CEO Series: Arcfield’s Kevin Kelly on making Golden Dome a success.

This transcript has been edited for clarity.

Mike Gruss
Hello and welcome to the Space Minds podcast. I’m Mike Gruss from SpaceNews. We’re here on the Space Symposium exhibition room floor. This is part of our CEO series, and I’m pleased to have Kevin Kelly join us today from Arcfield. Kevin, thanks so much for stopping by.

Kevin Kelly
I appreciate the opportunity.

Mike Gruss
Let’s start with one of the big buzzwords, maybe for the last year; longer: AI. Everyone wants to talk about it, but I want to start by asking you, How are you using AI at Arcfield? How is it changing how you think, how you operate, just even from a from a day to day perspective, and how you guys are incorporating it into what you do?

Kevin Kelly
Sure, I think, like everybody else, when we first got introduced to AI, the whole leadership team attended a bunch of seminars, and we had some speakers come in, and we talked about the art of the possible. We did some brainstorming, and the first takeaway that we had was, “holy cow, this changes everything.” I’m happy to say where we are today. It’s not changing everything, but we’re using AI in the back office, in our finance organization, in our human resources organization. We’re using AI in our proposals. It’s a huge force multiplier for us. It allows us to do more work. Each of our people to accomplish more in any given day. Most importantly, we’re applying it to the mission space. So we do a lot of model based systems engineering; digital engineering. It’s a labor intensive set of work, lot of hands on keyboards, punching in code, creating digital models out of reference specs and documentation. That’s a very time consuming, laborious process that we have built some tools. We call it intelligent model based systems engineering, IMBSE, that ingests those specifications, all the documentation for the legacy systems, and produces a digital model of representation in digital form of legacy systems and of current systems, so we can start manipulating them digitally right off the bat, it saves us thousands of hours of work.

Mike Gruss
Yeah, that’s what was gonna ask. If you didn’t have that, what would happen?

Kevin Kelly
I’d run out of people. Just to put it bluntly, we started looking at AI and said the only way we’re going to scale is with more and more software coders and systems engineers. I can’t find them fast enough. So this allows a team of systems engineers to do what a building used to do before, and it’s really been a huge force multiplier for us.

Mike Gruss
Is there a downside? Or what are people getting wrong when they when they think about AI, or when we’re at a place like this, and everyone says it’s going to solve all my problems.

Kevin Kelly
Yeah, it solves problems, and it creates problems, just like all technology does. I’ll tell you what keeps me up at night. Where we are with generative AI, it looks at a set of what it thinks is facts. Does not have reason built in. It looks at facts. It does the analysis. It does all the number crunching. It can pour through huge amounts of data and produce an output which establishes itself as a brand new fact. What if one of those facts was false? Happens all the time. They call that a hallucination. We are trusting it almost intrinsically right now. And there has to be hallucinations in the results and in the data. So we’re having to go back in a lot of our models and apply a lot more rigor to proving them to be correct. Otherwise we can have some serious risks.

Mike Gruss
What happens when you find those hallucinations, you you fix it, you go back, you run it again.

Kevin Kelly
It’s about improving the quality of the large language model that you’re drawing your facts from, and sometimes going back and tweaking the algorithms, the AI algorithms themselves. Sometimes the hallucination comes from the math. Sometimes it comes from the data source. And you say, “Oh, hold on. Hold on. From now on, it has to have these markings. Otherwise we’re not going to consider it a fact.” And you can start to narrow the LLMS a little bit.

Mike Gruss
I don’t want to put words in your mouth, but it sounds like it’s also it’s saving you time, but it’s also introducing rigor into a process that’s maybe a little bit new.

Kevin Kelly
Brand new. You know, look, we’re engineers. We do quality control. I always joke, and I say the engineers, they’re inventors, but they also can describe to you very accurately how that thing could kill you, whatever that thing is. So we worry about risk constantly. This is just a new set of risks, and we have to use the right word. We have to apply that same engineering rigor.

