Hey friends.

I want to take you back to March 27, near Riyadh.

An American E-3 Sentry AWACS, one of the most expensive and irreplaceable aircraft in the entire US inventory, a jet I personally served on for four years, sits parked at Prince Sultan Air Base alongside a handful of other assets.

Iran fires a barrage of missiles and drones.

When the smoke clears, twelve American service members are wounded, that AWACS is completely destroyed, and eight days later a US soldier who’d been hurt in the attack dies of his injuries.

I was pissed, as were many in my AWACS community.

For six months, the question hanging over OSINT world was how did Iran, a regional power without anything close to America’s own reconnaissance architecture, find and hit that specific aircraft with that much precision?

Many of my readers correctly summarized at the time that they must have had help from the Russians. President Zelensky was also warning the US publicly. And yet, we didn’t listen.

This week, CNN gave us confirmation of exactly how much help. Two US officials and additional intelligence sources told the network they have evidence Russian satellites fed targeting intelligence to Iran, and that the scale of the damage would have been nearly impossible to achieve without that Russian geospatial support.

CNN independently reconstructed the collection sequence using orbital data and outside space-tracking firms, Project Aurora and Kayhan Space, and found that fourteen Russian reconnaissance satellites, led by the advanced Cosmos 2573, passed over Prince Sultan in a coordinated sequence in the days before the strike.

A single satellite photograph is super useful. But fourteen satellites, running different sensor types in a coordinated pass over one target, is something else entirely: It’s building the first half of a kill chain, (find, fix, track).

A missile only knows where you tell it to go

People obsess over the missile itself. Things like range, warhead size, speed, accuracy. All beside the point if you don’t have precise, current targeting information to feed the weapon in the first place.

Knowing that an American base sits at a certain latitude and longitude is enough to hit the base in general terms.

It isn’t enough to hit the specific AWACS parked on a specific ramp (exactly in the rotodome no less), or to find the Patriot radar tucked behind a berm, or to know which shelter a THAAD battery moved into overnight, or to identify which radar is actively transmitting and from exactly where.

That level of precision requires intelligence, surveillance, and reconnaissance, and it requires it fresh, because military assets move, hide, and power down constantly.

The modern targeting cycle military planners use runs through six steps, sometimes shortened to F2T2EA. Find the target. Fix its precise location. Track how it changes over time. Develop an actual aimpoint (target). Engage it. Then assess whether the strike actually worked.

Based on CNN’s reporting, Russia appears to have helped Iran with several of those steps at once.

Let me start with the optical layer. Cosmos 2573 is one of Russia’s premier reconnaissance satellites, capable of collecting detailed, military-grade imagery.

That kind of imagery tells you where aircraft are parked, where vehicles have concentrated, where new structures have gone up, where defenses might be positioned, and maybe most important, what’s changed since the previous pass over the same spot.

Optical imagery has real limits, though. Clouds block it. Darkness blocks it. Which is why Russia didn’t rely on Cosmos 2573 alone, instead running a full collection architecture around it.

Russia used eight more satellites, listening instead of looking.

The array included satellites built specifically to intercept radio and radar emissions, alongside spacecraft carrying infrared sensors and synthetic-aperture radar.

A Patriot radar has to emit electromagnetic energy to do its job, and so does THAAD, and so do most military communications networks. Any system that emits announces itself to anyone listening, and it can be detected from a serious distance without ever needing a photograph of it.

With enough patient collection, analysts can start piecing together a radar’s operating frequency, its rough location, its activation schedule, its coverage envelope, and shifts in a defender’s overall posture.

That’s the picture you’d want before planning a strike on defended American assets, and it’s the picture Russia has spent years practicing against Ukraine’s own Patriot, NASAMS, IRIS-T, and Buk networks.

Two more satellites in that same sequence carried infrared sensors, and infrared answers a different question than a normal photograph does.

Heat shows activity: A running engine, a generator, a missile launcher warming up, machinery that’s recently been used all show up as heat; all proof of life.

