Training Example: The Blog of Palmer Luckey – Review the Data, Give Your Score & Compare to the Real AI Evaluation

This seems familiar, somehow.
As of today, Anduril Industries is taking the reins of the largest project of its kind in history: the United States Army’s Integrated Visual Augmentation System (IVAS) program.
For me, this announcement is deeply personal. Since my pre-Oculus days as a teenager who had the opportunity to do a tiny bit of work on the Army’s BRAVEMIND project, I’ve believed there would be a headset on every soldier long before there is a headset on every civilian.  Given that America loses more troops in training than combat, the Squad Immersive Virtual Trainer (SiVT) side of IVAS alone has the potential to save more lives than practically anything else we can imagine building.
Tactical heads-up-displays that turn warfighters into technomancers and pair us with weaponized robotics were one of the products in the original Anduril pitch deck for a reason.  The past eight years we have spent building Lattice have put Anduril in a position to make this type of thing actually useful in the way military strategists and technologists have long dreamed of, ever since Robert Heinlein’s 1959 novel Starship Troopers.  Not just day and night and thermal and ultraviolet, but peering into an idealized interactive real-time composite of past, present, and future that will quickly surpass traditional senses like vision and touch.  Put another way, Superman doesn’t use menus – he just sees and does.  If Anduril had been more than a dozen people when IVAS was first getting spun up all those years ago (at least the Tragic Heap guys didn’t win, our country really dodged a bullet there), I do believe our crazy pitch could have won this from the start – as things stand, though, there is no time like the present.
The IVAS program – one of the most important programs to the Army – represents just the beginnings of a new path in human augmentation, one that will allow America’s warfighters to surpass the limitations of human form and cognition, seamlessly teaming enhanced humans with large packs of robotic and biologic teammates.
We will REDACTED __________________________________________
_____________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________
_________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________________
____________________________________________________________________________________________speculative augmentation efforts I have been leading in the shadows over the past few years via our longstanding but never-disclosed Warfighter Systems business line.
I have to admit that I am almost equally chuffed to finally get a Microsoft Partner campus badge – I’ve always been a Microsoft fanboy after all, to the point of hosting a midnight launch party for the launch of Windows 7.  Yes, all of us had already been running the Release Candidate for years at that point, but there is something special about going gold, you know?  Microsoft has some pretty extraordinary things in the pipe on the AI and cloud computing side that will continue to power IVAS, and as the one person in the universe who bought a Surface Duo phone and then a Surface Duo 2, I am looking forward to letting them do what they do best so I can do what I do best.  I am, after all, the best head-mounted-display designer in the world.
This move has been so many years in the making, over a decade of hacking and scheming and dreaming and building with exactly this specific outcome clearly visualized in my mind’s eye.  I can hardly believe I managed to pull it off.  Everything I’ve done in my career — building Oculus out of a camper trailer, shipping VR to millions of consumers, getting run out of Silicon Valley by backstabbing snakes, betting that Anduril could tear people out of the bigtech megacorp matrix and put them to work on our nation’s most important problems  — has led to this moment. IVAS isn’t just another product, it is a once-in-a-generation opportunity to redefine how technology supports those who serve.  We have a shot to prove that this long-standing dream is no windmill, that this can expand far beyond one company or one headset and act as a a nexus for the best of the best to set a new standard for how a large collection of companies can work together to solve our nation’s most important problems.
Whatever you are imagining, however crazy you imagine I am, multiply it by ten and then do it again.  I am back, and I am only getting started.

Today is November 6th, 2022, the day of the SAO Incident.  Thousands of VRMMORPG gamers were trapped by a mad scientist inside a death game that could only be escaped through completion.  If their hit points dropped to zero, their brain would be bombarded by extraordinarily powerful microwaves, supposedly killing the user.  The same would happen if anyone in the real world tampered with their NerveGear, the virtual reality head-mounted-display that transported their minds and souls to Aincrad, the primary setting of Sword Art Online.
This type of scenario has been a staple of science-fiction for decades, but Sword Art Online exploded in popularity at exactly the right time to have a massive impact on the real world.  The already-popular web novel was adapted into an anime series that aired just as the Oculus Rift went from internet forum curiosity for turbonerds to siren of the games industry, endorsed by giants like John Carmack, Gabe Newell, and Cliff Bleszinski – the 4th episode of SAO, “The Black Swordsman” (黒の剣士) aired in Japan just as the Kickstarter for the Oculus Rift DK1 launched!
