Mass force has always been a defining principle of warfare. But what happens when mass is no longer limited by the number of available pilots? Advances in autonomy are allowing militaries to confront challenges like pilot shortages and increasingly contested environments. Swarms of intelligent, collaborative systems are moving from concept to reality, promising new levels of operational scale and greater safety for the warfighters who carry out those missions in the most dangerous environments.

At the center of this transformation is Brandon Tseng, co-founder and president of Shield AI. Drawing on his experience as a Navy SEAL, Tseng launched the company in 2015 with a singular mission: protect service members and civilians with AI-powered autonomy. Today, Shield AI is pioneering technology that can fly and fight in denied environments, scale from single aircraft to swarms, and—through its partnership with Booz Allen—help the U.S. field the intelligent mass needed to compete with peer adversaries.

In this conversation with Dr. Randy Yamada, Ph.D., Booz Allen’s vice president for autonomy, Tseng discusses the origins of Shield AI, the bold decisions that shaped its growth, and the future of AI-powered warfighting.

Since our founding in 2015, that has been our mission, and it comes from my time in the U.S. Navy SEAL teams. On deployments from Afghanistan to the Pacific to the Persian Gulf, I saw firsthand the need for this capability. At Shield AI, we are building the world's best AI pilot. The easiest way to think about an AI pilot is self-driving technology. In a military context, it enables unmanned systems to operate without GPS, without communications, and without a remote pilot. It also unlocks the concept of “teaming” or “swarming,” which, with mass being a fundamental concept of warfare, introduces the warfighting concept of unlimited mass. 

The first problem I wanted to solve was close-quarters combat. As a Navy SEAL, that’s one of the most dangerous missions we conducted: going room to room inside buildings. It looks exciting on TV, but in reality, there’s nothing more terrifying than being in a gunfight in such tight spaces. I began talking to people about putting self-driving technology onto a quadcopter to help with those missions.

One of those conversations was with a Naval Academy classmate of mine—a mechanical engineer and F-18 pilot. He told me that the problem I was trying to solve with quadcopters indoors was the same one they were facing in the air warfare community. When you fly a drone into a building, you lose GPS and communications. At that time, during operations in Syria, our forces were beginning to experience the same thing on a larger scale: jamming that disabled GPS and communications, grounding Predators and Reapers.

He explained that fighter pilots were now facing another threat: proliferated, integrated air defense systems. Against those systems, survivability rates were extremely low. You don’t send pilots into those areas because they simply won’t come back. He told me the odds of survival were similar to stepping on an IED. 

I’d break it into two parts. First, get ready for the pain train. Building a company is full of suffering, sleepless nights, and stress in every direction. I often tell veterans who start companies that one advantage they bring is knowing how to endure hardship and still accomplish the mission. That mindset helps.

Second, you have to be bold. Every entrepreneur will face moments when the right decision feels uncomfortable or even terrifying. At Shield AI, we’ve had to make several of those “bet the house” calls, and they’ve been critical to our growth. The reality is, if you’re not making bold decisions and taking big risks, you’re not really growing.

Two things set V-BAT apart. First is the hardware architecture. Its design creates a uniquely small logistics footprint and makes it easy to operate and transport. Features like the ducted fan give it advantages such as a payload-to-weight ratio roughly twice that of other vertical takeoff and landing aircraft. In practice, that means V-BAT can carry the same payload while being half the size and weight, making it a much more agile and versatile platform.

The second differentiator is software. Our autonomy is what truly makes V-BAT stand out. We’ve repeatedly demonstrated operations in GPS- and communications-denied environments, proving mission autonomy on the platform. We’ve also shown multi-agent mission autonomy, and we continue to expand those capabilities. Increasingly, software is becoming the defining factor for unmanned systems. Hardware innovation is valuable, but it’s hard to sustain differentiation there. Autonomy and mission software create a far greater competitive edge.

A good comparison is Tesla: The company disrupted hardware by committing to electric vehicles, but its real long-term advantage is in software. If you can differentiate on both hardware and software, the effect is powerful. That’s how we think about V-BAT today.