when an adversary military is doing an exercise. I sent another to tell me when this activity inside the area of interest occurs. I set up another agent to tell me when I have collection opportunities for these three satellites to take pictures. This was an example of a specific problem, however, Ursa Space has developed a general capability that would be of high interest to commercial companies as well. Commercial companies need to understand risks to their business, for example, understanding supply chain changes.” They have this whole thing running in a single framework — agentic AI — operational problem solving for intelligence, military intelligence applications. Ursa Space is out in front of developing that right now. It’s just so powerful when I think about those capabilities being available, right now.