The popularity of SAO led to massive otaku enthusiasm for Oculus, especially in Japan, which quickly became our 2nd largest market.  In turn, the existence of the Rift made SAO itself seem far more plausible and grounded – a story that had been written in a world where VR was a dead technology was now straight out of the gamer hype headlines.  This synergy had meaningful impact on our dev kit sales and adoption – literally thousands of people reached out to me asking variations of “Have you seen Sword Art Online?  When will you make the NerveGear real?!”.  The Oculus SDK was the closest thing to SAO’s World Seed, a software tool that enabled anyone to create incredible VR worlds.
There are dozens of stories I could tell, many of them worth a post of their own.  How we ended up with a Kirito mural on the wall of the first Oculus office, how it was eventually moved to Facebook HQ, why it was removed shortly after I was fired.  Personally showing Oculus DK2 to Reki Kawahara at Anime Expo, my wife meeting him while cosplaying as Sinon at the next Anime Expo, getting into minutiae of SAO canon at the Hollywood premiere of SAO: Ordinal Scale, the audience screaming as Yuuki came back from the dead to perform a two-user Mother’s Rosario attack combo with Asuna against the 100th floor boss of Aincrad.  Arguing with people who only watched the anime about the supposed plot and tech holes, explaining that the anime was only ever meant to act as a companion piece for fans of the novel series that so frequently predicts the future of AR, VR, AI, and more.  The beautiful wedding gift from Kawahara-san that shows he is a double-threat writer/artist combo!
But that isn’t what you are here for.  You want NerveGear, the incredible device that perfectly recreates reality using a direct neural interface that is also capable of killing the user.  The idea of tying your real life to your virtual avatar has always fascinated me – you instantly raise the stakes to the maximum level and force people to fundamentally rethink how they interact with the virtual world and the players inside it.  Pumped up graphics might make a game look more real, but only the threat of serious consequences can make a game feel real to you and every other person in the game.  This is an area of videogame mechanics that has never been explored, despite the long history of real-world sports revolving around similar stakes.
The good news is that we are halfway to making a true NerveGear  The bad news is that so far, I have only figured out the half that kills you.  The perfect-VR half of the equation is still many years out.
In SAO, the NerveGear contained a microwave emitter that could be overdriven to lethal levels, something the creator of SAO and the NerveGear itself (Akihiko Kayaba) was able to hide from his employees, regulators, and contract manufacturing partners.  I am a pretty smart guy, but I couldn’t come up with any way to make anything like this work, not without attaching the headset to gigantic pieces of equipment.  In lieu of this, I used three of the explosive charge modules I usually use for a different project, tying them to a narrow-band photosensor that can detect when the screen flashes red at a specific frequency, making game-over integration on the part of the developer very easy.  When an appropriate game-over screen is displayed, the charges fire, instantly destroying the brain of the user.
This isn’t a perfect system, of course.  I have plans for an anti-tamper mechanism that, like the NerveGear, will make it impossible to remove or destroy the headset.  Even so, there are a huge variety of failures that could occur and kill the user at the wrong time.  This is why I have not worked up the balls to actually use it myself, and also why I am convinced that, like in SAO, the final triggering should really be tied to a high-intelligence agent that can readily determine if conditions for termination are actually correct.
At this point, it is just a piece of office art, a thought-provoking reminder of unexplored avenues in game design.  It is also, as far as I know, the first non-fiction example of a VR device that can actually kill the user.  It won’t be the last.
See you in the metaverse.
-Palmer

Today marks ten years since the start of Oculus.  I decided to drop out of school when I was 19 to make a go at turning my extremely niche hobby of building high-performance head-mounted-displays into a business that could get everyone interested in VR and AR.  It took a couple hours to come up with the name, register the website, and spread the news on all the internet forums I frequented.  It would take almost two months to get back final paperwork for Oculus LLC from the California Franchise Tax Board, of course!
Just a few days short of five years later, I was fired from my own company.
