Subscription Required

Open to All Readers

By Vincent E. Bianco III

April 30, 2026, © Leeham News: Venture capital is flooding into autonomous air traffic control. The investor enthusiasm is rational. The regulatory assumptions behind it are not.

The money Is pouring in for Air Traffic Control (ATC), and it has nothing to do with the Federal Aviation Administration (FAA).

Over the past 18 months, a wave of startups has emerged claiming they can automate pieces of ATC using artificial intelligence. Y Combinator backed Enhanced Radar Systems. GTMfund led Oureon Technologies’ $3.5m pre-seed. NoamAI debuted at Airspace World 2025. Venture capital firms that wouldn’t have known what a TRACON was two years ago are now writing checks to companies promising AI-powered air traffic control.

On its face, the investor thesis makes sense. Only 23 of the FAA’s 313 staffed ATC facilities are fully staffed. The controller workforce is aging, recruitment is lagging, and training pipelines are years behind demand. NextGen—the FAA’s multi-billion-dollar modernization program—is roughly 16% complete after $7.5bn invested. The drone market is projected to see 800 million additional flights over the next decade.

The eVTOL companies, Joby, Archer, BETA Technologies, and Wisk, have collectively invested billions to certify aircraft that will operate in airspace where no traffic management infrastructure currently exists.

The market drivers are real. The pain is real. The technology is advancing. So why should anyone pump the brakes?

When Ideas Outrun Reality

For one thing, we’ve already seen what happens when aviation startups outrun their regulatory reality. Lilium, the German eVTOL developer, went public through a SPAC merger in 2021, burned through hundreds of millions in capital, and filed for bankruptcy in late 2024—before its aircraft ever reached certification. The technology was real. The timeline wasn’t. Autonomous ATC startups face the same fundamental risk: a product that works in a demo environment but hits a regulatory wall that the funding can’t outlast.

Because every one of these startups is selling investors a regulatory story. And most of those stories don’t survive contact with the FAA’s actual institutional architecture.

I say this not as someone opposed to innovation in ATC. I spent my career trying to modernize it from the inside. I say it as someone who knows what the FAA actually requires before new technology touches the National Airspace System. The gap between what these companies are promising and what the regulatory road actually looks like is something investors deserve to understand before they write checks.

What Investors Aren’t Reading

In July 2024, the FAA published the Roadmap for Artificial Intelligence Safety Assurance, Version I. It’s a 31-page document, signed by the Associate Administrator for Aviation Safety, that establishes the agency’s foundational principles for how AI enters aviation—airborne and ground-based systems alike. It’s public. It’s on faa.gov. Virtually no startup pitch deck or VC due diligence process I’ve encountered references it.

That’s a problem, because this document lays out several principles that directly constrain what autonomous ATC companies can and cannot do.

The first is Work Within the Aviation Ecosystem. The FAA is explicit: you don’t get to build your own regulatory pathway. You work within the existing certification framework, or you don’t deploy. For startups accustomed to the tech industry’s “move fast and break things” ethos, this is a cultural collision, not just a regulatory one.

The second is Avoid Personification. The FAA instructs industry to treat AI as a tool, not a digital human. Marketing language about “AI controllers that mirror human interaction” runs directly counter to the agency’s view of how the industry should think about these systems. This isn’t a branding suggestion. It’s a design philosophy that shapes how the FAA evaluates your product.

The third and most consequential distinction is between Learned AI and Learning AI. Learned AI, static models trained offline and deployed with fixed parameters, can be certified using adaptations of existing software assurance frameworks like DO-178C. Learning AI — adaptive models that continue to evolve in the operational environment, including reinforcement learning systems — presents a fundamentally different challenge. The FAA has no certification method for learning AI in safety-critical applications. None. The roadmap acknowledges this explicitly.

No Pathway for Certification

Several autonomous ATC startups are currently hiring reinforcement learning engineers. They are building learning AI for a regulatory environment that has no pathway to certify it. Their investors should be asking about this.

For context: the only FAA-approved machine-learning system in a safety-critical aviation application is ACAS X, the collision-avoidance system. Its development involved NASA, MIT Lincoln Labs, and the FAA Technical Center. It took over a decade. That’s the benchmark, not the exception.

The Four Phases—and the Walls Between Them

Autonomous ATC isn’t one product. It’s a spectrum, and each point on that spectrum has a different regulatory reality.

Phase 1: Data Aggregation. Collect publicly available data—ADS-B, weather, NOTAM feeds—and present it to users in a useful format. No FAA certification required. Deployable immediately. Revenue-generating today. This is where the smart companies are starting, and there’s nothing wrong with it. You’re building operational credibility and data assets without needing the FAA’s permission.

Phase 2: Advisory Services at Non-Towered Airports. This is the first wall. There are roughly 165 non-towered airports in the U.S. with scheduled commercial service and over 500 more with significant general aviation, training, or charter traffic, all operating without ATC.