Mike Gruss
Is it also giving you a competitive edge, do you think? Like, are you finding, “hey, we can do this now. We can do it arguably cheaper because we’re not paying for the people. But also we’re coming up with solutions maybe we wouldn’t have thought of or had the…”

Kevin Kelly
It definitely is and we’re we’re also trying to keep an eye on the marketplace, because we’re not the only ones developing in this space, and so if I can use AI to create brand new technical functionality in Category A, B or C, it’s really important that I understand what my competitors are doing, so I know where I want to invest my time and money and compete. If I’m third to the fight, probably not a good space for me to invest. But if I can take a leadership position and I can leapfrog my competition, or at least increase the gap between me and my competition. It benefits the mission and it benefits the company. So having that market awareness has been really key.

Mike Gruss
I want to switch topics a little bit, which is one of Arcfield’s, priorities or mission areas, is space superiority. That’s a that’s a term that’s getting, again, a lot of buzz here this week. It was a theme of one of President Trump’s executive orders in January. Tell me what that word means to you, and also, like, how maybe some of our listeners should be thinking about it.

Kevin Kelly
So, space superiority to us at Arcfield and to many of our mission partners on the government side, means the ability to operate with impunity without being challenged. To be able to surveil our adversaries in a similar environment, in a similar domain, to be able to communicate with privacy and protection and be able to provide, if it’s a war fighting domain, which space certainly is a war fighting domain, to be able to stay at least one step ahead of our adversary in that space. I’d say as a nation, we got very complacent. We allowed our nation state competitors to enter into that domain, to invest in that domain, to take a leadership position with certain technologies and arguably, and you’ll hear a lot of the leadership here at this conference say, we have lost space superiority. I don’t feel it’s quite that dire.

Mike Gruss
And we’re talking about China, really.

Kevin Kelly
China and Russia.

Mike Gruss
And Russia.

Kevin Kelly
Yeah. The Israelis are, you know, launching some interesting capabilities. Europeans are starting to invest a lot more in space. We’ll see how successful they are. Principally, China’s the biggest threat there.

Mike Gruss
So, what does that mean from a business standpoint, when you say, hey, space superiority is a priority for us. It’s a mission for us. It feels a little bit abstract, right? How do you make it feel tangible?

Kevin Kelly
Attach it to the mission. That’s the most important thing that you do there. I can’t sell anything to a mission partner unless it solves a hard problem, right? So we usually start with brainstorming. What can we do today? What can’t we do today? How do we regain space superiority and situational awareness in advanced orbitology, deception techniques, cryptography, covert communications, heat dissipation, compute. There’s a whole host of different technical challenges in space where we want to regain superiority or or further our superiority. And it really helps that we’re so intimately involved in some of the missions, certainly not all of them. And our experts come back and say, you know, this is one of the things that’s keeping our mission directors up at night. Okay, that’s an area we might be able to invest in. How do we start to solve that problem? If we think we have something compelling, we’re quick to demonstrate we invest money, our own R&D dollars, to solve that problem. And we demonstrate that to our mission partners, to say, are we going down the right track? Are we going to solve a problem here? And they’re pretty honest with us. It’s an open environment. They’re very transparent about I don’t want you to waste your time on that. Actually, that’s that’s a good problem to solve, I want you to double down on that. At the end of the day, that’s how you make a buck in this business.

Mike Gruss
Can you point to one? I don’t want to put you on the spot here, but can you point to one area where you’re like, hey, we helped do this, and now we believe it’s making a difference.

