Is Elon Musk right saying that fighter jets have become obsolete?

Elon Musk recently stirred up a frenzy in the airpower community with a handful of posts about the future of military aviation, claiming that crewed fighter jets are just an “inefficient way to extend the range” of missiles and bombs, and going on to say that it’s “laughably easy” to take down fighter jets, including stealth ones, citing the war in Ukraine as proof of his position. 

Elon Musk shared a video of a coordinated drone swarm with the caption, “Meanwhile, some idiots are still building manned fighters like the F-35.” This drew a great deal of attention as small drones like the ones shown in the video are utterly incapable of performing the sorts of missions the F-35 was designed to fly. Musk soon addressed these critics in a follow-up post:

This isn’t the first time Musk has come out against the idea of crewed fighters. Back in 2020, there was a similar conversation about his claim that the “era of fighters” was over and that remote-controlled fighters should be fielded to replace the F-35 which Sandboxx News addressed.

While Musk’s take on fighter aircraft is pretty divorced from the reality of today’s battlefield, that isn’t to say that there’s no truth at all to what he has to say. So, let’s go over Musk’s position piece by piece, and assess each in terms of realism and value in modern conflict. 

Crewed fighter jets are still essential for combat operations

The argument that crewed fighter jets aren’t essential for combat operations was first made in the Cold War by people who believed the U.S. should get rid of its strategic bomber fleets entirely once newly fielded ICBMs could deliver nuclear weapons to targets without any risk to a bomber crew. 

At the time, ICBM technology was not considered mature enough to be relied on alone without any other means of delivering nuclear weapons – similar to the way artificial intelligence and autonomous drone system technology are not mature enough today to step in and replace crewed fighters en masse. 

Air Force officers Lt. Col. Shane Praisewater and Maj, Matthew Guertin recently echoed these sentiments, referring to modern multi-role fighters as “stand-in” forces capable of operating inside contested airspace, versus drones and weapon systems that can be launched from stand-off distances:

“Drones and other novel capabilities add to the complexity of modern warfare, but they are not a replacement for the capabilities enjoyed by a stand-in force ensuring even limited control of the air over a battlefield. Artificial intelligence (AI) driven unmanned platforms are an intriguing option, but they have no history of battlefield success and are likely unavailable near-term. The USAF has attempted to start the transition to more unmanned platforms before, but seeking to define affordable mass in terms of lower risk is an inherent admission that stand-in benefits are still necessary. If the technology is not immediately available, the USAF cannot afford to gamble its entire future on it,” they wrote.

Today’s most advanced AI systems for combat aircraft, being developed and tested by the Air Force Research Lab as part of the ongoing Skyborg program, still rely on human intervention in several ways, including having a human in the cockpit to switch on specialized AI agents capable of performing different tasks. This is because, to date, no single AI agent is capable of managing everything a fighter aircraft may be tasked with doing. Specific agents can perform specific tasks, like air-to-air combat, but can’t perform others, like air-to-ground operations or even landing.

Further, until these systems are robust and trustworthy enough to operate entirely on their own, you have to manage concerns about jamming and other means of electronic warfare that could interfere with the function of autonomous systems in a combat zone. 

“Elon’s hot take about the F-35 being replaced by drones is because he understands technology, but he doesn’t understand tactics,” Sgt. 1st Class Ethan Long, a Patriot air defense system operator and instructor who’s worked in development testing for new DoD systems, told Sandboxx News.

“Every single system the United States employs can function without GPS signal because GPS jamming has become so common. We’re seeing this on the battlefields of Ukraine – it’s why Russia is using FPV drones that are guided by wire,” Long added. 

Even if such systems are going to operate untethered from human operators, then the American public needs to come to terms with the ethical implications of outsourcing the “kill decision” to a machine – something that is currently against DoD regulations. 

Related: What actually happens when fighter pilots take off their masks?

Drone swarms are terrifying, but they’re also limited in capability

A swarm of quadcopters, like the ones in Musk’s, video may be able to overwhelm an enemy air defense array but small drones of that sort are specifically designed to operate at low altitudes and short ranges (allowing for their compact form). So, in the best conditions, they need to be launched from within 20-30 miles of your target, which would mean getting a launch platform into that sort of vicinity. That could be done via aircraft, but at that point, you’re better off using a weapon designed to engage air defenses, like the AARGM-ER, which could wipe out the same array from more than a hundred miles out without the need to build a few hundred or even thousand replacement quadcopters for every mission.  

So let’s assume we’re talking about using larger drones that could offer similar capabilities to today’s fighter aircraft.  

Designing an aircraft around autonomous technologies that are rapidly maturing but remain in their relative infancy today would represent an even bigger financial boondoggle than the F-35’s acquisition. The more cutting-edge and experimental a technology, the more risk-laden its path from development to production to operation is. Because many of these technologies were not fully mature at the time, the F-35 program had to go back and make changes to the aircraft to accommodate issues and limitations as they arose, increasing costs and stretching out timelines. 

