“In a world characterized by unpredictability and increasingly frequent surprise, there are heavy penalties for ponderous decision making and slow execution.”
–Richard Danzig
The U.S. Air Force is losing the race to field a crewed sixth-generation fighter jet to China. If it loses this race, it will be qualitatively outmatched by the Chinese air force. The U.S. Air Force needs to resurrect the Digital Century Series, which would allow the service to buy and build aircraft using a process-centric, iterative acquisitions strategy. This would enable the Air Force to quickly buy and build new aircraft to remain responsive to the threat environment, incentivize contractors to build new aircraft instead of sustaining old aircraft, and attract new entrants to an atrophied tactical aircraft industrial base. Leaders in Washington are deliberating the direction of crewed sixth-generation fighter acquisitions as some are arguing against the need for this type of fighter, believing that the age of crewed fighter aircraft is giving way to fighter aircraft flown by artificial intelligence agents. The current state of AI-enabled autonomy is nowhere close to mature enough to cancel the acquisition of future crewed fighter aircraft.
How We Got Here
The U.S. Air Force has been working towards the acquisition of a crewed sixth-generation fighter aircraft for a decade. Different leaders have proposed competing visions for the program, and as a result, the Air Force has yet to buy a single operational aircraft. While the U.S. Air Force has been vacillating on how it wants to buy and build crewed sixth-generation fighter jets, the Chinese air force has shown compelling and concerning signs of progress in fielding their own capability.
In 2015, Pentagon acquisitions chief Frank Kendall announced that the Air Force and Navy would begin to pursue developing crewed sixth-generation fighter aircraft. In 2016, the Air Force released its Air Superiority 2030 Flight Plan, referring to a new, fighter-like capability called “penetrating counter air” and identifying autonomy as a “game-changing capability.” Eventually, the Air Force landed on a name: Next Generation Air Dominance.
In September 2020, Air Force acquisitions chief Will Roper announced that a record-breaking, full-scale demonstrator aircraft had flown in the physical world. Afterward, Roper laid out his vision for the program, announcing a bold new vision to build aircraft differently: the Digital Century Series. Roper was inspired by the Century Series of fighter aircraft that the Air Force had built in the 1950s when six different tactical aircraft had their first flights between 1953 and 1956. Roper did not want to re-create the airplanes. Rather, he wanted to re-create the same kinds of programmatic incentives that fielded the first operational aircraft that could fly twice the speed of sound just one year after fielding the first operational aircraft to fly faster than the speed of sound in level flight. Gen. Mark Kelly, the commander of Air Combat Command, supported Roper’s vision, comparing it to the development of the F-117 Nighthawk: the first stealth combat aircraft that was fielded in just five years.
In March 2022, Frank Kendall, who returned to government as secretary of the Air Force, identified Next Generation Air Dominance as one of his seven operational imperatives, saying it would “include a crewed platform teamed with much less expensive autonomous uncrewed combat aircraft.” Kendall had returned to government intent on pursuing select, priority programs that would make a difference in the Air Force’s readiness to contend with China, outlining seven operational imperatives that would be the focus of his tenure. In June 2022, Kendall announced that he would not be pursuing the Digital Century Series. Three months later, Kelly raised his concerns that China could beat the United States to the punch on fielding a crewed sixth-generation fighter aircraft, citing their iterative approach to aircraft development. In May 2023, the Air Force announced the planned award of a contract for a crewed sixth-generation fighter aircraft in 2024.
In July, Kendall announced the pause of the acquisition of a crewed sixth-generation fighter to do a program analysis. In September, Kendall proclaimed that he’d like to be able to field the aircraft for the same cost as an F-35, a marked departure from previous comments. In December, the Air Force announced that it would defer the decision on the program to the Trump administration. Afterward, news broke that the Air Force’s analysis supported the idea of building a crewed sixth-generation fighter aircraft. The Air Force had spent six months studying the problem only to arrive at the same conclusion, all while China was wasting no time pursuing the same capability.
The day after Christmas, videos of two new fighter jets flying over the skies of China were released. Afterward, outgoing Air Force acquisitions chief Andrew Hunter implied that the existence of these aircraft is not a surprise to the Air Force, but acknowledged the pace of Chinese aircraft development as “incredibly fast,” conceding that China could field the world’s first crewed sixth-generation fighter aircraft.
