What a Night on the Runway Reveals About AI and Critical Infrastructure
Last Sunday evening I was sitting on a fully loaded Alaska Airlines aircraft at Ronald Reagan Washington National Airport, engines running, positioned on the taxi runway and ready for departure to San Francisco. Then everything stopped.
Air traffic control issued a full ground stop. No aircraft landing. No aircraft taking off. Engines off. The reason, as our pilot eventually learned and relayed over the loudspeaker, was that a large-scale event on the South Lawn of the White House, involving military jet flyovers, paratroopers, and helicopters, had been communicated to neither ATC nor the airport authority in advance. The pilots had no information. The supervisors they escalated to had no information. Nobody in the chain of responsibility for one of the most heavily trafficked airspace corridors in the country had received advance notice until the event was already underway.
Compounding matters, multiple pilots in the preceding days had filed formal complaints with the FAA reporting that powerful lighting installed for the event was blinding flight crews during their approach into DCA. As the Washington Post reported this week, the disruption to DCA operations is not yet over. No flights are scheduled to depart after noon on July 4, with additional ground stops possible across the coming weeks as further events are confirmed. We sat on the runway for more than two hours that evening. Our pilot, a twenty-year veteran of commercial aviation, described it over the intercom as the most frustrating experience of his entire career.
I do not usually write about politics, and this is not a political piece. What I want to talk about is something else entirely: what this kind of failure reveals about the state of critical infrastructure, and why the race to bring AI into these systems is not a technology story. It is an operational necessity.
The Gap Between Where Aviation Is and Where It Needs to Be
The DCA incident is a vivid illustration of something that aviation professionals have been raising for years. The National Airspace System, for all its sophistication, remains fundamentally reactive and human-dependent in ways that create acute vulnerability when coordination breaks down. The consequences of that vulnerability are not abstractions. They play out on runways, in control towers, and in the lived experience of passengers and crews who have no visibility into what is happening or why.
This is precisely the gap that the FAA is now working to close with AI. As reported by The Air Current in April 2026, the FAA is quietly developing an AI-powered air traffic management system called SMART, Strategic Management of Airspace Routing Trajectories, which could enable the agency to plan for bottlenecks and anticipate scheduling conflicts before an aircraft even leaves the ground. It represents a fundamental shift from today’s human-centric, reactive ATC structure toward something genuinely proactive. FAA Administrator Bryan Bedford has described the agency’s current traffic management systems as “glorified calculators.” That candor is striking, and it underscores just how significant the modernization challenge actually is.
Why This Is an Investment Thesis, Not Just a Policy Story
At NextStar Venture Partners, we invest in AI for critical industries because the consequences of failure in those industries are not measured in lost revenue alone. They are measured in disrupted lives, compromised safety, and the erosion of public trust in systems that people depend on without being fully aware of how fragile the underlying infrastructure can be.
Aviation is a clear case. So are energy, healthcare, advanced manufacturing, and cybersecurity, each of which faces its own version of the same structural challenge: legacy systems built for a different era, now operating in an environment that demands something far more intelligent, adaptive, and predictive. The founders building AI for these sectors are not chasing incremental efficiency gains. They are addressing gaps that, as last Sunday evening demonstrated, can materialize without warning and affect thousands of people at once.
The deployment window for companies building in this space is opening, and the organizations that will ultimately procure these solutions are beginning to define what they need. That alignment between institutional readiness and technological capability is exactly where the most important venture opportunities tend to emerge.
We left DCA two hours and fifteen minutes late. The delay was an inconvenience. The conversation it points toward is considerably more important.
