AWS Summit DC 2026 | Josh Perrius, Booz Allen Hamilton & Greg Clifton, Intel

Clips
More from AWS Summit DC 2026

Greg Clifton

General Manager for the Defense & National Security Group

Intel

Josh Perrius

Senior Vice President

Booz Allen Hamilton

play_circle_outline Cloud-Driven Artemis: Booz Allen and Intel Fuel Lunar Missions as NASA Moves Trajectory Simulations to AWS

play_circle_outline Booz Allen Re-architects NASA Artemis Trajectory Simulations on Intel-Powered AWS Cloud

play_circle_outline Moore's Law and generational silicon advances driving orders-of-magnitude performance improvements.

play_circle_outline Modernizing Legacy Ground and Mission Software to Handle Commercial Satellite Growth, Data Deluge, and Constellation Complexity

play_circle_outline Edge AI and inference on lunar and space-based hardware under SWaP constraints

play_circle_outline Startups and Niche Spacecraft Providers Deliver Advanced Onboard Compute for Artemis 3 and 4

Info
Transcript

Josh Perrius, Booz Allen Hamilton & Greg Clifton, Intel

Greg Clifton

General Manager for the Defense & National Security Group Intel

Josh Perrius

Senior Vice President Booz Allen Hamilton

In this interview from AWS Summit Washington, D.C. 2026, Josh Perrius, senior vice president at Booz Allen Hamilton, joins Greg Clifton, general manager of government go to market and sales at Intel, to talk with theCUBE + NYSE Wired's Gemma Allen about how cloud computing and advanced silicon are enabling NASA's return to deep space with the Artemis program. Perrius explains how Booz Allen rearchitected Artemis trajectory calculation software — previously running on on-premises government servers — to operate in the cloud on AWS using Intel chips, enabling f... Read more

explore Keep Exploring

How does the partnership work behind the scenes, and what do people not realize about the computing, simulation, and hardware efforts required for the Artemis mission? add

How were the thousands of calculations and simulations required to plan and validate the Artemis mission beyond lunar orbit accelerated or migrated to the cloud? add

How are high-end compute resources, cloud bursting (e.g., using AWS), and AI being used for real-time trajectory analysis on recent missions, and how will increased compute power support future human lunar landing efforts? add

What is the primary driver of increasing complexity in modern space systems, and how does it affect computing, data analysis, and the need for system modernization? add

What approaches are being used to enable onboard AI and heterogeneous computing for space systems, and which hardware and software innovations are most relevant to future space missions? add

What plans does Booz Allen Hamilton have for upcoming Artemis lunar missions and for investing in and partnering with commercial space companies? add

bolt Powered by CUBE AI

Josh Perrius, Booz Allen Hamilton & Greg Clifton, Intel

search

Gemma Allen

>> Welcome all to theCUBE coverage here, live on the ground in Washington, DC for the AWS Summit 2026. It is our last two guests of the day, but we are truly saving one of the really feel good stories of 2026 so far for last. We bring on these next two. Joining me now is Josh Perrius from Booz Allen and Greg Clifton from Intel. Talk about Artemis. What's not to love about that? Welcome, folks.

Greg Clifton

>> Yeah. Thank you. Thanks for having us.

Josh Perrius

>> Thank you. Appreciate it.

Gemma Allen

>> So when we think about these missions and the news around Artemis, and it was so exciting, I watched it on Instagram. I don't know about you. What a lot of folks don't realize is it takes a lot to do this, right? But it's not just a space mission, it's a data mission, it's a hardware mission. It's a meeting of many minds.

Josh Perrius

>> Right.

Gemma Allen

>> Talk to me, maybe I'll start with you, Josh, about how this partnership works underneath the hood. What do people not realize?