I learned a lot of lessons and made a lot of money along the way, and my new company Anduril wouldn’t be where it is if it wasn’t for Oculus.  Over the next year, I am going to try to distill some of those lessons into vignettes others might find useful, posting more or less in time with the decennial of various Oculus milestones (crowdfunding, fundraising, shipping, acquisition, termination, etc).  I have quite a bit to share, especially regarding portions of Oculus history that have been subject to extensive revisionism in public accounts.  This year is also the right time to finally unveil some VR technologies I haven’t been able to talk about for a variety of reasons.  For now, though, I just want to share the first official communication from Oculus VR, the “About” page text from oculusvr.com posted on April 15th, 2012:
Let me get this out of the way:  I am a huge HMD nut.  As far as I know, I have the largest private collection of unique HMDs in the world, totaling 43 units, not including the ones I have built myself.  Part of this HMD obsession, as with any other obsession, is trying to convince other people to join in.  And what a seductive dream, that technology can transport us into worlds we cannot hope to experience in real life, or augment our reality to shape it closer to our desires!
Unfortunately, virtual reality has risen and fallen many times, with a lot more emphasis on the latter portion.  The tech has never gotten far enough to be truly convincing, and great VR hardware has been far out of reach for the average person… Until now.
Oculus is my tilt at trying to change that.  The tech has improved, and we can build hardware and software that is better, stronger, and faster than the old guard, companies that create niche, wildly expensive products.  Don’t get me wrong, these companies are important, and they have to solve some very tough engineering challenges to satisfy their customers.  But the reality is that as gamers and dreamers, we have a different set of challenges to meet.  Massive field of view to engulf your visual senses, low latency tracking to maximize presence, light weight and comfortable for long term use, and perhaps most importantly, prices measured in the hundreds of dollars, not  tens of thousands.  I have worked long and hard with a lot of brilliant people to try and meet those challenges, and now it is time to put it in your hands.
Why the name “Oculus?  Because it is the Latin word for “eye”, and someone used the word in a meeting several months ago.  I thought it was a nifty word, and was better than the alternative, “StepN2theGAME”.
And there you have it.
Best,
PalmerTech

https://vimeo.com/456288740
anduril.com/ghost
I’ve been going to the Consumer Electronics Show (CES) every January ever since I wasn’t allowed to, back when I was 16 years old with a specious ModRetro business card and little tech industry experience beyond my very first prototype of what would eventually become the Oculus Rift. While Anduril doesn’t make consumer electronics, I still attend every year to scout out new technology and meet with friends, partners, and suppliers who brave the Vegas crowds in search of tech esoterica.
Twelve years of watching drone technology at CES emerge and then boom has made one thing abundantly clear: China Won.  American companies have built a lot of cool science fair projects that get lots of media attention, but minimal real-world deployment – from light shows to food delivery.  We also design a lot of interesting low-volume consumer drones, but most of them lean heavily on Chinese R&D and then go on to have their product manufactured in China.  Yes, I know, I am fully aware I did the same thing years ago with Oculus.
This situation would piss me off a lot less if China was only winning the sales game.  As a PC gamer, I intimat

Today is November 6th, 2022, the day of the SAO Incident.  Thousands of VRMMORPG gamers were trapped by a mad scientist inside a death game that could only be escaped through completion.  If their hit points dropped to zero, their brain would be bombarded by extraordinarily powerful microwaves, supposedly killing the user.  The same would happen if anyone in the real world tampered with their NerveGear, the virtual reality head-mounted-display that transported their minds and souls to Aincrad, the primary setting of Sword Art Online.
This type of scenario has been a staple of science-fiction for decades, but Sword Art Online exploded in popularity at exactly the right time to have a massive impact on the real world.  The already-popular web novel was adapted into an anime series that aired just as the Oculus Rift went from internet forum curiosity for turbonerds to siren of the games industry, endorsed by giants like John Carmack, Gabe Newell, and Cliff Bleszinski – the 4th episode of SAO, “The Black Swordsman” (黒の剣士) aired in Japan just as the Kickstarter for the Oculus Rift DK1 launched!
The popularity of SAO led to massive otaku enthusiasm for Oculus, especially in Japan, which quickly became our 2nd largest market.  In turn, the existence of the Rift made SAO itself seem far more plausible and grounded – a story that had been written in a world where VR was a dead technology was now straight out of the gamer hype headlines.  This synergy had meaningful impact on our dev kit sales and adoption – literally thousands of people reached out to me asking variations of “Have you seen Sword Art Online?  When will you make the NerveGear real?!”.  The Oculus SDK was the closest thing to SAO’s World Seed, a software tool that enabled anyone to create incredible VR worlds.
