Interview with US FAA Administrator Bryan Bedford

Transcript

Kris Van Cleave:Bryan, thanks for joining all of us here. Very busy man here. You're coming up on 1 year as the FAA Administrator, and, you know, about 13 months ago, DOT announced this major overhaul of air traffic control. You guys have had the money for About 7, 8 months now. Am I doing the math right?

Bryan Bedford:Got it in July.

Kris Van Cleave:Okay. So let's, let's just start with a status report. Where are we in the effort to modernize air traffic control, spend that $12.5 billion on making the system better?

Bryan Bedford:Well, so the, the process itself of modernization is more complicated than just what's in the $12.5 billion. So when we think about modernizing the system, there's What you hear referred to as BNATS, the Brand New Air Traffic Control System, and that's where the $12.5 billion investment in infrastructure is focused. That's pulling out, you know, 10 million miles of copper, you know, telephone lines and replacing it with either high-speed data, low-Earth orbit satellite, 5G, you know, looking for ways to improve the data transport system for the FAA system. That process on just the telecom piece, again, one of 14 discrete workstreams, is Right. About 53% complete. When we started the process, it was about 10% complete. So we've moved the bar about 40% over the last 12 months, and we'll finish the process by 3rd quarter of 2027. That's on time. Then we have to actually start building all new voice and data switches. They have to be able to speak analog and digital because guess what? The FAA today is a completely analog creature, and we need to move it into the 21st century, into a digital And we have to do that at the same time we're moving 18 million IFR flight trajectories a year and 1 billion passengers. So we have to run the current system while we're building the new system and then ultimately move from, you know, TDMI to VOIP. So all of those workstreams are either on or ahead of schedule to complete the project by 2028.

Kris Van Cleave:And where are you from a reliability standpoint? You know, last summer was rough. Saw the issues around Newark and outages sporadically in several spots. Where do you feel like the operational reliability is today?

Bryan Bedford:Better. Certainly far from optimum, but we're better today than we were last year. We'll be much better next year than we are now. So we are still far, far away from what I think we would consider to be the reliability in the system that we want. But again, I think we're going to have a much better summer in 2026 than we did last year.

Kris Van Cleave:And I should say, we are using the Slido system. So if there are questions, I'll try to get to some of those too as we're going through here. So I'm sure that they will put the barcode up, the QR code up, if you don't have it already. And we'll keep that in mind too as we're talking here. One of the things I thought was really interesting was the talk about replacing the radar systems. But those are also systems that nobody builds here. So how do you, how do you stand up a radar manufacturing industry, if you will, in the US and get those by the hundreds installed in 2.5 years?

Bryan Bedford:I think this was one of my favorite moments, you know, watching as we contract for the supply chain on radar. The initial government cost estimate to replace 612 radars was just under $1.9 billion. So we started the process with 3 providers, Thales, Indra, Raytheon, and the team was negotiating, you know, pretty hard. And I won't reveal all of our tactics, but suffice it to say, over the course of the last 30 days of that negotiation, we took the price down from $1.8 billion down to $1.06 billion, and with strong economic incentives for the, the radar providers to deliver on schedule, and with enough flexibility that if we see someone falling behind, we can actually pull supply from them and give it to the overproducers. So there's a lot of incentives for them to meet our targets. And the FAA, you know, we set an internal deadline to finish all supply chain acquisition by December 31st, 2025. And they closed that contract, literally signed the last supply chain contract at 11:59 PM on New Year's Eve. So the team really worked very, very hard to get it done and meet our deadlines. But we will, as part of those contracts, reshore radar production. So we have a facility that is building radars now in Kansas City and one in St. Petersburg, Florida.

Kris Van Cleave:Okay. You know, when this is done, when this $12.5 billion is spent and these projects are completed, what should industry and the traveler expect? What's going to be better? What's going to be different?

Bryan Bedford:So if we just stopped at BNATS, the $12.5 billion, deploy it, you know, remove all the '70s, '80s, and '90s technology and replace it with, you know, 21st century digital tech, we'll have a really super reliable system. But all of the inefficiencies that are essentially designed into the current, current airspace system, you know, that will be unchanged. So we'll have a very reliable but still very inefficient system. National airspace system.

Kris Van Cleave:So you'd have the same system with the same the same places where it's maxing out now.

Bryan Bedford:That's correct.

Kris Van Cleave:That would continue to be there. That the limits would would stay the same.