Josh Perrius

>> Sure. Yeah. Space in general is hard and people in the space community love to say that, right? But this is a particularly complex mission. We have not been out beyond really LEO or low earth orbit with humans since the '60s when we did... In the early '70s with the original Apollo program. So there's thousands of calculations that have to be done to kind of prove out in simulations to make sure that these paths are going to work for taking the Artemis out past the moon and back. And so that's a huge amount of computation that has to get done and it has to be done over and over again to make sure it's right. So we've worked to take that capability, which was running in on-prem servers at a government facility with NASA and rearchitected it to put it into the cloud on AWS and running on Intel chips. And so that's kind of how it came together. We're the software and they're the hardware and getting that to speed up that whole process so that we could do that better and faster and in a more reliable way.

Gemma Allen

>> Intel, I mean, what a company. You've guys been on quite the journey.

Greg Clifton

>> Yeah.

Gemma Allen

>> I jumped off-screen. I wish I bought that stock. I'll tell you that much. Talk about, you mentioned the original Apollo mission, right? Neil Armstrong didn't have a lot of the technology, a lot of the hardware, a lot of the metal, the visualization that we have now. Talk about what's been evolving from the perspective of hardware, space tech. And I'm also interested to ask this next, why did it take so many years to do this again? But let's start with the hardware.

Greg Clifton

>> Yeah. I think on the hardware side, Moore's law is still alive and I think we're even moving faster than that. So as Josh said, the math to do this and to do that trajectory analysis and do it in real time takes high end compute capability. But also I think the beauty of this working with AWS was the ability to burst beyond that on-prem environment to get that capacity on demand. That was for that trip. I think the cool thing too is what it's going to take for the next journey to actually land a human on the moon is going to take even more compute power. So we're actively working that now, looking at the latest technology and really, the technology matters. As you move from an older generation to new generation, you get orders of magnitude better performance, particularly from an AI perspective, which I think you guys are using AI in a tremendous way to plan this journey and what we're trying to do. So yeah, we're always trying to push the envelope and we're trying to push it with technology that's designed, engineered and manufactured here in the US.

Gemma Allen

>> So you both, both of these companies have been, had a longstanding relationship with NASA.

Greg Clifton

>> Right.

Gemma Allen

>> Back when NASA was the only show in town really, right, from the perspective of space tech, Space tech has gotten very popular, very excitable, very quickly. We see that. I mentioned to you off camera that I talk to a lot of entrepreneurs in this space and they all kind of lead with the same thing, which is that yes, space is extremely innovative and radical for many ways, but the technology and the operating systems and all of the hardware and software that goes into it is often decades old. Talk about what's evolving though. Talk about what those conversations look like as these external pressures enter the space too. And I'll start with you, Josh.

Josh Perrius

>> Sure. Yeah. So we've been a big part of actually that modernization, especially on the ground and in the software world at Booz Allen, so both in NASA but also in the broader Department of Defense and in the intelligence community as well. So there's a lot of those systems needed to modernize for many different mission reasons. I mean, but the biggest one I would say if we had to pick one thing driving the complexity is really the reduction in cost around launch that has led to this proliferation of so many more satellites in space, which makes for much more complex constellations to be managed, but also so much more data coming down. And so when you put those two things together, it just really ramps up the computing that's really needed from both an analysis perspective, and management, the algorithms that go into scheduling and sort of figuring out what these sensors are going to look at and when, all the sensors that are used to gather data around even things like Artemis where the main focus is accrued human flight to go do this thing, but it's covered in sensors and so there's tons and tons of data coming down that needs to be analyzed and understood and shared between different entities who are doing different parts of the mission.

Gemma Allen

>> And let's talk about the silicon for space, right? Because sure, it has its own nuances, its own requirements, et cetera. It varies. That type of technical capability sits on different types of chips, right? It needs specific requirements. What's happening? Where is the innovation happening from the perspective of Intel? What do we not see when we think about the metal?