There are dozens of stories I could tell, many of them worth a post of their own.  How we ended up with a Kirito mural on the wall of the first Oculus office, how it was eventually moved to Facebook HQ, why it was removed shortly after I was fired.  Personally showing Oculus DK2 to Reki Kawahara at Anime Expo, my wife meeting him while cosplaying as Sinon at the next Anime Expo, getting into minutiae of SAO canon at the Hollywood premiere of SAO: Ordinal Scale, the audience screaming as Yuuki came back from the dead to perform a two-user Mother’s Rosario attack combo with Asuna against the 100th floor boss of Aincrad.  Arguing with people who only watched the anime about the supposed plot and tech holes, explaining that the anime was only ever meant to act as a companion piece for fans of the novel series that so frequently predicts the future of AR, VR, AI, and more.  The beautiful wedding gift from Kawahara-san that shows he is a double-threat writer/artist combo!
But that isn’t what you are here for.  You want NerveGear, the incredible device that perfectly recreates reality using a direct neural interface that is also capable of killing the user.  The idea of tying your real life to your virtual avatar has always fascinated me – you instantly raise the stakes to the maximum level and force people to fundamentally rethink how they interact with the virtual world and the players inside it.  Pumped up graphics might make a game look more real, but only the threat of serious consequences can make a game feel real to you and every other person in the game.  This is an area of videogame mechanics that has never been explored, despite the long history of real-world sports revolving around similar stakes.
The good news is that we are halfway to making a true NerveGear  The bad news is that so far, I have only figured out the half that kills you.  The perfect-VR half of the equation is still many years out.
In SAO, the NerveGear contained a microwave emitter that could be overdriven to lethal levels, something the creator of SAO and the NerveGear itself (Akihiko Kayaba) was able to hide from his employees, regulators, and contract manufacturing partners.  I am a pretty smart guy, but I couldn’t come up with any way to make anything like this work, not without attaching the headset to gigantic pieces of equipment.  In lieu of this, I used three of the explosive charge modules I usually use for a different project, tying them to a narrow-band photosensor that can detect when the screen flashes red at a specific frequency, making game-over integration on the part of the developer very easy.  When an appropriate game-over screen is displayed, the charges fire, instantly destroying the brain of the user.
This isn’t a perfect system, of course.  I have plans for an anti-tamper mechanism that, like the NerveGear, will make it impossible to remove or destroy the headset.  Even so, there are a huge variety of failures that could occur and kill the user at the wrong time.  This is why I have not worked up the balls to actually use it myself, and also why I am convinced that, like in SAO, the final triggering should really be tied to a high-intelligence agent that can readily determine if conditions for termination are actually correct.
At this point, it is just a piece of office art, a thought-provoking reminder of unexplored avenues in game design.  It is also, as far as I know, the first non-fiction example of a VR device that can actually kill the user.  It won’t be the last.
See you in the metaverse.
-Palmer

Today marks ten years since the start of Oculus.  I decided to drop out of school when I was 19 to make a go at turning my extremely niche hobby of building high-performance head-mounted-displays into a business that could get everyone interested in VR and AR.  It took a couple hours to come up with the name, register the website, and spread the news on all the internet forums I frequented.  It would take almost two months to get back final paperwork for Oculus LLC from the California Franchise Tax Board, of course!
Just a few days short of five years later, I was fired from my own company.
I learned a lot of lessons and made a lot of money along the way, and my new company Anduril wouldn’t be where it is if it wasn’t for Oculus.  Over the next year, I am going to try to distill some of those lessons into vignettes others might find useful, posting more or less in time with the decennial of various Oculus milestones (crowdfunding, fundraising, shipping, acquisition, termination, etc).  I have quite a bit to share, especially regarding portions of Oculus history that have been subject to extensive revisionism in public accounts.  This year is also the right time to finally unveil some VR technologies I haven’t been able to talk about for a variety of reasons.  For now, though, I just want to share the first official communication from Oculus VR, the “About” page text from oculusvr.com posted on April 15th, 2012:
Let me get this out of the way:  I am a huge HMD nut.  As far as I know, I have the largest private collection of unique HMDs in the world, totaling 43 units, not including the ones I have built myself.  Part of this HMD obsession, as with any other obsession, is trying to convince other people to join in.  And what a seductive dream, that technology can transport us into worlds we cannot hope to experience in real life, or augment our reality to shape it closer to our desires!