Developing a new fighter today comes with those same risks, and incorporating as-yet unproven artificial intelligence into a fighter that’s already designed to bring a laundry list of other emerging technologies to bear would only increase its complexity and cost and extend the development timeline. 

Yet, to be fair to Musk, the technology will eventually get there but even then, there’s no guarantee that these drones will be cheaper to operate in the near term. The reality of global military operations means countless variables need to be considered alongside platform capability to determine what’s the most efficient use of funds.  

Pulling the pilot out of the cockpit doesn’t automatically result in cost savings

The RQ-4 Global Hawk is an excellent example of how the transition to uncrewed aircraft doesn’t necessarily come with a reduction in cost. This high-flying remotely operated surveillance aircraft entered service more than two decades ago and has long been touted as the Air Force’s eventual replacement for the U-2 Spy Plane. Yet, nearly 70 years after the Dragon Lady first started flying, it remains in service because it’s often still cheaper to stuff a pilot in a space suit and send them up in the U-2 than it is to send a remotely operated RQ-4 to do the same job. 

That cost ratio has improved in the RQ-4’s favor over the years and the U-2 is now slated to retire in 2026, but that’s still notably more than a quarter century into Global Hawk operations.

Notably, it takes a great deal of time and money to integrate new technologies and platforms into America’s sprawling defense apparatus and warfighting strategy, and only once that transition is completed can new technologies begin to offer anything close to cost savings. The further the new technology or method of operations departs from existing infrastructure, training, procedure, or doctrine, the longer and more expensive that transition will be. 

That’s not an argument against making these technological changes, but it is nonetheless a reality you must account for as a military planner. You can’t simply throw old systems, platforms, or practices away and start over with new ones. You need to keep your existing systems online as new ones are fielded and their technology matures. That means having the infrastructure to support both legacy systems and emerging ones at the time time, including all the personnel required to build, operate, and maintain both. 

While we’re only talking about dozens of aircraft in the case of the U-2 and RQ-4, we’re talking about thousands of aircraft when it comes to fighters. The transition toward uncrewed fighters has already begun, with contracts already awarded for continued development on several collaborative combat aircraft, or CCAs, to fly fighter missions alongside crewed aircraft, but even some of these groundbreaking designs are being fielded with cockpits so developers can focus on the complexities of fielding new fighter drones separate from the challenges of fielding AI control systems robust enough for the job, let alone capable of managing the testing regime it takes to get there. 

At this stage, saying it would be more “efficient” to transition to uncrewed fighters would only be true if the U.S. could immediately divest its existing fleets of tactical aircraft and focus all of its time and resources on developing new AI-piloted jets, with the assurance that there would be no threats America might need fighters to address in the meantime. That, of course, isn’t realistic, and as such, the only path forward is to continue maturing these technologies toward service while simultaneously maintaining a high level of capability delivered through legacy platforms. Musk isn’t wrong that drones will eventually be cheaper to build, buy, and operate, they just aren’t today and won’t be for years to come, but that is the general direction we’re headed. 

Fighter pilots do a lot more than just fly to a specific point and drop bombs

Beyond the question of fiscal efficiency, we also need to address the underlying assumption Musk’s post makes about the role pilots play in combat when he says, “Crewed fighter jets are an inefficient way to extend the range of missiles or drop bombs. A reusable drone can do so without all the overhead of a human pilot.” 

In the right context, Musk is right that drones can be a cheaper way to deliver ordnance to a target than crewed fighters, and we can look to data collected over two decades of asymmetric conflict in the Middle East to substantiate his claim. But where this claim fails to embrace the nuances of reality is in the assertion that delivering ordnance to a set point in space is all that fighter pilots do. 

As F-35 pilot Hasard Lee wrote in his book The Art of Clear Thinking, “boiled down, a fighter pilot’s job is to make decisions – thousands of them each flight, often with incomplete information and lives on the line.” He later goes on to explain that, “the flight must be executed under the fog and friction of war, where no matter how well a mission is planned, it will change. This means that despite the immense effort that is put into planning a mission, there will always be difficult decisions that need to be made in the air that haven’t been anticipated or that don’t have textbook answers.” 

This speaks to the problem artificial intelligence has with absorbing the fighter pilot role. AI is exceedingly good at completing straightforward tasks it’s been trained to manage, but faces significant limitations when asked to make decisions outside of its prescribed training data set. 

Related: How fighter pilots plan combat missions

AI is far more limited than tech-bros want you to believe

On October 7, 2024, a team of researchers from Apple published a paper called GSM-Symbolic: Understanding the Limitations of Mathematical Reasoning in Large Language Models. The paper highlighted the limitations of current artificial intelligence and went on to posit that these limitations are inherent to these systems themselves. 