Then there’s synthetic-aperture radar, or SAR, which I’ve written about before. SAR uses radar rather than visible light, which means it works straight through darkness and cloud cover without any degradation.

Its real strength, though, shows up over repeated passes rather than a single snapshot, because it can flag exactly what’s physically changed on the ground between one orbit and the next like a battery that relocated, a new concentration of vehicles that wasn’t there yesterday, a freshly built revetment, or an aircraft that’s suddenly parked where nothing was before.

Three of the fourteen satellites in Russia’s Prince Sultan pass carried this capability.

Put all four sensor types together, optical for identification, signals intelligence for listening, infrared for activity, SAR for change detection, and you’re assembling an actual target package.

Then Russia came back after the strike.

Ukraine’s own intelligence assessment described continued Russian satellite activity around targets after Iranian attacks had already occurred, documenting at least two dozen reconnaissance surveys covering dozens of sites across multiple Middle Eastern countries within a single ten-day stretch in late March.

That’s the final, closing piece of the kill chain: battle damage assessment.

Did the missile actually hit?

Did the radar survive intact?

Did surviving aircraft or equipment relocate afterward?

Does the target need a second strike?

If a satellite revisits the same site after a strike, it can answer every one of those questions, and answering them turns a single attack into a repeatable, improvable cycle.

Reconnaissance feeds target development, target development feeds a strike, the strike gets followed by another satellite revisit, that revisit produces a damage assessment, and the damage assessment feeds straight back into retargeting for whatever comes next.

So how did Ukraine know first?

Ukraine flagged this pattern months before CNN independently proved much of it.

Zelensky told NBC, in an interview following the Prince Sultan strike, that he’d received daily presidential intelligence briefings showing Russian satellites photographing the Saudi base on March 20, March 23, and March 25, a sequence that all but announced an attack was being planned.

“I think that it’s in Russia’s interest to help Iranians,” Zelensky said. “And I don’t believe, I know, that they share information.”

Five months later, CNN independently tracked the same satellites using entirely different methods and arrived at essentially the same conclusion.

We don’t know exactly how Ukraine’s intelligence services built that early picture, but a few plausible avenues stand out.

Spy satellites aren’t invisible, and their orbits can often be tracked by anyone with the right equipment, the same way CNN later reconstructed Cosmos 2573’s flight path using civilian space-tracking firms.

Ukraine’s own intelligence services, with four plus years of practice watching Russian military behavior in exhaustive detail, could have simply asked the obvious question: Why are Russian reconnaissance satellites suddenly clustering over an American base in Saudi Arabia, and why does an Iranian strike follow shortly after?

Add in four years of intercepting Russian communications and military networks and Ukraine’s extensive intelligence-sharing relationships with Western partners, and it seems obvious in hindsight.

Why didn’t Washington listen?

This pisses me off more than the actual strike.

Ukraine didn’t wait for CNN’s September investigation to raise the alarm. Zelensky went public with almost this exact warning on March 28, one day after the Prince Sultan strike, posting on X that Russian satellites had photographed multiple strategic sites “in the interests of Iran,” and telling NBC News he was “100 percent” certain Russia was sharing that intelligence with Tehran.

He didn’t stop there.

He explained Russia’s own pattern to American reporters in plain, specific language. “We know that if they make images once, they are preparing,” he said. “If they make images a second time, it’s like a simulation. The third time it means that in one or two days, they will attack.”

That’s a targeting cadence, described by the one country on Earth with four years of firsthand experience reading it, handed to American journalists in real time.

By July, months after Prince Sultan, Zelensky was posting again, this time far more specifically. “Since the beginning of July, we have recorded active Russian satellite surveillance of the Gulf states and US military facilities located there,” he wrote. “These images subsequently appear in Iran... there is a clear correlation between Russia’s satellite imagery of these sites and Iranian strikes, both before the attacks, in preparation for them, and afterward, to assess the damage inflicted.”