Bryan Bedford:That's right. Which we don't like, right? I think there's a follow-on question. So what are you going to do about that? You've heard Secretary Duffy speak about a follow-on investment. The 12.5 is going to be fully deployed. It's going to get us. You know, to the finish line on equipment modernization. But that still leaves a— it leaves a data architecture that today is managed by 313 compact computers. Right. The way the system was designed back in the '70s and '80s essentially looked at the compute power that we had, and it built the system around the data limitations. Everything is inside the facility. So on-prem compute power. And those data limits are essentially lifted into the upper airspace. So we have, you know, thousands of— or not thousands, hundreds of sectors that are being managed by controllers based on their limited view of data. So we don't have in any of our facilities the ability to see the entire national airspace picture in a unified way. So by spending this next $10 billion to re-architect the data systems to get it from the facilities into the cloud, And to pull in a common automation platform that will allow us then to build significant efficiency into how we build trajectories for the national airspace.

Kris Van Cleave:How confident are you that you can get $10 billion more from Congress?

Bryan Bedford:Well, I trust the president and the secretary. They promised they were going to get us the funding, and I believe they're going to deliver the goods. I don't know how Congress could look at the system today, and we've invited them into our facilities. And many have taken me up on that offer. I have done many congressional visits to our facilities, and it's eye-opening just to see the mass of copper wires that are running in, to actually see the floppy disks sitting on top of the, you know, the computer server. It's a shocking recognition of the fact that the system is woefully obsolete and needs to be brought into the 21st century.

Kris Van Cleave:Of this $10 billion, the Secretary is very excited about, uh, you implementing, integrating AI into the, the systems to, I guess, do just do everything better. Um, can you talk about what, where that AI software, how that factors into building a better system?

Bryan Bedford:So there's, I guess, 2 critical use cases here for machine learning or artificial intelligence, none of which, by the way, are removing, you know, controllers from the decision-making process. One way is something that will actually happen this year, and it has nothing to do with the $10 billion. You start thinking about how we build trajectories today. Essentially, they're— it's no different than the federal highway system.

Kris Van Cleave:Right.

Bryan Bedford:All the airlines are flying essentially the same trajectories to get from, you know, Miami to New York. And you can imagine there are times of the day when the highway's congested, and therefore the FAA starts to implement programs. I'm sure I get it almost every week, edict time of 20 past the hour, or, hey, we're doing extended miles in trail, or we're doing airborne holds. So we started a process that we call SMART, so Strategic Management of Aviation Routing Trajectories. And in the SMART system, we challenged 3 vendors essentially to bring machine learning into the design and construction of how those trajectories would be managed. And all 3 of them built digital twins for us. You know, they're both— or all 3 of them are producing representations of how the airspace actually works at about a 96% confidence level. So we— And they're all getting the same answers. So I think that's reassuring to us that, that it works. And what we're seeing is a system that is woefully unorganized. And so what happens is we are allowing all the airspace users today essentially to schedule however they want. And it's a free-for-all that ultimately the FAA and frankly the controller in a facility, arriving facility, has to sort out. We can do a much better job of sorting that out on the front end. And when we, you know, pick a day, we looked at February 12th, 2026, because it was a beautiful, clear day with no weather in North America. And we still ran at about 82% on-time rate. And I think we had 0.2% cancellation. So it was a pretty good day from that effect. But all the, all the delays are driven by the fact that we peak the schedules to the constraint that the airport can't support the runway capacity, the gate capacity, the taxiway capacities completely in gridlock. And so we can sort all of that stuff out before the flights leave the ground. The other thing that we saw that was really, frankly, surprising to me is on a, on a normal day when we have, say, 50,000 flight trajectories, we were seeing anywhere between 3,000 and 8,000 of those trajectories were actually designed to collide.

Kris Van Cleave:Wow.

Bryan Bedford:With one another because they're all using the same, you know, flight planning systems. So what we have is, of course, tactically deconflicting traffic with our en route facilities. So again, super safe. But what if we could actually build trajectories before the airplanes are actually crewed and scheduled? What if we could look at the schedule, say, in September, deconflict it, and optimize those trajectories, removing all of the block hour pads that are in there?

Kris Van Cleave:Right.

Bryan Bedford:The savings on that schedule is about $1 billion a month just in terms of crew cost, fuel savings, and maintenance. So I think we've got a real opportunity to do something transformational with AI. The second thing that we see it, of course, is in a common automation platform. So today, controllers are using legacy technology stacks— ERAM, STARS, ATOPS. The conflict probes in those systems today give us about an 18-minute line of sight on where the aircraft are going. Of course, radar's painting a picture of where the airplane was 60 seconds ago. ADS-B gives us about a 6-second latency on data. So we know where the plane was, you know, but AI can tell us not only where the airplane will be in 18 minutes, but where the airplane's entire line of flight is and where all of those lines of flight intersect. And we could deconflict. One, we can deconflict before we leave the ground. But, you know, bringing machine learning into the actual Common Automation Platform will give controllers a line of sight that when they deconflict something, it's deconflicted for the entirety of the remaining route, as opposed to just their small sliver of the airspace.