Kevin Kelly
I can. Much of what we do is classified. What I’m about to tell you, of course, is not. The Earth’s aura is a dynamic environment, a plasma and electron field around the Earth that changes in an unpredictable manner. That changing aura affects Synthetic Aperture Radar over the horizon radar, HF communications, and the drag on a spacecraft. We have been for decades, compensating for that with more delta V, more thrust, changing vectors, increasing the power on radar systems, waiting for the aura to change into a lower density model so that we can get certain HF signals through. We haven’t been able to predict it, and we haven’t been able to model it. We invested heavily in that technology to both monitor, predict and model the Earth’s aura. And we’ve produced a product. We produced a software product that we heard from NASA, we heard from the Space Force, we heard from the intelligence community, Department of War. They don’t have anything like this in terms of accuracy, of being able to — we call it dragster — it’s a fun name, but it’s a software model that integrates nicely into the software development toolkit and allows people who are planning and doing orbital architecture and analysis to have a much higher fidelity model for how the drag is going to impact your spacecraft.

Mike Gruss
Interesting. I want to talk about one of these ideas that came up even I think it was hinted at this morning, but this idea of dynamic space operations, which can mean a lot of different things to a lot of different people. You all are involved in one of these Tetra refueling missions for the Space Force. Yeah, talk a little bit about what’s going on there and and also how you think that gets back to space superiority, because it feels to me like that’s all part of the same pillar.

Kevin Kelly
It is. So we are prime on the Tetra five and Tetra six missions. We’re integrating the spacecraft, working very closely with some of the blue chip names that you see in a conference like this, or building different components that go on that spacecraft. If you think about the challenge in the Tetra five mission, it is to provide on orbit refueling at the geosynchronous level, at the geosynchronous orbit to extend the life of our most sophisticated–

Mike Gruss
Crown jewels.

Kevin Kelly
That’s right. Ultimately, what ends a spacecraft’s life is usually you run out of fuel, or you have some anomaly that can’t be fixed from the ground. This isn’t going to deal with the anomalies, but it is going to provide refueling. If you can extend the life of those Crown Jewels, that is a huge cost, imposing advantage you have over your adversary if you can keep them in operation. But there’s a lot of physics challenges with that, rendezvous and proximity operation, transference of fuel, docking, undocking, being able to perform those refueling activities covertly in some type, in some cases, or at least with discretion. So there’s operational mission challenges, there’s a ton of physics challenges. The best thing I can say about that program is we’re very proud that the Space Force has trusted us with such a complex and important mission. Number two, we benefit immensely from the transparent, almost badgeless environment that we work with the Space Force and our subcontractors putting all the problems on the table. A lot of engineering talents being applied by both government, AFRL Space Force, us to solve these hard problems and you know, not worried about who’s going to look bad at the end of the day. Or, geez, I want to, I don’t want to be the first one that delays launch. No, no. We want to get ahead of all of those things. To me, is one of the most refreshing engineering programs I’ve seen.

Mike Gruss
I know we spoke with with General Bratton earlier this year, and he said he has some skepticism about the idea of refueling in space, and that there’s some people who are just wondering, like, Hey, does the business case close? Does does it make the most sense from a technological standpoint. How do you, obviously, you’re spending a lot of time and money on this, but how do you respond to those kind of, I wouldn’t necessarily say criticisms, but concerns about winning hearts and minds there?

Kevin Kelly
I would say those comments apply to every first time mission. Nobody’s done this before, so there’s a lot of challenges.

Mike Gruss
Not a surprise.

Kevin Kelly
No, yeah, that’s right. And we don’t know what we don’t know, right? So we’re gonna figure it out along the way, but I’m thoroughly impressed with that’s not a foregone conclusion. But I’m thoroughly impressed with the talent both the government and the industry has applied to this problem. I have a strong feeling we’re gonna solve this all right.

Mike Gruss
We’re hitting all the buzzwords today, but the other big topic is Golden Dome, missile defense. Tell me — you guys have some capabilities you’re working on there, particularly with sensors when it comes to hypersonic vehicles or missiles. Tell me just on a broad level, how is Golden Dome a success to you? Like, what does success look like to you for that program and what has to happen to get there?