He said explicitly that he intended to share evidence directly with the United States. That’s the exact same battle-damage-assessment loop CNN would later independently confirm, described by Ukraine two months before it happened again, to a different base, with three more Americans dead.

So, we’re not speculating about whether Ukraine tried to warn Washington. We know it did, repeatedly, publicly, with enough technical specificity to be useful.

So why the hell didn’t we listen?

Trump had already, on the record, told multiple outlets that the United States did not need Ukraine’s assistance in its war with Iran.

That’s a direct position taken by the president himself, offered while Zelensky was actively trying to hand over the kind of warning that might have saved an AWACS or a radar or soldiers’ lives.

A year earlier, in March 2025, the Trump administration had briefly cut off intelligence-sharing with Ukraine entirely. When a reporter asked Trump directly whether Putin was exploiting that pause to intensify strikes on Ukraine, his answer was almost breezy. “I think he’s doing what anybody else would.”

Whatever you take from that exchange, it establishes a documented pattern. This administration has, at multiple points, treated the intelligence relationship with Ukraine as leverage or as an afterthought, rather than as a serious channel worth protecting and using.

A president publicly stating he doesn’t need a country’s help isn’t the ideal precondition for that country’s intelligence getting taken seriously inside the interagency process, regardless of how good the underlying tradecraft is.

Intelligence doesn’t arrive stamped with a neutral, unimpeachable label. It arrives from a source, and if the political relationship with that source has soured, the information gets filtered through that souring whether anyone admits it or not.

That’s the “beautiful reports” dynamic I wrote about with Putin and his own generals, except run in reverse.

Instead of subordinates shading the truth to please a leader, you get a leader who’s already decided a source isn’t worth fully trusting, so accurate information from that source struggles to travel upward with the urgency it deserves.

Compartmentalization is worth mentioning too.

American intelligence runs on classification stovepipes, and a warning about Russian satellite tasking against a Saudi-hosted base sits at an awkward intersection of at least three different agencies’ equities like foreign intelligence liaison, Gulf regional security, and tracking Russia’s own space assets.

Getting a fast-moving, foreign-sourced warning through that structure in time to matter operationally is a hard problem even when everyone involved is acting in good faith, and it becomes a much harder problem when Trump dislikes Zelensky simply because Ukraine reminds him that he was once impeached over his dealings with the country.

China may now be filling the same role

Finally, some extra-shitty news just for funsies…

On September 11, the Wall Street Journal reported that US officials have linked high-resolution satellite imagery supplied to Iran by Chinese entities to the July 17 strike on Muwaffaq Salti Air Base in Jordan, an attack that killed three American soldiers, Army First Lieutenant Tyler James Feehan, Private Isabella Gonzales, and Sergeant Angel S., and wounded four more.

Iran reportedly obtained high-resolution images of the base both before and after that strike, the same before-and-after pattern the satellite kill chain runs on.

US officials haven’t publicly identified the Chinese entities involved, and they haven’t accused Beijing’s government of directing or ordering that support.

Not surprisingly, Beijing has denied the allegations, with China’s Foreign Ministry accusing unnamed parties of trying to drag the country into the conflict, and a Chinese Embassy spokesperson calling the claims a malicious smear.

And this isn’t the first American loss tied to intelligence at that particular base. BBC Verify separately confirmed, through its own satellite analysis, that an American AN/TPY-2 radar, a core THAAD component, was destroyed at the same Jordanian installation back in March.

Asked about the Chinese imagery report aboard Air Force One, Trump waved it off again. “They basically do what we do,” he told reporters. “When they say China spies on us, I say you’re right and we spy on them too.”

We don’t know what Russia transferred to Iran, whether it was raw satellite imagery, fully processed intelligence products, specific coordinates, complete target folders, or direct human analytical support walking Iranian planners through the process.

We don’t know who inside Russia’s government or military actually authorized each individual sharing decision.

We don’t know exactly how much Russian assistance improved Iranian strike accuracy compared to what Iran’s own indigenous reconnaissance could already manage on its own.