Kris Van Cleave:And do you envision the AI more as a force multiplier, or is— are you going to be looking for the machine to be controlling traffic, directing traffic?

Bryan Bedford:So again, come into the lab and I'll show you some of the possibilities. But the The setup today is essentially the machine is probing all of the potential routing trajectories and looking for those conflicts and then advising us that it sees something that needs attention. If we do it right on the beginning, we shouldn't have many of those exceptions, but the controllers today have a maximum of 19 planes that they can have under management at any one time. Imagine if we could take that to 50 or 100 by having AI essentially focus the controllers on where the real need is, as opposed to probing every single flight that's in their, their space.

Kris Van Cleave:And if you're saying that, you know, thousands of flights basically get put on potential collisions because their courses— because they're routed the same way every day, deconflicting that is a— becomes a huge part of the controller workload every day, right? Like, you're Somebody's got to adjust, shave, shave seconds off, move, move somebody, whatever. That's all workload at your controller level.

Bryan Bedford:That's today's workload, yes, which is why we limit the amount of airplanes that are in any one sector to a number that's, you know, 19 or lower.

Kris Van Cleave:So if you alleviate that, you are, you are addressing one of the concerns with your workforce in that just the, the workload on each controller.

Bryan Bedford:The workload should go down. Therefore, the amount of work that they could could manage in any one particular sector could increase. Yes.

Kris Van Cleave:Let's take a couple of questions that that have shown up on on the Slido here. Are you talking to the military at all about freeing up airspace off the East Coast? You know, particularly on those routes to Florida during peak times.

Bryan Bedford:I will tell you the the. FAA has a great relationship with the Department of War. We have people embedded in terms of their airspace domain needs, and we have military embedded with us. So there are very strong lines of communication there, and there is always a tension between airspace availability and closure. Obviously, we'd like more of it, and they would like us to have less of it. And that tension is always going to be there. So I can't tell you that I see any change in the current posture from what we've experienced in prior years. I can tell you the friction's there.

Kris Van Cleave:And in theory, if you're able to integrate, build a smarter system, you could use your, your, the commercial lanes more efficiently, right? Which Would that— that could address some of the congestion, particularly on the—

Bryan Bedford:So the airspace redesign process, I think, is going to evolve over 3 years. First, we have to sort of build the foundation, you know, the data architecture, the technology stacks to support what we ultimately want to do in terms of really unlocking all the technology that are on the aircraft to actually maximize, you know, the efficiency from point A to point B. AKA reduce time and space, right? And then the follow-on to that will then be building many, many more RNP procedures. Really, how can we start to sequence traffic at top of descent so that we get efficient sort of power-off descent into the terminal area right to the runway? Again, AI can help us with this because it's just— it's a math puzzle. And guess what? Machine learning is really good at solving math puzzles. Complicated math problems.

Kris Van Cleave:Let's see. Do you have any update on ATC and CMO staffing levels?

Bryan Bedford:Well, we published our 2026 controller workforce plan. Huge brouhaha over, I think, some misreporting that the FAA wants to slash staffing by 2,000 positions. So let's just cover the facts. Today we have about 11,000 certified professional controllers deployed across the United States. That's the highest level that we've had since, I don't know, 2000. So we're, we're, we're in a better place than we have been, but it's still woefully inadequate. We have a defined need now of 12,600 controllers. So from that perspective, you know, the FAA thinks it's 1,600 controllers short. The Air Traffic Controller Union believes that number should be 14,006. I think that's a red herring. The fact of the matter is we're short, and we are hiring and training just as fast as we can. So the academy is full, the pipeline is full. We have 1,000 fully certified controllers that have changed facility locations, and when they do that, they go from a productive status to a nonproductive training status. So I've got a lot of talent in the pipeline that has demonstrated that they can certify as a controller, but they're in a new facility. So I think we're— again, we can see light at the end of the tunnel on controller staffing. When it— when on the CMO side, we're again hiring as fast as we can find people that are qualified to do the work.

Kris Van Cleave:Let me be a contrarian for a minute. If If 12,600 is good, I had a science teacher in middle school that would say if, if some is good, more is better, usually before he blew something up in a lab. Why not set the goal for, for more controllers than you need?