Defense Secretary Pete Hegseth has called China’s progress on fielding a sixth-generation fighter “a dangerous development,” vowing to “look under the hood” of the Next Generation Air Dominance program. Deputy Defense Secretary nominee Steve Feinberg called the decision on the program’s fate “a controversial issue” and indicated that the final decision will rest with the defense secretary and the president. Major General Joseph Kunkel, the architect of the Air Force’s future force design, has said that “the fight looks much better when [Next Generation Air Dominance] is in it.”
Autonomy and Combat Aviation
In the past decade, multiple programs run by defense research and development organizations have developed artificial intelligence for tactical aviation. When Frank Kendall announced that a crewed sixth-generation fighter jet would be paired with lower-cost, autonomous aircraft, the Air Force consolidated multiple programs, selecting General Atomics and Anduril Industries to develop prototypes.
Venture capitalists, defense tech startup founders, and Elon Musk have declared crewed fighter aircraft obsolete on today’s battlefields, advocating for the military to fully pivot to autonomous aircraft. These individuals are ignoring that Ukrainian F-16s have been used to great effect against Russia, and in October, Israeli F-35s led a strike package of over 100 aircraft that crippled Iran’s air defenses and ballistic missile factories, losing no aircraft during the mission.
There is a need for more autonomy. Today’s Air Force faces a compelling adversary in the Chinese air force. The U.S. Air Force is the smallest, oldest, and least ready that it has ever been and is suffering a perpetual pilot shortage. It cannot afford to rely solely on crewed aircraft to project power.
In November 2022, the Air Force announced their plan to build Collaborative Combat Aircraft. Unlike remotely piloted platforms like the MQ-1 Predator and MQ-9 Reaper, Collaborative Combat Aircraft would be flown by AI agents, commanded by pilots in crewed fighters during combat, and have many of the same mission systems and performance characteristics of a crewed fighter aircraft. Today, the Air Force is testing AI agents on the X-62A Vista and the XQ-58 Valkyrie, with entire teams of test professionals monitoring flights. Bill Gray, the Air Force Test Pilot School’s chief pilot, has seen AI agents make a number of mistakes during a dozen flights in the X-62.
This current state of AI-enabled autonomy is nowhere close to mature enough to justify the decision to cancel the acquisition of a crewed sixth-generation fighter aircraft. Fighter pilots have three core cognitive functions when flying in a tactical environment: perception, decision, and execution. Perception is the function of observing one’s environment, assessing the disposition of the adversary, and remaining oriented to one’s surroundings, including terrain and other friendly aircraft. Decision is the function of choosing to take a particular action, including employing lethal force by dropping a bomb or firing an air-to-air missile. Execution is the function of following through on one’s decision. For fighter pilots, execution is often the easiest part of flying in a tactical environment. The most difficult form of execution is dogfighting: maneuvering in three-dimensional space to a position to shoot down an enemy fighter aircraft.
The victory of an AI agent against a human fighter pilot in a virtual dogfight in 2020 is not representative of how AI agents perform when dogfighting in the physical world. The agent flying against the human fighter pilot in the virtual dogfight had perfect awareness of its environment, as well as the velocity, angular acceleration, and vectors of the human fighter pilot’s virtual airplane. In the physical world, such awareness is exceptionally rare. The artificial intelligence agents flying the XQ-58 and X-62A in the physical world are just beginning to build toward the functions of decision and execution, to say nothing of solving for the function of perception.
Prototype Collaborative Combat Aircraft are scheduled for their first flights late this year, with operational aircraft scheduled for delivery by the end of the decade. Like every other combat aircraft, they will face countermeasures. Algorithmic countermeasures are already being seen in a few areas, including when an amateur Go player defeated a top-ranked AI algorithm by exploiting weaknesses in the algorithm’s decision-making, marines wearing cardboard boxes on their heads to defeat AI surveillance systems, and activists disabling self-driving cars with parking cones.
The best way to hedge against algorithmic countermeasures is to marry human agency with autonomy. The Air Force should continue to work towards determining the tasks that autonomous aircraft can be relied upon to perform in combat and calibrate trust in the autonomy accordingly.
Autonomy and Self-Driving Cars
The commercial technology sector’s ongoing struggle to field self-driving cars should serve as a cautionary tale for the dangers of overestimating the potential of autonomous combat aircraft.