Unfortunately, virtual reality has risen and fallen many times, with a lot more emphasis on the latter portion.  The tech has never gotten far enough to be truly convincing, and great VR hardware has been far out of reach for the average person… Until now.
Oculus is my tilt at trying to change that.  The tech has improved, and we can build hardware and software that is better, stronger, and faster than the old guard, companies that create niche, wildly expensive products.  Don’t get me wrong, these companies are important, and they have to solve some very tough engineering challenges to satisfy their customers.  But the reality is that as gamers and dreamers, we have a different set of challenges to meet.  Massive field of view to engulf your visual senses, low latency tracking to maximize presence, light weight and comfortable for long term use, and perhaps most importantly, prices measured in the hundreds of dollars, not  tens of thousands.  I have worked long and hard with a lot of brilliant people to try and meet those challenges, and now it is time to put it in your hands.
Why the name “Oculus?  Because it is the Latin word for “eye”, and someone used the word in a meeting several months ago.  I thought it was a nifty word, and was better than the alternative, “StepN2theGAME”.
And there you have it.
Best,
PalmerTech

https://vimeo.com/456288740
anduril.com/ghost
I’ve been going to the Consumer Electronics Show (CES) every January ever since I wasn’t allowed to, back when I was 16 years old with a specious ModRetro business card and little tech industry experience beyond my very first prototype of what would eventually become the Oculus Rift. While Anduril doesn’t make consumer electronics, I still attend every year to scout out new technology and meet with friends, partners, and suppliers who brave the Vegas crowds in search of tech esoterica.
Twelve years of watching drone technology at CES emerge and then boom has made one thing abundantly clear: China Won.  American companies have built a lot of cool science fair projects that get lots of media attention, but minimal real-world deployment – from light shows to food delivery.  We also design a lot of interesting low-volume consumer drones, but most of them lean heavily on Chinese R&D and then go on to have their product manufactured in China.  Yes, I know, I am fully aware I did the same thing years ago with Oculus.
This situation would piss me off a lot less if China was only winning the sales game.  As a PC gamer, I intimately understand that the cheaper option (consoles) sell very well given their ease of use and affordability.  In the case of drones, though, China leads the global drone industry across the entire spectrum, not only in quantity, but in quality!  DJI dominates the high end by building incredible imaging, networking, power, and aero systems while hundreds of Shenzhen-located companies nip at their heels by cranking out cheaper mid-range and low-end drones, along with most of the offerings designed by Western companies.  DJI might get temporarily beaten here and there by some upstart focusing on a specific feature or specification, but when it comes to the overall package, they rarely even get close.  When it comes to consumers, that overall package is what matters.
Sadly, this is also true for militaries around the world, including the United States and our allies.  Given the amazing capabilities offered by the latest consumer drone technology, many units with any sort of purchasing authority discarded US-made drones that cost literally hundreds of thousands of dollars in favor of Chinese drones that cost just a few hundred dollars, driven by desire for compact size, great image quality, ease of use (especially during takeoff and landing) and the ability to hover in place rather than flying orbits (circles) around ever-changing views of their targets.  The 2019-2020 National Defense Authorization Act banned most government entities from purchasing or using Chinese drones, which inspired quite a few well-meaning but half-hearted attempts by US drone startups to adapt hobbyist and consumer drone designs for military service.  Designs that were optimized around the needs of consumers and hobbyists are rarely the right answer for brave users who spend a little more and demand a heck of a lot more, from security to durability to endurance and beyond.  In other words, the men and women who serve on the front lines.
This is why we built Ghost.  It is American-designed, American-made, and built from the ground up for the needs of warfighters, the most demanding users in the world.  We threw out the template everyone has been using for a decade and started from first principles, using world-class design to make something extraordinary.  We didn’t invent any new chemistry, metallurgy, or physics, we just used the same materials available to everyone in a very new way.  Warfighters have already been using previous versions of Ghost for a while now, but we are finally able to talk more publicly about our work in this space as we launch Ghost 4.
