A Marine tiltrotor pilot walks into a coordination gap between thirty vendors' worth of drones and decides the interesting job is not building another drone. It is telling the drones what to do next.
Philong Duong runs NODA AI out of Austin, a company that would like to be the reasoning engine of the autonomous force. That is a phrase from the company website, and like most phrases from company websites it is easier to say than to prove. What Duong actually built, and what Bessemer, Booz Allen Ventures, Draper Associates, Bloomberg Beta, and Alumni Ventures collectively wrote a $25 million check for in February 2026, is a piece of software called Urza that pretends none of the drone companies exist. Or rather, it pretends they all exist equally, which for the drone companies amounts to the same thing.
Urza is an orchestration layer. It sits above thirty-plus OEM platforms - a category that in defense includes small unmanned aircraft, ground vehicles, and various kinds of sensor packages that companies would prefer you buy exclusively - and hands out tasks. If a Group 3 UAV from vendor A can do the mission cheaper than the ground robot from vendor B, Urza sends the job to vendor A. If the situation changes and vendor B is closer, Urza sends the job to vendor B. The vendors do not have to agree to this arrangement, which is the whole point.
This is a strategic choice that also functions as a business model. In defense hardware, the last decade of venture money went into building better drones. Duong's bet, articulated in an interview with Tectonic Defense, is that having better drones eventually becomes table stakes, at which point the interesting question is who tells thirty different drones from twelve different vendors to move at the same time. He phrases this in chess terms, which is a common metaphor for founders selling to the Pentagon and an increasingly literal one. NODA wants to be the player. Everyone else can be the pieces.
Then Duong extends the metaphor into territory most founders leave alone. “After the nuclear weapon,” he told Tectonic, “the next strongest deterrence model is going to be the person who builds the best orchestrator.” This is a large claim. It is not the sort of thing you say if you are trying to sound modest, and Duong is not, but it is also not the sort of thing you say if you have not thought about it. Orchestration as deterrent is a specific argument: an adversary is not deterred by the number of your assets but by the coherence with which those assets can be sequenced. The scary thing is the tempo.
NODA's product roadmap suggests they take this seriously. Alongside Urza there is Laria, described as the weapons and tactics school for autonomy - essentially a synthetic environment for developing plays. And then there are the algorithms themselves, packaged and named the way military capabilities have always been named: MONET, KRAKEN, ANGELSCAPE, COUNTER SHORAD, URBAN ISR NETWORK, ELECTRONIC ATTACK FORMATION. If you have spent time inside an operations center, this vocabulary reads as familiar. If you have spent time inside a Silicon Valley company, it reads as branding. Both are correct.
The Department of War has, per NODA's own disclosures, previously assessed and selected the company as sole orchestrator for the Multi-domain Collaborative Autonomy program. Sole orchestrator is a phrase worth pausing on. In defense procurement, sole anything is unusual. It typically implies either capture (bad) or a genuinely uncontested capability (rare, and what NODA is claiming). The UK Ministry of Defence is also a customer. Pentagon programs LUCAS and the collaborative autonomy initiatives have run through the platform.
Duong was, before all this, C3.ai's first Federal Product Manager. He arrived from active duty and built the company's federal practice up from under $1 million in ARR to nearly $100 million in under four years. He also helped C3.ai IPO at the end of 2020, though on a personal roster of accomplishments IPOs and ARR ramps are perhaps the less interesting entries. Before C3.ai, Duong flew CH-46E and MV-22B aircraft for the Marine Corps. He served as a JTAC - Joint Terminal Attack Controller - on multiple Middle East deployments, which means he was the guy on the ground calling in the aircraft. He served as a foreign military advisor. He is currently, per his LinkedIn, in the Selected Marine Corps Reserves.
His co-founder is Dave Corbett, who was also a Marine Corps joint fires instructor and also a colleague on C3.ai's Defense and Intelligence Products team. Joint fires instructor is where they both, independently and then together, saw the coordination gap that NODA now sells against. The story founders tell about seeing the problem firsthand is usually a story. In this case, per Bessemer's own writeup of the investment, it appears to be the actual sequence of events.
The company was demonstrating Urza within months of being founded. The pre-seed was $4 million. Nine months later the Series A closed at $125 million post-money, a pace that is either evidence of a hot market or evidence that the founders had a working product before they had a fundraising deck. Both are probably true. Bessemer's Janelle Teng Wade, who led the deal, describes NODA as “building the AI-native connective tissue for defense autonomy,” which is investor-speak for the same chess metaphor, filtered through a term sheet.
The most interesting biographical fact about Duong is not any of the above, though. It is that he co-founded The Wingman Foundation, a nonprofit dedicated to serving the families of fallen naval aviators and maintainers, and he remains on its board. His callsign there, listed on the foundation site, is “Donger.” He is the son of Vietnamese immigrants who arrived in the United States with little money and less English, in a family whose members had fought for their freedom. He grew up in Silver Spring, Maryland. He studied computer science and minored in philosophy at Elon. He has a Columbia MBA from 2017. He is, in other words, the specific American type that shows up at the intersection of duty, code, and capital - a type that is currently reshaping a lot of the defense-tech cap tables.
In our mind, the key differential in future conflict is not going to be what the best Group 3 UAV is. It's going to be whoever is building the best decision models and executing the right algorithms, at the right time, at the most relevant speeds.
The $25 million Series A closed February 2026 at a $125 million post-money valuation. Bessemer Venture Partners led. Booz Allen Ventures, the corporate venture arm of the $12 billion federal contractor, joined - a signal both that NODA has real Pentagon adoption and that Booz Allen would like to integrate the platform into its own delivery channels. Draper Associates and Bloomberg Beta added credibility from the classical Silicon Valley side. Alumni Ventures filled out the syndicate. Paige Craig of Outlander, an early backer, described NODA as “the fastest growing company in our portfolio.”
This is nine months after the $4 million pre-seed. It is, by any conventional metric, fast. It is also the current pace of defense-tech company formation, which has been reshaped by the reindustrialization thesis, the shift in Pentagon procurement toward software-defined capability, and a specific class of founder that Duong exemplifies - operator, engineer, veteran, and comfortable inside a term sheet.
Vendor-agnostic tactics engine across mixed fleets. Real-time task brokering. Thirty-plus OEM vehicle integrations. Designed to be the layer that no drone company gets to own.
Synthetic environment that combines simulators to develop and test autonomous tactics and procedures. Called the “weapons & tactics school for autonomy” on NODA's own site.
Named plays for specific tactical applications. Counter SHORAD, urban ISR network, electronic attack formation. The vocabulary reads like a doctrinal library, which appears to be the point.
After the nuclear weapon, the next strongest deterrence model is going to be the person who builds the best orchestrator.
CEO and co-founder of NODA AI in Austin. Marine Corps veteran, former CH-46E and MV-22B pilot, JTAC, and C3.ai's first Federal Product Manager.
Builds Urza, an AI-native orchestration platform that coordinates heterogeneous manned and unmanned systems across vendors and domains for defense customers.
$25M Series A led by Bessemer Venture Partners in February 2026 at a $125M post-money valuation, following a $4M pre-seed nine months earlier.
Dave Corbett - fellow Marine Corps joint fires instructor and former colleague on C3.ai's Defense and Intelligence Products team.
Yes, in the Selected Marine Corps Reserves.