In April 2025, President Trump and Defense Secretary Pete Hegseth stood in the White House and gave the Air Force’s next-generation fighter a name: the Boeing F-47. The designation made official what the Pentagon had been building toward for years. The F-47 is not designed to win dogfights alone. It is the crewed command node at the center of a much larger system, one that pairs a human pilot with a team of AI-driven drone wingmen known as Collaborative Combat Aircraft, or CCAs.
If the program delivers on its planning targets, each F-47 pilot would direct up to eight of those autonomous drones in flight. With roughly 185 crewed fighters in the projected fleet, the math points to a combat network of approximately 1,500 platforms. Those figures, drawn from congressional testimony and Air Force budget documents over the past two years, remain planning estimates rather than locked procurement numbers. But they sketch the scale of ambition behind the Next Generation Air Dominance program, known as NGAD, and they explain why the Pentagon is moving fast to get both halves of the system airborne.
The CCA drones are already flying
The uncrewed side of the equation hit a milestone before the crewed jet was even publicly named. A CCA prototype completed flight testing less than two years after the program launched, a pace that Air Force Secretary Troy Meink called evidence of “acquisition speed and delivery of combat capability.” That timeline is notable in a department where major aircraft programs routinely slip by years.
The Air Force has split CCA development into two increments. Increment 1 contracts went to Anduril Industries and General Atomics, each tasked with producing autonomous aircraft that can fly alongside crewed fighters and execute missions with limited pilot input. Increment 2 is expected to field more capable variants. The first flight proved the airframe works. What it did not prove is whether the full teaming architecture, with one pilot managing multiple drones through contested electronic environments, functions at operational scale.
Why the Air Force wants this network
The strategic logic is straightforward: mass matters, and the United States no longer has it. The Air Force’s fighter fleet has been shrinking for decades, from over 3,000 tactical aircraft during the Cold War to fewer than 2,000 today. Meanwhile, China has been expanding its own air force and is developing what U.S. officials describe as a sixth-generation fighter alongside its own loyal wingman drones. Beijing’s defense buildup is the explicit reason NGAD exists.
Autonomous wingmen offer a way to rebuild mass without rebuilding pilot pipelines, which take years and cost millions per aviator. A CCA drone does not need a decade of training, does not fatigue on long sorties, and can be risked in missions too dangerous for crewed aircraft. If the drones are cheap enough and capable enough, a relatively small number of F-47 pilots could project far more combat power than the same number of legacy fighters flying alone.
That “if” carries significant weight. The F-47 itself is not cheap. When the Air Force briefly considered restructuring NGAD in 2024, one driver was the crewed fighter’s projected unit cost, which reporting at the time placed above $300 million per jet. The CCA drones are intended to cost a fraction of that, but the Air Force has not published firm unit-cost targets for fleet-scale production. The total price tag for a 1,500-platform network, including sustainment, software updates, and communications infrastructure, remains an open question.
The pilot workload problem
Controlling one or two drones from a cockpit is a manageable cognitive task. Orchestrating eight across a threat environment where adversary electronic warfare could sever data links is something else entirely. The Air Force has discussed autonomy levels that would let CCAs execute pre-planned tasks without constant pilot oversight, essentially giving each drone a set of instructions and letting its onboard AI adapt to changing conditions. But the specifics of that software architecture, its failure modes, and the rules of engagement for autonomous weapons employment have not been detailed publicly.
The challenge is not just technical. It is doctrinal. How much decision-making authority will reside onboard the drone versus in the F-47 cockpit? What happens when a CCA loses its data link and must act independently? Who is accountable when an autonomous system acts on imperfect sensor data? These questions sit at the intersection of software engineering, military ethics, and operational law, and the Pentagon has not yet answered them in any public forum.
Secretary Meink’s emphasis on acquisition speed suggests the Air Force plans to answer many of these questions through rapid flight testing rather than years of study. Each test sortie generates data on how the drones perceive their environment, how reliably they follow intent-based commands, and how gracefully they degrade when sensors or communications are jammed. That approach borrows from commercial software development, where frequent iteration replaces long design cycles. In a combat-aircraft context, though, the stakes of an undetected software flaw are considerably higher than a crashed app.
What the public record does and does not confirm
As of June 2026, two facts rest on solid ground. The F-47 is a named, funded program with explicit White House backing, and it is designed from the outset as a crewed hub for autonomous teammates. At least one CCA prototype has flown, demonstrating that the drone hardware works and that the Air Force can move from contract to first flight faster than its historical average.
What is not confirmed is the scale of the eventual network. The 185-fighter, 1,500-platform figure reflects planning estimates that have circulated in budget documents and congressional testimony, but the Pentagon has not issued a formal program-of-record commitment to those numbers. Fleet size will ultimately depend on unit costs, production capacity, and whether the autonomy software matures fast enough to justify buying drones in bulk.
The next set of milestones will be more revealing than the first. CCA flight tests are expected to shift from basic airworthiness to autonomous mission execution and multi-drone coordination. Those tests will show whether a single pilot can meaningfully command a team of AI wingmen or whether the concept needs to be scaled back to a smaller, more manageable ratio. They will also begin to expose how resilient the network is when an adversary actively tries to break it.
A bet on software as much as steel
At its core, NGAD is less an airplane program than a software program that happens to involve airplanes. The F-47 airframe gives the pilot a survivable platform to fly into contested airspace. The CCA airframes give the network its mass. But the element that determines whether the whole enterprise works is the autonomy stack running inside each drone and the command-and-control layer linking them to the crewed fighter.
The Pentagon’s track record with ambitious software-driven programs is mixed. The F-35’s software suite took years longer than planned to reach full capability. The Army’s Future Combat Systems, which also promised networked autonomous vehicles, was canceled after billions in spending. NGAD’s advocates argue the technology has matured since those earlier efforts, and the CCA’s rapid first flight supports that claim. But validating autonomy at scale, in an environment where an adversary is actively jamming, spoofing, and shooting, is a test that no prototype flight can replicate.
For now, the F-47 and its planned drone wingmen represent the most concrete step the United States has taken toward a networked, human-machine air combat force. Whether that step leads to a 1,500-platform fleet or something smaller and more cautious will depend on answers the Air Force has not yet provided, and on tests it has not yet run. The hardware is moving. The harder question is whether the software, the doctrine, and the budget can keep pace.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