Greg Clifton

>> Yeah. I think it's in a lot of areas. Even you talked about the commercialization, which is bringing costs down, but it's also moving innovation forward. It's the same thing we've seen in any device, whether it was the Internet of Things world, other spaces where the capital that gets put in and the speed drives that innovation. I think what we're doing is try to bring state-of-the-art capabilities to that environment, whether it's on the ground or in space We're working with partners to enable that so that they can move fast and again, use the latest capability rather than we all heard the stories about the space shuttles had, less compute than an iPhone, that's changed dramatically. There's some significant capability going on these devices up in space and in the ground.

Gemma Allen

>> So we hear a lot about AI on the edge. It's having a unique moment, you could say. I don't think anyone knew what that meant two years ago. Now it's like a big, big, big billion, billion, multi-billion dollar bet. There's nothing more on the edge than the moon, right?

Josh Perrius

>> Right.

Gemma Allen

>> I mean, it's hard to imagine how far edge you could go. But from the perspective of what's required, like we think about things like lower compute, like power, all of those things. How does that kind of change from the perspective of the innovation cycle, both from a software inference perspective and from a hardware perspective? What makes the build and the innovation unique?

Greg Clifton

>> I mean, I'll answer from our perspective. One of the things we're doing is we're building system on a chip. So we're putting CPUs, GPUs, and NPU, a neural processing unit all on the same device. So now you can have a multiple capability. I can run a mission thread on the CPU, I can do inference on the GPU and I can do a longstanding AI run on the NPU. So trying to give that heterogeneous architecture flexibility so you can do lots of different workloads on that one compute device.

Gemma Allen

>> And from a software perspective, right? Because again, it's as remote as you can possibly imagine.

Josh Perrius

>> Right.

Gemma Allen

>> How do you think about that in terms of the future, like from an innovation perspective? So much is happening so quick, there's some phenomenal innovations. What are most relevant to the space world?

Josh Perrius

>> Yeah. Well, I think what probably a lot of people didn't know before is that satellites traditionally and the things we put into space were not that smart because of those constraints around SWaP, size, weight and power and some of the challenges around cooling, those things that you barely had a sensor up there that put data on the ground and a lot of the smarts would happen in the software that was written on the ground. But as emissions have gotten more complex and you have like... When you're putting then actual humans and things up there and not just satellites, that really moves the need to put that compute closer to the edge, but also the size and because of the innovations we talked about around launch and stuff like that and the ability to put more and more up is giving you the ability to put a more and more robust compute set at the edge. And so that's what's exciting is, so there's things we have to do around the software around making sure it can run in smaller environments and Booz Allen has done that. We tailored a large language model to run on the International Space Station to get it small enough to kind of fit on the compute there. We're doing space domain awareness type algorithms that can run on camera sets on small SATs and things like that. But very excited about the lunar mission is that there's going to be actually a lot of really good compute there, but the challenges are going to be a lot around latency and the ability to kind of move data back and forth and the communication paths. So trying to figure out all that architecture is really exciting and figuring out what are the... Constantly working with the manufacturers and hardware partners on what exactly is going to be available at the edge so that we can tune and customize the software to work best in that environment.

Greg Clifton

>> And I think you're seeing that big time, even it started on AI PCs where it's developed from the ground up to operate disconnected. And so that technology that started out here and the ISV ecosystem that built it is moving into edge, into robotics and drones and now into other harsher environments like space.

Gemma Allen

>> Let's talk about the partnership for a second because you mentioned that there's a lot of conversations happening at a hardware software level. This is like a huge multifaceted, fast innovation, right? But at the end of the day, you both have been in business with NASA for a long time. Space is a competitive space. There's a lot of entrepreneurs with a lot of promise, but you guys are steady, reliable, long term innovators. How do you think about that from the perspective of this entire ecosystem we're in right now and the competitive forces at play?