An AI advisory system at these airports would provide genuine safety value. But here’s the catch: the FAA has a documented pattern of permitting new technologies to operate in a regulatory gray zone, then asserting authority retroactively after an incident. Companies that engage the FAA proactively in this phase will survive the transition to Phase 3. Companies that don’t will discover the FAA’s enforcement posture the hard way.

Phase 3: Certified Advisory at Staffed Facilities. The wall gets higher. NATCA enters the picture. The National Air Traffic Controllers Association represents approximately 20,000 FAA controllers and holds contractual authority over technology deployment at staffed ATC facilities. This isn’t optional; it’s baked into the labor agreement. NATCA will support tools that reduce workload and improve safety. They will oppose anything positioned as reducing headcount. The difference between “force multiplier” and “controller replacement” isn’t just marketing. It’s the difference between adoption and political death.

Phase 4: NAS Infrastructure Integration. The wall becomes a fortress. Integrating with STARS, TFDM, or ERAM requires DO-178C software certification at the highest assurance levels, cooperation from platform owners like Raytheon and Leidos, Department of Homeland Security and DoD security clearance for national critical infrastructure, and validation at the FAA’s William J. Hughes Technical Center. No autonomous ATC startup has achieved any of these milestones. Terminal Flight Data Manager, or TFDM, the FAA’s own billion-dollar program to replace paper flight strips with electronic ones, has been in development since 2016, was cut from 89 sites to 49, slipped its completion date from 2028 to 2031, and still hasn’t delivered its core surface metering capability. That’s the institutional velocity any company planning NAS integration needs to calibrate against.

The critical insight is this: each phase transition isn’t a product upgrade. It’s a regulatory threshold that requires different FAA engagement, different certification standards, and different stakeholder relationships. Startups that pitch investors a continuous product evolution curve are misrepresenting the landscape. The curve has cliffs.

The Airspace Nobody Is Managing

There is, however, one market segment that most startups and investors are overlooking — and it may be the most investable near-term opportunity in the entire autonomous ATC landscape.

eVTOL companies have invested billions in approaching Part 21 type certification. Joby is in the final stages. Archer is close behind. Upon certification, these aircraft will operate commercially in an altitude band between the Part 107 drone ceiling,  400 feet AGL, and the base of traditional ATC-managed airspace. That middle layer has zero traffic management infrastructure today. No controllers, no UTM services, no separation standards, no deconfliction protocols designed for vehicles that take off vertically, transition to forward flight, and share corridors with both drones below and manned aircraft above.

And the altitude band isn’t uniform. It varies by location depending on the underlying airspace classification. Near a Class B airport, the overlying shelf might start at 1,500 feet. In rural uncontrolled airspace, Class E typically begins at 700 or 1,200 feet Above Ground Level (AGL). The corridor changes shape with the geography, which makes the traffic management problem harder — and strengthens the case for an intelligent coordination platform.

The company that builds the coordination layer for that airspace doesn’t just have a product. It has a platform the entire Advanced Air Mobility ecosystem depends on. And unlike Phases 3 and 4 of the traditional ATC autonomy roadmap, this market doesn’t require displacing human controllers or integrating with legacy NAS infrastructure. It requires building something new in the airspace that nobody currently manages.

Technology That Fills a Gap

I spent 23 years inside the FAA watching technology programs die because they tried to replace something that already existed. The programs that succeeded — TCAS, LAANC, data comm — filled gaps that controllers and pilots could feel. The eVTOL integration layer is that kind of gap. Nobody is being displaced. Nobody’s workflow is being disrupted. You’re providing a service in airspace where no service exists, for aircraft types that didn’t exist five years ago. That’s a fundamentally different political and regulatory proposition than walking into an FAA facility and telling controllers their job is about to change.

The FAA is already moving on this. In January 2026, the agency created a dedicated Office of Advanced Aviation Technologies to oversee drones, eVTOLs, and other AAM systems — the first new top-level office in the FAA’s organizational chart in years. The Beyond Visual Line of Sight (BVLOS) rulemaking, proposed Parts 108 and 146, is in active public comment with over 900,000 submissions and a presidential executive order mandating finalization.

Part 146 in particular creates the regulatory framework for Automated Data Service Providers,  the formal term for what the UTM community has been calling UAS Service Suppliers. For the first time, the FAA is building a certification pathway specifically for the kind of third-party traffic management services that autonomous ATC companies want to provide.

The eVTOL Integration Pilot Program selected eight projects across 26 states in March 2026, with first operations expected by summer. Joby, Archer, BETA, Wisk, Electra, and Reliable Robotics are all participating. These companies will be flying pre-certified aircraft in Class B and C airspace alongside commercial traffic. Someone needs to manage the coordination. The question is who.

Contenders and Pretenders

What does a credible autonomous ATC startup actually look like?