Current testing for Collaborative Combat Aircraft is comparable to the state of self-driving cars in 2010 when Google first started testing. In 2012, Google began tests on public roads with human supervision. Industry leaders predicted that most ridesharing vehicles would be self-driving by 2021, and car ownership would be all but extinct in major cities by 2025.
Despite easy access to processing power and an extraordinary volume of data, the engineering challenges of self-driving cars have proven to be too great for such widespread adoption. They have caused crashes, traffic jams, and collided with pedestrians. Multiple self-driving car startups backed by major automotive manufacturers, including Argo AI and Cruise, have shut down.
Waymo has successfully developed and deployed self-driving cars that are limited to a defined geographic area, which is a remarkable accomplishment. However, Waymo cars are occasionally crashing into utility poles and randomly doing donuts in parking lots with human passengers inside. Despite the RoboTaxi’s appearance, Tesla is apparently retaining the ability to take control of the vehicle from a remote location in case something goes wrong.
Self-driving cars are an easier engineering problem than autonomous combat aircraft for three reasons. First, engineers building self-driving cars have an immense amount of data from the same environment their algorithms are being built to drive the car in. Engineers building autonomous combat aircraft have a comparatively minuscule amount of real-world data to train their algorithms. The data they do have come from controlled environments, not real air battles. Second, self-driving car companies have much more alacrity when testing their cars. The safety standards for operating military aircraft are incredibly stringent. Crashes of a single military aircraft can ground entire fleets for months. Third, America’s roads are governed by a plethora of traffic laws that allow self-driving cars to assume a great degree of predictability in the behavior of the other cars on the road. Air combat is an unpredictable, unfair, and competitive environment where the laws of physics are the main laws that matter. As such, taking control of an autonomous combat aircraft from a remote location will be made extremely difficult in future air battles by counter-space weapons.
The Digital Century Series
When Will Roper was looking to disrupt the status quo with Next Generation Air Dominance, he identified a need to shift away from a product-centric acquisitions strategy and towards a process-centric acquisitions strategy.
Traditional, product-centric acquisitions strategies first define requirements for a new aircraft based on what the government thinks it will need for decades to come. Second, they prescribe those requirements to industry, allowing for little creativity and original ideas. Third, the government will select one aircraft, buying hundreds of them, incentivizing the contractor to get the flyaway cost as low as possible, and keeping the aircraft in service for a generation or more.
A process-centric acquisitions strategy would identify the operational problem that it needs to solve with a new aircraft, avoiding the self-defeating tendency to ask for too much out of one aircraft. Second, it would describe that problem to industry instead of prescribing requirements, empowering industry to be more creative about how they’d design the aircraft. Third, the government would compete designs against each other, buying aircraft in smaller quantities, soliciting new proposals from industry in a matter of years rather than decades, and retiring aircraft in less than a generation. This would incentivize industry to make its money in developing, producing, and fielding new platforms, eliminating the economic incentive for contractors to keep legacy programs alive.
The one drawback of a process-centric acquisitions strategy would be that the aircraft would have a higher flyaway cost. However, the total cost of the program would be lower since the aircraft would be retired earlier and replaced by a new platform.
A process-centric strategy would be enabled by a “digital trinity” of agile software development, modular open systems architecture, and digital engineering. Roper was inspired to pursue this approach after being exposed to the methods that Formula One teams use to digitally design and test thousands of racecars in virtual simulations before deciding the best design for a particular racetrack, going to “print” on the new design, gathering data on the performance of the car from a plethora of sensors during the race, feeding those data back into their models and simulations, and iterating the next design. By the end of the season, 85 percent of the 15,000 components of a Formula One racecar will have been changed and improved since the beginning of the season. Successful Formula One teams are victorious not because they focus on building a single racecar better than every other team, but because of their pursuit of perfecting a process that allows them to continuously iterate and improve on the design and construction of a racecar.
The Digital Century Series was intended to achieve three outcomes. First, quickly develop, produce, and field new aircraft to remain responsive to the threat environment by leveraging a variety of methods, including manufacturing technology that can 3D print racecars with top speeds of 253 miles an hour. Second, flip the business model by incentivizing industry to build new aircraft rather than maintain old aircraft by buying smaller numbers of new aircraft more frequently and retiring them earlier. Third, attract new entrants to a shrinking industrial base by having a greater frequency of opportunity and opening the market for maintenance and sustainment to third-party vendors.