I can’t possibly cover all of the elements of Ghost that make it the best drone there is (that a person can carry, anyway – Predators don’t count) because there are just too many things to cover.  The best way to try is probably comparisons with a few elements of other drones people are familiar with:
Ghost is a single-rotor aircraft.  The mechanicals of the rotor system are harder to design and manufacture than a DJI-style quadcopter, and flight controls are harder to program, but the result is much lower aerodynamic disk loading that results in longer endurance (over 100 minutes with real mission payloads), near-silent acoustic signature, high max payload capacity (dozens of pounds), and high speed.  Compare this with a DJI Inspire 2, a high-end offering that costs $3000 that can lumber around with a couple pounds for 24 to 27 minutes, props screaming.  If we wanted to create the best marketing numbers, we could fly a drone with nothing on it for even longer, but that isn’t my style.  Multicopters are better and easier for some applications, but not this.
Ghost is autonomous, powered by an onboard AI Core that can perform 32 trillion operations per second.  That is hundreds of times more than most drones this size, and we use it to power complex computer vision and sensor fusion algorithms that make Ghost into something less like a manually operated RC plane or blob-following quadcopter and more like a real pilot, capable of making decisions and acting on them with minimal human guidance.  The operator can rely on Ghost to reliably perform all kinds of tasks for them, even when communications are jammed or turned off to avoid detection.  It is hard to overstate how much software work is involved in this capability – the future of warfare will be superficially hardware-driven with a buttload of software smarts under the hood making it all actually work.
Ghost is compact, not just for the capabilities provided, but in absolute terms.  To make another comparison, Ghost packs up into a bag that is even smaller than the DJI Inspire carry case!  This is important when you want to ship, carry, and deploy dozens of Ghosts from a single truck, or just carry a single unit with you into the field.
Ghost is practically invisible to the targets it observes, with a frontal cross-section smaller than some of the phones I have owned over the years.  Stay tuned for true invisibility.
Ghost sees you no matter what, using stabilized visible spectrum imaging, thermal imaging, and other specialized sensors all fused into a single picture by our AI Core.
Ghost is incredibly rugged, a mix of well-machined metals and carbon fiber composites with a direct-drive motor and minimal wear components.  We weren’t trying to set flight endurance records – our top priority was to make something that could survive neglect, desert heat, arctic cold, high winds, heavy shock, abusive combat environments, and submersion in salt water.  Most of our competition will get taken out by any one of those things, and we still manage to fly far longer.
Ghost is modular and never obsolete, with payload rails on the top and bottom of the aircraft that provide access to power and data for new capabilities from both Anduril and third-parties, from electronic warfare to loudspeakers to satcom to laser designators.  Unlike other drones that have a single accessory attachment point (or none), we accommodate five standard payload modules on the top and as many as space and weight allow on the bottom.
Ghost doesn’t use a crazy power source (yet). Our battery packs are utilizing the same type of 21700 battery cells used by companies like Tesla and Samsung – not the most energy-dense cells out there, but we can (and have) changed our endurance using modular-rail mounted power supplies with more exotic battery chemistries, gasoline, and hydrogen fuel.  We didn’t get great endurance with science fair project technology, we did it with the most stable, affordable, and logistically-sustainable tech out there, which was much harder.  More to come on other power sources in the future.
Ghost is designed from the start for swarming and teaming, allowing a single operator to control large numbers of mesh-networked airframes across vast distances, sometimes from the other side of the world.  This is so much harder than locally controlling a single drone as a single operator, but the tactical rewards are obvious.  This is only possible because of our previous work on Lattice, which allows us to fuse data from thousands of sources into a single AI-driven collaborative common operating picture.  The work we have done with Ghost is going into our other unannounced drone projects as well.
All those other things you imagine Ghost is going to do in the future, it is going to do.
It is hard to call any device “the best” without qualification.  There will always be specific comparisons where a particular feature was prioritized at the expense of others, or arguments about semantics, definitions, and research lab demonstrations of immature technologies.  In this case, though, I feel very comfortable saying that Ghost 4 is the best drone there is.  I can’t wait to get the latest version into the skies with US warfighters and their allies across the globe – orders start right now, shipments start in Q1 2021. I owe a huge thank you to the partners who have helped us go from janky prototype to production, especially the Future Commando Force.
Oh, and I am really bummed that CES is not happening this year.  I wanted to show up with a big booth for Ghost 4 in the middle of the drone section, with plenty of flags and a big banner proclaiming “THE BEST DRONES ARE AMERICAN DRONES”.  Maybe Ghost 5?
anduril.com/ghost