Kevin Kelly
So I’ll take it up a level. Success is like the Strategic Defense Initiative. It is demonstrating the capability in front of the world that, again, serves as a cost imposing strategy that either discourages them from continuing their efforts to create the offensive weapons, or, at a minimum, cost them so much time and money that they either lose focus or fall into economic failure. In order to do that, we have to technically succeed. So now you get back to the program itself. President Trump has been very clear, certainly, General Guetlein and his whole team about a schedule and demonstrable capability within a certain period of time, and then an architecture that stitches all of that together, inclusive of all the multi jurisdictional domains, the FAA, the FCC, the Coast Guard, you name it. NOAA. There’s tons of non warfighting elements that need to be woven into the architecture to do command and control, because if you are talking about a hypersonic inbound, you have precious seconds to track, identify, target and destroy, so there’s the physics challenge, there’s the political challenge. First things first, how do you eat an elephant? One bite at a time. Start with the technology. How do I detect it? How do I target it? How do I destroy it? Let’s start demonstrating all of those capabilities. Let’s get an architecture up and running.

Mike Gruss
A lot of that’s already there.

Kevin Kelly
A lot of it’s already there. And I applaud General Guetlein and his team for taking a look at what we have. Instead of saying, let’s start over, and imagine that we’re starting from scratch.

Mike Gruss
And I think that’s the way a lot of industry has been talking about it. Hey, the tools are there, the capabilities are there. What we have to work on now is how we integrate.

Kevin Kelly
Yeah. In addition to that, I see industry coming together ahead of even written requirements at this point. Say, Hey, you guys have that sensor. You demonstrated that, and I’ve got this on orbit capability, and it links with, you know, it does long range fires, or does sensor to shooter capability. I have some analytics software. Industry started collaborating right off the bat to solve the problem. Of course, there’ll be a dollar to be made.

Mike Gruss
The budget’s there for it.

Kevin Kelly
Yeah, the budget’s there for it. But I’ve been impressed with both government and industry pulling together and saying, What can we do with what we have?

Mike Gruss
Have you started working with industry or competimates that maybe you didn’t expect to?

Kevin Kelly
Yes, and you know what? It’s not terribly different than other engagements we’ve had. We’ll see where it ends up. We might end up being competitors. We might we might collaborate with them, but it gives us an opportunity to decide. Is there a business there, and have we solved the technical problem? Can I solve it without them? Is it in our collective best interest to solve it together? The usual business question

Mike Gruss
What do you still want to see from General Guetlein’s office? Obviously, he’s been meeting with a lot of industries, doing, you know, private sessions or classified sessions. But what are you still hoping to learn about Golden Dome in the next year, let’s say?

Kevin Kelly
I’m impressed with how they’re meeting with the investment community, private equity startups. I’m impressed with how they’re meeting with universities; finding out what we have today, identifying the gaps. They’re doing much of that in the classified domain. Perfectly reasonable that they do that. I’m not seeing a broad area architecture. That is one, and it’s a bit self serving. We do that. We’re not the only company, but we, we do those system of systems, digital models of the entire architecture. I think it’s time. We should be be turning our attention to development.

Mike Gruss
Especially with the deadlines that are coming up.

Kevin Kelly
Yeah. So those sensor programs, those directed energy programs, those intercept programs, can all understand what the architecture is likely to look like.

Mike Gruss
I want to talk about the space economy for a second, because, as everyone knows, we’re here at Space Symposium, the question I wrote down is, “Is there too much optimism right now?” And so I’ll ask that, but that seems like a funny question, because only a year ago, there was not as much optimism. I think there were questions about what DOGE would look like, how it would impact, for lack of a better term, the space community and the space economy. But right now, it’s a completely different, you know, tell me if you feel differently, but it feels like it’s a completely different mood. And so when you talk to folks who aren’t part of the space community, what do you tell them? What do you how do you describe this moment we’re in now?

Kevin Kelly
I liken it to the early days of the Internet, the dot com…

Mike Gruss
The mid 90s, late 90s?