Adopting best practices from Formula One to build a sixth-generation fighter will present some challenges. First, the right people will need to be on the inside of the Next Generation Air Dominance program. The government can address this by using Highly Qualified Expert appointing authority to hire engineers from Formula One teams to work on the program. Second, Formula One teams don’t have to worry about protecting intellectual property or classified information. The components of a fighter aircraft are made by a variety of companies that often have a stranglehold on their intellectual property. Additionally, there will be highly classified capabilities in a sixth-generation fighter. Creating a decentralized architecture that can share relevant information across organizational siloes whilst respecting intellectual property and protecting classified information is the biggest technical challenge to building the Digital Century Series.
The Nighthawk, the Raven, and the Dangers of Over-Reliance on one Capability
During the First Gulf War, the F-117 Nighthawk achieved an 80 percent mission success rate and suffered no losses, flying more than 1,200 missions. America’s decisive victory in that conflict coincided with the dissolution of the Soviet Union. Facing a peace dividend and armed with a diabolically effective capability in the F-117, the Air Force retired its fleet of EF-111 Ravens, becoming dependent on Navy EA-6 Prowlers for electronic attack.
In 1999, an F-117 was shot down over Belgrade, Serbia by a surface-to-air missile designed in the 1960s. Afterward, an Air Force investigation discovered multiple contributing factors, including the absence of a Prowler escort. Stealth aircraft had proven to be vulnerable, and electronic attack aircraft still had a critical, albeit secondary role to play. The Air Force was unable to reverse course on the decision to divest from electronic attack aircraft, relying on Navy electronic attack aircraft for almost two decades until the F-35 became operational.
The downing of an F-117 in Belgrade in 1999 shows that today’s technological overmatch will lead to tomorrow’s countermeasures. The advent of stealth aircraft in the 1980s did not spell the end of the need for electronic attack aircraft, nor does the advent of AI-enabled autonomy spell the end of the need for crewed fighter aircraft.
Looking Ahead
It is tempting to view the future of combat aviation as a binary choice between crewed aircraft and autonomous aircraft. The AI agents flying experimental aircraft over American Air Force bases are not nearly proven enough to warrant the end of crewed fighter aircraft. Those who advocate for divesting from crewed fighter aircraft are not basing their argument on factual evidence of the utility and superiority of autonomous combat aircraft.
The last several years of ponderous deliberation over a crewed sixth-generation fighter has left the U.S. Air Force to face the real possibility that it will see China field an operational crewed sixth-generation fighter before it fields its own. The decision to abandon the Digital Century Series and pursue a product-centric acquisition strategy for a crewed sixth-generation fighter has proven to be a serious mistake that has cost the United States precious time as it prepares for a potential conflict with China. Fortunately, there is a chance to adjust course.
The Digital Century Series should be resurrected by the Air Force. It is the only way to expeditiously field a crewed sixth-generation fighter and break the cycle of forcing a dwindling number of contractors to lose more money on building new aircraft and make more money on sustaining old aircraft. These contractors are not the first to blame for this trend. They do not design the programs that they sell their aircraft to. The U.S. government is the only customer for new fighter jets in America. If it decides to do things differently, industry will have to respond in kind.
If the decision to resurrect the Digital Century Series comes down to cost, then American leaders should acknowledge the fact that the Air Force has been underfunded for decades, and today’s security environment demands that the Air Force receive the generational investment it requires to prepare for and deter future conflicts.
Elected and uniformed leaders in the United States face a consequential choice: rediscover the same ingenuity and courage that built the world’s fastest crewed aircraft in four years and the first stealth combat aircraft in five years, or continue to cower behind the same broken central planning system that has suffocated the defense industrial base and prevented the full might of American free enterprise from being unleashed on the military’s most important missions.
The former may result in some occasional missteps and setbacks, but the latter runs a serious risk of ceding American predominance in developing and fielding tactical aircraft to China. Such an outcome is more than possible. It has already happened in shipbuilding.
Ben McNally is a consultant in the defense technology industry. He has worked at the Defense Innovation Unit, helped startups and small businesses win millions of dollars in government contracts, and led events, programming, and outreach for the Center for Defense Innovation at the Capital Factory in Austin, Texas. He is a graduate of the University of Texas at Austin, where he was a fellow at the Clements Center for National Security. The opinions expressed here have no connection to his consultancy work.
Image: Midjourney
Jesus Christ is King
