Introduction
I spent a decade in cybersecurity and data engineering before I started building defense software at Empyrean Defense. If there's one thing that experience taught me - beyond dealing with annoying data - it's how to recognize when an industry is lying to itself.
Cyber was, and still is, drowning in it. Every product is a silver bullet. Every vendor deck opens with the same scare slide: breach costs, dwell time, nation-state threat actors. They close with the same promise: buy our thing, and the bad stuff stops. It never stops. The products don't do what the slides say. The customers know it. The vendors (probably) know it. The analysts who write about the Magic Quadrants know it.
Everyone seemingly performs the ritual anyway because no one has figured out how to sell a tool by saying "this will make one part of your problem 15% less terrible, under conditions that may or may not match yours." That sentence is honest. However, in enterprise SaaS sales, especially in security, honesty doesn't close deals.
Each problem has a solution, and that’s what I wanted to write about. Empyrean Defense Research has been kicked off to at least bring some auditable honesty to analysis. If you don’t care to read further: check out this link to our dedicated Research page where we’ll be publishing artifacts.
The Defense Analysis Landscape
Defense publishing has its own version of the same disease. Congressional testimony that selectively cites parametrics to justify a program that's already funded. Vendor white papers where the simulation always concludes "you need more of the thing we sell." Think tank reports that hedge every finding into a fog of "may" and "could" and "further study is recommended" until the reader can't extract a single actionable claim. OSINT Twitter threads from people who have strong opinions about weapons systems they've never touched, citing sources they haven't read, reaching conclusions that feel good but wouldn't survive a high school physics exam.
None of these are useful. Some of it is actively harmful - not because the authors are malicious, but because the incentive structures reward ambiguity and punish clarity. If you say something specific, you can be wrong. If you're wrong, you lose credibility. If you lose credibility...well, need we go further? S,o you don't say anything specific. You say "capability gap" and "emerging threat" and "multi-domain challenge" and everyone nods, and nothing changes, and the warfighter is no better informed than before your PDF existed.
We think there's room for something different.
What Empyrean Defense Research Is
Empyrean Defense Research is our attempt to do defense analysis the way we think it should be done; specific, transparent, reproducible within a platform, and published in full without sponsorships or sales gating. We do this with public domain and open-source data from a myriad of sources. We don’t have access to classified pKill tables or false alert rates for a certain class of threats. We want to use what is widely reported about a system and see if the physics checks out.
Every piece of research we publish follows the theme of eliminating variables which setup the foundation for other work. We don’t have access to an anechoic chamber, the United States government will not allow us to install a THAAD battery in the driveway, nor is any other nation interested in loaning us fighter aircraft. The next best thing we can do is follow the same combat physics that has been around for several decades and run public doctrinally aligned engagements within the confines of what the physics allows.
The engine behind this work is the Empyrean Defense Wargaming & Simulation Cyber Range; what started as an organic training module inside our platform and grew into a high-fidelity simulation environment that models engagement physics across domains. Sea states, RF propagation, blast overpressure, atmospheric extinction, beamforming, thermoclines. Dozens of publicly available, peer-reviewed physics models stitched together into a pipeline that resolves complex engagement chains end-to-end. The simulation runs on the same COP and fusion pipeline operators use in live operations - the only thing that changes is the data source.
Every entity in our scenarios is built from open-source data: U.S. and allied defense publications, academic research, OSINT, manufacturer specifications. We call these entity files "playing cards," and every field on every card carries source attribution.
We do not claim these numbers will survive a real fight. We do claim the math is honest, the assumptions are stated, and the methodology is transparent. Every simulation run is reproducible from the scenario file and seed (conditionally available under MNDA). Where we approximate, we say so. Where we're wrong, we want to know, and we'd rather find out from a surface warfare officer or a program office evaluator telling us where our drag model breaks than from the silence of people who read it and moved on.
The First Publication: Quantifying Layered Naval Defense Against Hypersonic Glide Vehicles
Our first research brief, Quantifying Layered Naval Defense Against Hypersonic Glide Vehicles, is a Monte Carlo simulation study of a single Arleigh Burke Flight IIA destroyer defending against DF-17 hypersonic glide vehicle salvos. 1,500 runs. Three defensive loadout variants - including the PAC-3 MSE/Aegis integration the Navy plans to field by FY2027.