Josh Perrius

>> Well, I mean, I think from our perspective, we try not to rest on our laurels, right? We have this long history over 70 years with NASA and in the broader space community, but the way you stay there is you have to keep changing for the mission, right? Adapt to what new technology is out there. And that's why the partnerships are really important to us because we think we're smart and we like to come up with great ideas, but it's always better when you have more people contributing. And so a great example was in this, in the Artemis mission. So we were using Xeon processors from Intel that were specific for high performance computing and it worked really well and we were able to see this huge improvement in performance. But as we look to the future missions for Artemis, like we talked about, they're going to be even more and more complex. So we're partnering on how to tune and adjust those algorithms to be even more performant, take better advantage of the capabilities of the new chips that's coming and kind of looking forward with that. And we wouldn't be able to do without that partnership. We wouldn't see those things ahead of time, wouldn't be able to do that. And so that's a win for both our companies because we're both learning and working on a better outcome, but it's a huge win on the mission for the government because they're getting a much better solution.

Gemma Allen

>> And we think about the communication rhythm, right, the building process, the co-build, the co-innovation that needs to happen. Time is of the essence here, right? So the cadence, I'm sure is much shorter than it was before where folks built in silos, put them together and hoped for the best, right?

Josh Perrius

>> Right.

Gemma Allen

>> No, that's not what's happened now. Talk about how this relationship iterates even, like how you guys innovate together as a partner network with NASA and with many other partners

Greg Clifton

>> Yeah. I think it's a lot of testing, trial, error, but that's much faster now. Even you do your DevOps up in the cloud, you don't have to stand up your own environment to do that anymore. Even on the last question, I think you look at space, it's just another environment, another challenge. We've done that over the years when we first went out to the edge or when we first developed cloud or now in AI. It's just another interesting problem to go after and see how can we connect hardware, software, networking to solve a unique, in this case, size, weight, power, environmental challenge.

Gemma Allen

>> Well, it's certainly a great story. What other great stories are on the horizon, folks? Are there any big bets that you both have on the calendar on the roadmap for the rest of 2026, 2027, something we could hope to be talking about a year from now, and start with you, Josh.

Josh Perrius

>> Well, yeah, I mean, I think the future Artemis missions are one. I mean, if you've seen what the national administrators talked about, they're rapidly changing. Even the plans that we thought we had in place a year or two ago are going to be greatly accelerated with many more Artemis missions coming. And so we're working really hard on making sure we're ready for all those. The Artemis III will be a low earth orbit test, but then Artemis IV, again, back to the actually testing, getting onto the moon again and actually landing there and the future missions like that. So there'll just be more and more complexity around that. We're continuing at Booz Allen too to invest in space. We invest in the startups as well, to your point about all the entrepreneurs in that space, we're trying to find additional partners. We have long-term partners who are obviously huge multinational conglomerates that we work with, but we're also trying to find these little niche players as well to see who we can partner with. So there's new spacecraft providers who are bringing new compute into space that we're interested in. So we're working with them on the different sensors and missions that we can put onto their hardware as well.

Gemma Allen

>> Wow.

Greg Clifton

>> And I think for us, just edge in general, we're spending a lot of time in physical AI, robotics. We're working in healthcare, we're working in manufacturing. So it's going to be interesting, with what we've done from a compute perspective, we're solving a lot of the size, weight, power challenges. As I mentioned that multiple different workloads on one device. We're going to keep pushing the envelope. I think there's all kinds of problems that we can solve across lots of different mission spaces, whether it's government, industry, education.

Gemma Allen

>> Well, the world is certainly your oyster or the moon.

Josh Perrius

>> Sure.

Gemma Allen

>> Folks, thank you so much for joining us on theCUBE.

Josh Perrius

>> All right.

Greg Clifton

>> Thank you.

Josh Perrius

>> Thank you very much. It was great.

Gemma Allen

>> I'm Gemma Allen and that is a wrap on day two here at the AWS Summit in Washington, DC. We had a phenomenal today's talk to some great minds of industry, leaving with a lot of hope for what's ahead for this community and this company for 2026 to 2027. Thank you so much for watching.