It starts with the team. A founding team with FAA operational experience, aviation certifications, or a demonstrated history of navigating federal certification processes brings institutional credibility that no amount of seed funding can buy. A team with no aviation credentials isn’t necessarily doomed, but it faces a steep learning curve and a credibility deficit with the FAA that adds years to any realistic timeline.

A credible company has decomposed its product vision into specific automated tasks with identified safety criticality levels, as the FAA’s Safety Framework for Aircraft Automation explicitly requires. It has engaged specific FAA offices through defined regulatory mechanisms—not “expanding engagement with the FAA” in vague terms, but naming which office, which pathway, which submission.

It has a NATCA-awareness and early-engagement strategy. It carries aviation product liability insurance. And its deployment timeline acknowledges the phase transitions rather than compressing a decade of certification into a three-year Series A runway.

A credible company also understands which FAA offices it needs to talk to. The UAS Integration Office, specifically AUS-400, the Safety and Integration Division, runs the NTAP process and is the front door for UTM service providers. The Aircraft Certification Service runs the AI/ML technical discipline through the STEP Program, where Dr. Trung Pham’s team is developing the certification methodology for AI systems. These are two different offices, under the same Associate Administrator, with different processes and different institutional cultures. A company that doesn’t know this distinction is a company that hasn’t done its homework.

Safety Risk Management Plan

The Near-Term Approval Process itself is instructive. NTAP requires a Service Provider and Champion Operator to submit jointly a concept of use, concept of operations, service level agreement, requirements traceability matrix, and a safety risk management plan. The FAA doesn’t evaluate technology in isolation. It evaluates whether a specific service reduces risk in a specific operational context. Companies that show up to the FAA with a platform, but no operational partner, are starting the conversation wrong.

The warning signs are equally clear: regulatory strategies built on FAA orders that don’t apply to the product’s actual function. Marketing language about “replacing controllers” or “AI that mirrors human interaction.” Timeline claims are disconnected from every precedent in FAA certification history. The absence of any mention of NATCA, the 20,000-member union with contractual authority over exactly the technology deployment these companies are proposing.

The contrast between discipline and hype is already playing out in the broader electric aviation market. BETA Technologies, the Vermont-based eVTOL and eCTOL developer, waited until it had a flying product, a manufacturing facility, a 50-site charging network, and strategic backing from General Dynamics and GE Aerospace before going public in November 2025.

Its CX300 had already received special airworthiness certification from the FAA. The company has flown over 100,000 nautical miles, made the first passenger-carrying electric flight in the U.S., and was selected for seven of eight FAA eIPP projects.

Lilium went public four years earlier with a vision deck and ran out of money before certification. The autonomous ATC space will produce its own versions of both stories. Investors should be asking which one they’re funding.

The most fundable companies in this space right now are those with modest near-term ambitions and realistic regulatory roadmaps, not those with the boldest vision decks and the vaguest compliance stories.

Deploying capital, wasting capital

The autonomous ATC market is real. The demand is urgent. The technology is maturing. Capital should flow into this space.

But capital deployed without understanding the regulatory terrain will be capital wasted.

The FAA is not the enemy of innovation in air traffic control. The agency published an AI safety roadmap, established a technical discipline for AI and machine learning, created a dedicated office for advanced aviation technologies, and is actively developing certification frameworks for autonomous systems. But the FAA operates on aviation time, not Silicon Valley time. The National Airspace System is the most complex, safety-critical system humans have ever built, and the agency charged with protecting it will not shortcut the process because a startup’s Series A runway is running short.

Europe is moving faster. EASA issued its first U-space service provider certificate in May 2025. The SESAR Joint Undertaking published the U-space Implementation Handbook in January 2026. Startups that can deploy in European markets first while building their FAA relationship may find the fastest path to revenue and operational credibility. The global regulatory landscape matters, not just the American one.

Capturing value

The investors who understand the regulatory architecture — and fund companies built for aviation timelines — will capture the value this market creates. The investors who believe pitch deck timelines will fund expensive lessons.

The companies that will successfully automate pieces of the NAS will be the ones that respect its complexity rather than dismiss it. That’s not a constraint on innovation. That’s the price of admission.

I’ve watched FAA technology programs come and go for over three decades. The ones that worked all had something in common: they were built by people who understood the system they were trying to change. The ones that failed were built by people who assumed the system would change for them.

The autonomous ATC gold rush is on. The gold is real. But the miners who understand the terrain will be the ones still standing when the dust settles.

---

About the Author

Vincent E. Bianco III is a 23-year veteran of the FAA Air Traffic Organization who served at five ATC facilities across California and Nevada, held six years of facility management positions, including Air Traffic Manager, and completed two FAA Headquarters tours focused on NAS technology integration. He is an FAA Safety Risk Management Practitioner, Boeing Certified SMS Champion, and FAA-licensed Private Pilot. He is Principal Consultant at Marivin Consulting Services LLC in Santa Rosa, California.