Bryan Bedford:It's a great question, and it really comes down to how we deploy them. If, if a facility is at a healthy level, which is defined in the collective bargaining agreement, then it can no longer receive new controllers. And so our deploying controllers really depends on how much unhealthy staffing at facilities are out there. So if I say the Buffalo tower is not healthy when it really is, I can continue to take training academy resources or other fully certified controllers and allow them to redeploy to Buffalo when I really need them at New York Center. And so it's important to actually try to get the staffing right where we can so that I'm not putting controller resources in places, frankly, there just isn't traffic to manage.

Kris Van Cleave:And it's too complicated of a system for there to be like a ready reserve of controllers that are sitting reserve, so to speak?

Bryan Bedford:So in a, in a perfect world, we would have full staffing at a location and not need to drive overtime or roster somebody to come in and work on a day off. But those are the tools essentially. We can either increase staffing synthetically through overtime or junior-manning them to work on a day off, or we can reduce demand, right, by, you know—

Kris Van Cleave:Nobody likes that suggestion. Exactly.

Bryan Bedford:Well, those are the, those are the two options we have today.

Kris Van Cleave:Let's see, what, what else are people wondering? Um, how can, how can we as an industry help you achieve your goals? Um, and is there anything specific you think you need from industry to, to help succeed, particularly in modernizing the system?

Bryan Bedford:So the Modern Skies Coalition, I think, has been invaluable. Uh, if we go back to 2017, you may recall President Trump in his first administration announced that he wanted to, you know, build a brand new air traffic control system. And then the industry hijacked that narrative and turned it into a privatization debate. I think we learned the lesson that we don't want to have that debate until the system is actually modernized, and then we can argue over should it be a government entity or a private business organization and who should pay for it. So I think Modern Skies, you having a unified industry voice on the Hill, within the administration, pushing modernization, this next $10 billion is critical to get the job done. Again, if we stop with just Replacing the equipment will have great reliability, but we'll still have massive inefficiency. So if we really want to get the true benefit of modern architecture, we need that second tranche of investment.

Kris Van Cleave:What's your biggest concern that we're not talking about, whether that be the industry, the media, the public?

Bryan Bedford:Well, so maybe it's the sense of urgency to get to the, to the right end state. If what I see from my perspective is You know, drones going from 1,000 operations— or sorry, 100,000 operations last year to 1 million this year to 10 million next year. So drones are for real. They're happening. And it's going to be a growing industry across the airspace. You know, advanced air mobility, it's going to happen. We've got, you know, 8 EIPP sites now. We're going to start to see these things, I think, deploy in 2027. And then, you know, SpaceX, Blue Origin, ULA. I mean, there is a massive increase in space launch cadence. I think our team will reduce the impact of the space launches on the system, but that's only going to be consumed by more cadence. So we have got to get better tools so that we can manage airspace more efficiently because the innovation that is coming into the airspace environment, and we're going to have a food fight over who gets those resources when what we really need to be arguing about is how we grow the pie. And so I want to focus on growing the pie, not arguing over who gets what slice of it.

Kris Van Cleave:But while you're slicing the pie, who does get the bigger slice? I mean, where are the priorities as we look to bring in drones and eVTOLs and manage that demand with You know, the airlines and the space launches and all of the other— the general aviation, the business aviation?

Bryan Bedford:Well, so for drones, we'll use hard rules. Stay below 400 feet. Stay within, you know, outside 10 miles of terminal control space. So we'll sandbox drones in, in a way that we can all feel comfortable that they're separated from you know, commercial traffic or general aviation traffic. But there will come a point in time where drones are going to go places that will need deconfliction. And that's going to be an intersection between, you know, our CPC workforce versus unmanned traffic management. And we haven't figured that out yet. We're working on that as we speak.

Kris Van Cleave:And is that a place where ADS-B or some level of standardization—

Bryan Bedford:Electronic conspicuity? I always have a hard time saying conspicuity. But yes, I think this gets to a larger issue of just broadcasting in general. I think this is where I think sometimes the FAA gets in its own way. We want perfection when we can We could actually deploy things that are really good, maybe not perfect. ADS-B In is one of those solutions. We can— I mean, at a very low price, you can get antennas and receivers and electronic flight bag-enabled solutions that have both visual and aural alerting, which is clearly better than no ADS-B In situational awareness. So I do think that we are going to have to nudge, push, or perhaps even rulemake our way into a solution.

Kris Van Cleave:Some, some, some level of don't let good be the enemy of great on bringing safety to play. You know, one of the things that, that, that sort of struck me when you were talking to Congress was this idea that, you know, the system as we know it has reached its limits at points. And we sort of saw that— see that play out at Chicago O'Hare this summer. There seems to be, if you listen to some of the airline executives, limitless demand for their product. But there are limits to what the skies can control. Are we really— are the skies really full? And if we are reaching our limits, what is— what is that growth, that whether it's drones, eVTOLs, or just, you know, airline to be named later's aspirations of more service?