Kevin Kelly
Yeah, mid 90s, early days of digital wireless communications, CDMA, and the 3G 4G 5G transition, optical communications. There’s a there’s a bit of a dark cloud in there, though, and that is in every one of those instances, we overbuilt the industry. But the way capitalism works, you need options, right? So you’re going to build three, four, five competing solutions in order to neck it down to the two that matter. And I think we’re in that early phase. There’s a lot of investment, a lot of optimism. We’re solving a lot of hard problems. But will all of these companies and these solutions be here 10 years from now, in the same space? Probably not.

Mike Gruss
No.

Kevin Kelly
There’ll be some combination and mergers.

Mike Gruss
But the floor here looked very different 10 years ago too.

Kevin Kelly
That’s it’s a good point. It’s going to continue to evolve. It doesn’t discourage me at all. I like the idea that we as a country can innovate faster than anybody else. We seem to have limitless interest in investing capital to solve hard problems, and we’re okay with failure.

Mike Gruss
How will you know if it’s going too fast, though, if there is that “building too much,” the numbers aren’t meeting. What will that look like? How will we know?

Kevin Kelly
Two possible scenarios. One is a peace dividend. The bad thing goes away. Yeah, we have plenty of opportunity to invest in human spaceflight. There’s a lot of dual use here in this hall, which is great.

Mike Gruss
Another hot term.

Kevin Kelly
But if the bad thing goes away, the proverbial, you know, Berlin Wall falls, and you end up scratching your head saying, I’m not sure I’m going to invest in the same thing next year that I did this year. That’s, that’s one potential sign. Or just a lot of business combinations, a lot of mergers, a lot of synergies within the marketplace. That’s a sign that we’re starting to find that rhythm, that, you know, coalescence, and we’re going to settle into what the industry will be. We’re on such a rapid up slope right now, I can’t imagine when that’s going to be.

Mike Gruss
You mentioned innovating faster. Speed is always a theme at one of these conferences. It’s always a theme with the space community, particularly national security. Right now, I think there is this emphasis on speed and getting products delivered, but that’s part of that matrix, that school we hear a lot about, where there’s also capability, there’s also cost. What’s changing? How are you thinking about this moment when it comes to maybe speed is king. Or maybe you don’t think it is.

Kevin Kelly
No, we definitely do. We’re a country that doesn’t pay for second place or third place. So first to market,

Mike Gruss
Someone made a Ricky Bobby reference.

Kevin Kelly
There you go! That’s perfect. Speed is always a factor. And when we decide we do not have limitless dollars or limitless people to invest in solving a problem, we take a look at how quickly can we get there, and how fast is my competitor getting there? We all want to solve the mission challenge, but there’s more mission challenges than we have solutions for at the moment. So speed is probably the number one factor. That is going back to where we started this conversation, an area where AI has a huge impact on what we can do. So where we can apply AI and pick up an additional pace, you know, let one person do the work of 10 people. That’s a huge force multiplier, and it increases our speed. That’s probably the simplest way to answer that. But speed is always a factor.

Mike Gruss
Last question here, what’s one thing you think folks are getting wrong about space right now?

Kevin Kelly
This is gonna sound… I should decide if I want to say this. Space is more than [low Earth orbit (LEO)].

Mike Gruss
You want to elaborate?

Kevin Kelly
Low Earth orbit is where most of I see most of the investment going. It is very important. It is the easiest from a communication standpoint, has the least amount of delay. It’s also going to become the most congested. It already is the most congested. It is the most vulnerable to our adversaries. So we have to think about orbits more than the low Earth orbit. I liked some of your previous conversations around the [medium Earth orbit] with SES and some of the capabilities that we can do there, going to [Geostationary orbit] and going to cislunar and thinking beyond… To the layman. Space is LEO. That’s where the space station is. That’s where, you know, that’s where astronauts orbit. That’s where Starlink is. You got to think beyond. I think that’s one of the things, just as an industry, when I speak to investors and people who are new, they think everything is low Earth orbit. A lot of things are; it’s not everything. We got to think beyond it.

Mike Gruss
Kevin, thanks so much. This was a lot of fun. Thanks so much for joining us. This has been Space Minds.