We Saw A New AI-Piloted Fighter Drone About To Transform Warfare

Military Unmanned Aerial Vehicles (UAVs)
9 followers
1 Video View·Nov 26, 2025
Multiple startups all the way to defense giants like Lockheed Martin are betting on a new
crop of drones to be the future of war.
I'm here at Shield AI getting an exclusive first look at the XBAT.
It's an autonomous fighter jet, meaning there's no pilot.
It's entirely powered by an AI system called HiveMind.
XBAT is a next-generation warfighter.
It has capabilities to fly long distance, carry munitions, be fully autonomous,
and be anywhere that defense needs it.
Drone combat has shaped the war in Ukraine, shedding a light on the shifting landscape of the modern battlefield.
And Shield AI is one of the many companies vying to lead the change.
There is going to be a world filled of autonomous systems.
Self-driving cars, humanoid robots, self-driving aircraft like what Shield AI has been doing
CNBC visited Frisco, Texas to see how Shield AI is working to shake up the defense industry.
Shield AI was founded in 2015 and has built its business on drone development.
First with this quadcopter and eventually with this drone called the VBAT,
an intelligence and surveillance drone with VTOL or vertical takeoff and landing.
Shield secured a nearly $200 million contract with the U.S. Coast Guard for the VBAT.
So the VBAT has been successfully deployed for a while now.
But the next thing that you guys are doing is deploying the XBAT.
Yeah.
The model of it right here, right? Tell me about that.
This is an XBAT.
XBAT is the first airplane in the world that is AI piloted and vertical takeoff launch and land.
We, Shield AI, have flown the F-16 autonomously, but those two things,
AI piloted and vertical takeoff launch and land,
have never come together in the form of a next-generation aircraft.
The XBAT is a fighter jet that can be equipped with missiles and used for combat.
The company says that one of its driving missions
is to save the lives of service members with pilot-free aircrafts.
What is the timeline that you expect these to actually be deployed into battlefields?
So first flight of the XBAT, we're doing a subsystem flights in 2026.
We're doing a full system flight in 2027.
We've been doing engine testing already, radar cross-section testing already,
wind tunnel testing already, but going to production in 2029.
We got to visit the wind tunnel testing site in San Diego, California.
This is where Shield is scaling down models of its aircrafts
to test in a tightly controlled environment.
Simulations are only so good, there's still no substitute for testing.
So we have to come to the wind tunnel.
So this is 17% scale, smaller than the full-scale aircraft to get it inside here.
But once we settle on the scale and build the model,
we can blow air over the model and measure that lift, the drag,
and see if it matches our prediction.
This is key as it allows adjustments to be made before building the real thing.
If you flight test first, you've gone too far into the process
and it's very costly to start making changes at that point.
So we do as much testing with models as we can.
These are expensive, but they're still much cheaper than building a full working aircraft.
Still, as with many startups, the company says it's investing a lot of money in development
and not yet profitable.
Even though it's generating a lot of revenue.
Today we're generating hundreds of millions of dollars of revenue
and we anticipate doubling our revenue scale this year
and we see a strong growth path in the future as well.
In June 2025, President Trump issued an executive order called
Unleashing American Drone Dominance,
which aims to accelerate commercialization of drone technologies
and integrate them into the national airspace system.
Although no direct dollar amount was attached to that order,
the big beautiful bill has allocated billions of dollars
for unmanned aerial systems and AI development.
Drones are cheap and they're everywhere, but it's more than just cost.
They inflict approximately 60 to 70 percent of damaged weapons on the adversary side
and they inflict up to 80 percent of injuries for the adversary side as well.
The USA is going to field millions of drones.
Key to fielding millions of drones is AI and autonomy.
I can't field millions of drone pilots.
So the company said it ran the numbers and the expat will cost them about $27 million to make,
which I know that sounds like a lot, but it's actually a fraction of the cost
of what the military has previously been spending on fighter jets,
especially when you factor in the costs of training pilots,
which this aircraft does not have because it's run entirely by AI.
That's important because the U.S. Air Force is already facing a pilot shortage.
It's a highly skilled job that requires significant investment from the government.
A report from RAND Corporation estimates that training a basic qualified pilot for an F-35,
which is a widely used modern military fighter jet, can cost over $10 million.
That's in addition to the cost of the aircraft itself, which,
depending on the type, costs in the range of $80 to $100 million.
Lockheed Martin finalized a contract at the end of September 2025
to deliver another nearly 300 F-35s to both the U.S. military and other international customers.
Is the goal ultimately to replace the F-35?
I don't want to say it's replacing fighter jets or fighter pilots anytime soon,
but the way that I think about this aircraft is our aim is for it to be this generation's F-16.
F-16 is the most widely proliferated fighter jet on the planet,
but there are a lot of companies hoping to do the same thing.
S.H.I.E.L.D. is relatively small compared to competitors like General Atomics and Anduril,
who were selected by the U.S. Air Force in 2024
to develop a fleet of drones meant to fly alongside manned fighter jets,
beating out major players like Boeing and Lockheed Martin,
who had hoped to secure the development funding.