Bryan Bedford:So I think our long-term goal is for the FAA not to be the resource constraint on the National Airspace System, whether that's due to technology or workforce size or how they're deployed or the airspace design itself. But at some point, we have to recognize that airports are limited by concrete and gates. And as we get— and I think that's where we're seeing, you know, peak demand that can't be satisfied. I don't know the last time you guys have been in O'Hare. A lot of construction going on there. No end-around movements for aircraft. Everything is a runway intersection to migrate through the airport. We don't like that, honestly. I mean, you know, departures and arrivals happening simultaneous with aircraft crossing runways in order to get to gates is not the best recipe. But that's what we're trying to manage there. So I will tell you, I believe, and our conclusion was that the ramp situation at O'Hare was not supportive of the increase in operations that we saw.

Kris Van Cleave:Did the unique use-or-lose gate lease structure there sort of facilitate that, what looked like an arms race?

Bryan Bedford:I think the way the gate allocation works in O'Hare leads to a misalignment of incentives. And we've certainly had that conversation with the Chicago Airport Authority.

Kris Van Cleave:Is that something that FAA could step in and say, stop that?

Bryan Bedford:We cannot. We are not a party to that contract.

Kris Van Cleave:And so the capping the flights was the only— the best tool you had?

Bryan Bedford:Yeah, it's heavy-handed, but we got to the right result. And we're going to see, I think, a great operation in Chicago this summer, which was not the case last year. And of course, you know, that there's a ripple effect, you know, when things start to bog down, whether it's New York airspace, D.C., or Chicago, or Atlanta. As things bog down in one space, it affects the entire efficiency of the NAS. So it's important to get these high-density markets right.

Kris Van Cleave:We've got 2 minutes. So let's, in our lightning bonus round, Where, where, where do you see the state of certification in general, whether it's seats or, or new airplanes?

Bryan Bedford:Great question. We have put a lot of resources into our, our aircraft cert team in anticipation of, of eVTOLs primarily, but certainly I think we want the FAA to be much more collaborative. There are some folks who would look at that and say, oh, that's, that's not the way that relationship should work. You're a regulator and a compliance organization. You're not a collaborative organization. It's self-interesting, frankly, for us to want to partner with industry to understand what they're trying to accomplish, because it helps us align our resources so that we're not the bottleneck. And it's uncomfortable for industry to actually want to share that information with the FAA because they do assume we're going to use it in some way to penalize or prevent them from reaching their goals. So creating dialogue between our aircraft cert team and some of our manufacturers in the United States, it's been a slow process, but I think we've evolved to a level of trust now, and it's helping us again make sure we understand what's the future work in process look like so that we can manage the and ultimately increase the throughput of our activity that we can support. So again, it's a change of mindset, but I think it's designed to help unlock innovation. We'd love to see Boeing produce, you know, the next mid-market aircraft here in the US. We'd like to see us first to market. So how can we start having that conversation today to make sure that we aren't the bottleneck for achieving that outcome?

Kris Van Cleave:And certainly it's balancing safety, but also with some level of predictability, right?

Bryan Bedford:Look, I think this is where, you know, Kelly's got the right mindset. You know, Boeing, first and foremost, has to design great aircraft, and then they have to build them at a high level of quality. And that, frankly, is the lowest-cost solution for Boeing. Build a great airplane, manufacture it right the first time, not having to do the rework. I think he's totally bought into that, and our team and their ODA now, I think, are rising to Do you foresee, uh, the 737s certified by the end of the year?

Kris Van Cleave:I do.

Bryan Bedford:I think we'll get the MAX 7 first, um, and then followed by the 10, and hopefully the 777 early next year.

Kris Van Cleave:We are out of time. Any final thoughts?

Bryan Bedford:No. Hey, I appreciate, again, all the support here, um, especially when you speak with one voice through Modern Skies. So I just encourage you to continue to support our efforts. Out in the field, and, and we'll try to deliver this before the end of 2028.

Kris Van Cleave:FAA Administrator Bryan Bradford, thank you for your time, sir.

Bryan Bedford:Thank you.

Kris Van Cleave:Appreciate it.

Copyright policy: All transcripts on this site are the copyright of CAPA - Centre for Aviation. Our reproduction policy is as follows: you may quote up to 400 words of any transcript on the condition that you attribute the transcript to CAPA - Centre for Aviation and link to the original video page. All other use is prohibited. While we aim for 100% accuracy in the transcript, there may be some minor transcribing errors.