We gave the defender every advantage we could think of. Perfect radar, perfect guidance, perfect crew, no jamming, no multi-axis attack, calm seas, clear weather. __The ship died in every run.__ Every variant. Every evasion level. Every time.
PAC-3 MSE achieved zero kills across all evasion levels - including non-maneuvering ballistic reentry. The hit-to-kill guidance designed for Mach 3-4 theater ballistic missiles cannot close the miss distance gap at Mach 10 closing velocity. The physics are brutal and uncaring, the type of indifference only mathematics can bring.
The data also surfaced a finding we didn't expect; terminal evasion maneuver helps the defender, not the attacker. Maneuvering HGVs extend their time in the SM-6 engagement envelope, producing more intercept opportunities for blast-fragmentation warheads. We do realize there are more opportunities to eliminate hypersonic threats - ideally before they leave the TEL, or before terminal - but we cannot assume there won’t be leakers.
The piece that inspired us to test this scenario was Patriot PAC-3 Missiles To Arm Navy Arleigh Burke Class Destroyers by Joseph Trevithick at The War Zone. I must have been living under a rock, because I did not realize the program was so far along, and the thought of Patriot missiles on my favorite naval vessel is legitimately cool. So, we set out to mockup what the magazines may look like, against the hypersonic threat mentioned in the article (and in many others).
Today, before publishing this piece and the research, Sakshi Tiwari at The EurAsian Times published his piece Patriot Missiles: After Combat Success on Land, U.S. To Navalize PAC-3 MSE, Boosting Navy’s Layered Air Defense that runs counter to our findings, regarding terminal hypersonic threats at least. This is purely coincidental and very welcome; the author had his own thoughts on the strengths of the whole naval air defense system, which is something we want to explore more broadly too!
The full brief - including the complete methodology, entity specifications, engagement timelines, cost-exchange analysis, and a detailed physics appendix with worked examples - is available now, ungated, in full.
Why Ungated, Why Free
In cyber, gated content is the norm. You hand over your email, your title, your company name, and in exchange you get a PDF that could have been a blog post. Then the SDR calls start. The BDR emails. The "just checking in" LinkedIn messages. The content isn't a resource; it's a trap. The person who downloaded it didn't want a relationship with your sales team. They wanted to understand the problem. You punished them for trying.
We don’t want to punish people who are curious or want data to help with their decision-making capability. I mean, for crying out loud, we call our software the Decision Dominance Engine - it would not be very on brand to hide the research. Also, it’s all based on public domain data that we extrapolate to build the playing cards, further built upon by peer reviewed journal entries and formulas for the physics that underpin the whole thing. We realize we are not the first to come up with a physics-backed wargaming platform, and it will only become more featured as we develop the module further.
If the research compels you to want to see a demo, see how Empyrean Defense can help you organize whether you're in commercial or defense, then we have a contact form for that very reason. At the end of the day, we just want to put out cool content, and for us it’s very cool to be able to dogfood our own product.
What's Next
__Quantifying Layered Naval Defense Against Hypersonic Glide Vehicles__ is a naval scenario, but not everything will be. We're building a compendium of wargaming scenarios across the full spectrum of Joint All-Domain Operations: air defense, counter-UAS, electromagnetic warfare, space domain effects, and the sensor fusion problem that ties all of it together. It has given us the idea to do follow-ons of the same engagement and layer in more ships, more threats, EW effects, and more.
Not every bit of research artifact produced will be so devastating, at the end of the day we’re Americans, it doesn’t serve us much good to only publish the bad. So, when we model out what a 250KW HEL can do against a hypersonic glide vehicle, or when we demonstrate some physics-backed capabilities of equipment out in the field - we will publish that as well.
We're not here to sell you a conclusion. We're here to show you math and let physics speak, with some thoughts and conclusions from our own unique perspective. Research artifacts are the same as our Insights Blog content, so expect at least one new piece of material from either vertical every 2 to 3 weeks, time permitting.
Empyrean Defense Research publications are UNCLASSIFIED and use only open-source, publicly available data. No classified or restricted information is used at any point in our research. Distribution is unlimited.
Stay Dangerous.