The team from Apple studied their own AI systems and those of top-tier competitors like OpenAI and Meta, and concluded that none of them are actually capable of performing genuine logical reasoning, and can only seemingly demonstrate reasoning by repeating steps they’ve previously been trained on. But more specific to our discussion, they also concluded that the addition of “seemingly relevant but ultimately inconsequential information” to questions posed resulted in “substantial performance drops of up to 65% across all state-of-the-art models.” 

“The high variance in LLM performance on different versions of the same question, their substantial drop in performance with a minor increase in difficulty, and their sensitivity to inconsequential information indicate that their reasoning is fragile,” the paper concludes. “It may resemble sophisticated pattern matching more than true logical reasoning […]”

Being a fighter pilot requires making split-second decisions repeatedly sometimes without complete information, and always with a glut of related but arguably inconsequential data pouring in from internal and external sensors on the aircraft. Effectively managing this firehose of information, choosing where to place one’s focus, and quickly coming up with novel solutions to problems you haven’t seen before is the very foundation of fighter pilot training – but this sort of creative problem-solving is outside the realm of what AI can reliably perform. That’s not to say AI will never have the right answer, but when it comes to deciding where to launch a missile, drop a bomb, or evade a threat, getting it wrong once has significant implications. 

When it comes to managing the complexity of a modern battlespace, these systems will need to rely on human operators to give specific directives for a long time to come. Having local human operators provide those directives without concerns about latency means having crewed aircraft in the battlespace, flying alongside their AI-controlled wingmen. And while Musk is right that drones are a big part of the future of airpower, he’s objectively wrong in assuming humans won’t be. 

As Long, who also goes by “Habitual Linecrosser” on YouTube and Tiktok explained to Sandboxx News, the U.S. military doesn’t just buy systems or platforms from “a buddy,” instead, capabilities are developed through an intensive developmental testing process meant to meet specific criteria and mission requirements.

Is it actually easy to shoot fighter jets down?

Likewise, Musk’s claim that fighter jets will be “shot down very quickly” by adversaries with advanced air defenses is a sort of half-truth that depends entirely on circumstance. Musk suggests that Ukraine has shown how vulnerable aircraft are to modern air defense systems. Yet, the conflict in Ukraine has largely been between two forces using similarly dated Soviet hardware, and despite Russia’s use of more advanced and modern systems like the S-400 Triumpf, Ukraine still has dated but operational MiG-29s, Su-27s, and Su-24s flying combat operations regularly. 

Air defense is complex and nuanced, with even the best systems only capable of providing coverage for a set amount of area; only able to engage a set number of threats; and always operating under the impending threat of airstrike. If air defense were as simple an enterprise as Musk suggests, Russia could have secured air superiority over the entirety of Ukraine, which is only about the size of Texas and rests right on the Russian border, at some point in the last three years of fighting. 

No air defense system is infallible and even if there was such a system, no nation on earth could afford to build so many of them that they could blanket all of their airspace with it.

At the end of the day, defense is the sort of thing everyone believes they could do better as long as they can trade in general statements and ignore the granular details of logistics and warfare. But once you find yourself knee-deep in the complexities of combat operations at scale, many of those seemingly insightful generalizations begin to sound almost silly.

However, Elon Musk wrong is not entirely wrong when he says that drones are the future of air warfare. They will certainly play a large and growing role in conflict in the decades ahead, but will complement not replace humans. In some simple mission sets drones will be able to replace humans, but not in the more complex ones, at least, not any time soon. 

“If the technology was a little further along, I wouldn’t necessarily disagree with Elon Musk. But right now? We’re not quite there. So, as for now, we will always need an operator on the battlefield in that aircraft, tank, APC, wherever. Let technology progress a little bit, and we’ll see where we’re at,” Long told Sandboxx News. 

Large-scale war is one of the most complex endeavors humanity has devised and it grows more complicated every day. Scholars, engineers, and warfighters dedicate their lives to studying tactics and doctrine, developing new technologies or capabilities, and finding creative ways to win when even the odds seem stacked against them. Elon Musk can play a role in that discourse, but his shower-thought tweets aren’t going to get that job done, and shouldn’t be seen as anything more than what they are: the opinions of a bright guy with no actual experience in the field. 

For ideas to be good, they need to be actionable. Everything else is just a conversation. 

Read more from Sandboxx News

• Marines deploy new system to take out ships in the Pacific
• How does China’s new J-35 stealth fighter compare to America’s F-35?
• Who dares wins: The importance of defeat in being successful
• How the military bounced back from the recruiting crisis
• These were 5 outrageous uses of nuclear power during the Cold War