I think it's a new challenge because the concept of crewed-uncrewed systems is great in theory,
but we're yet to see how this is practically actually implemented
and how, most importantly, it is showing in the battlefield.
It's a crowded space. How do you guys set yourself apart?
There are definitely a lot of drones, and where we have focus is leveraging
AI capabilities to ensure that we deliver great mission outcomes for our customers.
And in the AI context, it's being able to operate in contested environments, meaning no GPS.
You have to use software and intelligence to be able to
deliver targets and have insights about what you're focused on.
SHIELD.AI has long been focused on building drones,
but now it's hinging a lot of its future on the AI software that's used to power them,
like the HiveMind in the XBAT.
The software is a cornerstone and foundation for everything we do.
It will ultimately be the long-term growth driver of this business because
it enables the development of this next generation aircraft.
We have to empower the defense industrial base
with the exact same development tools, infrastructure, and pipelines
that SHIELD.AI has used to make AI autonomy.
So we work directly with the major defense prime contractors of the world.
This is where we manufacture our autonomous systems, but most importantly,
this is also where we do our engineering to bring the best of autonomy
and marry that with world-class aircraft design.
Do you feel AI is at the point where it can be reliably
making decisions autonomously in a war zone?
Definitely, but we always assume there's a human in the loop somewhere.
And we're seeing the impact and positive aspects of that because HiveMind is
deployed today, for example, in the Ukraine,
helping us deliver great outcomes for the Ukrainians.
So we've got a lot of battle-tested experience
that gives us confidence in the capabilities that we have.
SHIELD says its mission is to deter war,
or as Tseng calls it, peace through strength,
by enabling countries to keep their adversaries in check.
But AI is advancing quickly and creating increasing concerns
about everything from job replacement to security threats.
And the prospect of AI-powered weapons systems doesn't come without risk.
What do you say to people that are scared about the prospects?
What AI could lead to if it fell into the wrong hands
or if it was used for bad intentions?
Yeah, I tell people as a former Navy SEAL that has
had to make the moral decision about the use of lethal force on the battlefield.
I don't believe that autonomous systems should be
making any moral decisions about the use of lethal force.
SHIELD AI does not believe that. That's SHIELD AI policy.
That is U.S. military policy. That is NATO policy.
And so I am less concerned about this feature of autonomous killer robots.
I think it gets overblown by Hollywood.
Drones have been used in war zones since as early as World War I,
but their importance has grown immeasurably since then.
70% of recent conflicts used drones.
And just for a second, in 2010, only three countries possessed armed drones.
In 2025, these numbers increased to 118 countries and they continuously grow.
So definitely what we see from the war in Ukraine or Middle East,
they are tactically, operationally and strategically absolutely important weapons.
And they have become central, not peripheral.
The war in Ukraine has shed light on the explosion of drone deployments in modern warfare
and the various special functions that they can perform.
However, who actually wins from this drone arm race is China.
Because both Ukraine and Russia are using Chinese components still to some extent.
We have seen a lack of those systems from the U.S.
Particularly, we have not really seen the presence of many American companies in the real battlefield.
One of the defense tactics against drones, which is utilized by both Ukraine and Russia,
is GPS jamming signals that render many autonomous drones ineffective.
That is a problem that Shield is solving with AI.
What we have done in Ukraine with our VBAT is,
we've done hundreds of operations now where GPS and communications is jammed.
That is a singular point of success where for the first time since that war started,
they've had the ability to conduct long-range reconnaissance, intelligence,
surveillance and targeting operations while GPS is jammed.
A big focus for you guys is the vertical takeoff and landings.
Why is that so important?
The number one benefit of being able to take off vertically
is that you are no longer constrained by a runway.
They are massive, stationary, expensive infrastructure targets for the enemy.
Usually, you pay a price of either being able to take off vertically and land vertically,
or being able to have range and being able to carry useful payloads.
What we're trying to do is break that curve a little bit
to be able to carry useful payloads for long ranges and being independent of a runway.
So that's the nut that's tough to crack.
It's pretty small. This could take off or land really anywhere.
Oh yeah, we pack this thing in the back of a truck.
We're launching off small ships all the time.
But these advancements at Shield have had some bumps along the way,
notably in 2024 when a U.S. surface member's fingers were partially severed
during a drone landing accident.
Forbes reported that the company had been overlooking safety precautions for years.
What has changed since then?
We've been very much focused on safety and building safety into the culture of the company.
And this is something we take incredibly seriously.
Did you guys lose contracts over that injury, though?
I read some reports that it threw profitability targets off.
Through that process, there was some loss of confidence from customers.
But I think we've done a phenomenal job of recovering from that and rebuilding momentum.
And today, as we sit here, we're very confident in our ability to deliver great products that are safe.
Shield AI is a CNBC Disruptor 50 company.
It's our list of the most innovative private companies that are shaping this new generation of AI.
Is the goal to one day take Shield AI public?
We're really proud to be part of the Disruptor 50 list.
And we think about the opportunity about being public.
We're not in a rush to do that,
but we think that that is something that could be highly valuable for our shareholders.
And we're going to think about the timing of that and when we'd want to do that.
But it's definitely a goal over the long haul for the company.