But I do know that Iran’s targeting was never good enough alone to hit an AWACS in the rotodome, a nine-meter circle, from 500 km away… Unless a hell of a lot of luck was involved.

Iran already knew how to build missiles long before any of this. It already knew how to build Shahed-derived drones. The hard problem was knowing where the valuable target would be sitting the moment that weapon finally arrived.

Russia helped solve that problem for Iran.

The strategic lesson here should sound familiar to anyone who’s read the smut I write for a while now: Modern combat is decided by who can close the kill chain first, from detection to strike to confirmed damage.

Apparently, our adversaries still believe in the power of “allies” as combat enhancers, at a time when the US is trying to ditch ours as quickly as possible.

Слава Україні!

Hey friends.

Somewhere in a Chinese defense lab, a team of engineers looked at the modern battle tank, arguably one of the most mature, over-engineered, thoroughly optimized weapons categories humans have ever built, and decided it needed a nuclear reactor and a rail gun.

This is actually one of the more interesting stories in miltech this week.

(Although, I did see a story that Polish military forensic officials were having difficulty scanning a Russian Gerbera drone with X-rays. That’s probably worth a look. I’ll dig into that later this week if nothing else hits the news before then.)

Besides, I don’t talk about China near as much as I should. I just saw a Chinese product that uses Lidar to find the signature of mosquitos flying around your house, then uses lasers to shoot them out of the air.

So, like a miniature directed-energy air defense system for your home. It costs $5,000.

My point is, if we sleep on China, one day we’re going to wake up and have been outengineered.

Okay, so what about this nuclear tank?

This is a peer-reviewed concept study, published in June in the Chinese journal Acta Armamentarii, written by researchers from Inner Mongolia First Machinery Group, a real state-linked defense contractor (similar to General Dynamics) with actual decades of tank-building experience, alongside academics from the Beijing Institute of Technology.

Nobody is bending metal on this thing yet.

But the concept itself is worth looking at, because of what it says about where armored warfare might be headed, and because China has already quietly developed several of the individual pieces this design depends on.

Wait Wes, aren’t tanks on the modern battlefield dead? With drones and whatnot?

Well, the Chinese engineers have a plan for that too: Electromagnetic armor.

What they’re proposing

Here’s the core idea: A 60-ton armored vehicle, roughly in line with a modern main battle tank’s weight class, powered by a 10-megawatt-class small modular reactor and armed with a 50-megajoule electromagnetic railgun.

Chinese researchers estimate that railgun could hurl a projectile to roughly 3.5 kilometers per second, giving it a theoretical strike range somewhere around 280 miles, closer to an artillery rocket than anything currently mounted on a tank turret.

For scale, a modern tank’s main gun typically reaches out a few miles, maybe low double digits under ideal conditions with a guided round.

This concept is describing a tank that could theoretically hit a target the distance from Boston to Washington, D.C. Whether that ever becomes militarily useful is a separate question we’ll get to.

First, let’s talk about how you’d even build the thing.

Here’s the part I appreciate about this paper… The researchers have laid out a very plausible three-stage development path, each stage building the electrical infrastructure the next one will need.

Read more

There are a lot of different ways for a writer to find their “writer’s voice.” On Friday September 18, I’ll be hosting a session at Medium Day 2026 about how I found mine, the journey that got me there, and six questions to ask yourself… (a ‘voice audit’ of sorts), that will help you dial it in.

I’m not a coach.

I’m not a guru.

I will never try to sell you courses.

I typically never write about writing. But I think this is a good opportunity for aspiring writers to learn how their human “voice” can really help their work stand out in the age of slop.

I’ve been writing professionally (for money), since 2015, and on Medium since November of 2018. From a purely technical perspective, my writing back then was on point; I knew how to structure an argument, I could write clean sentences, and I could sound polished, measured, and credible.

My mentor was the incomparable writing instructor Shani Raja, former editor for Wall Street Journal. He now publishes courses on Udemy. (highly recommend)

But at the time, I also sounded like every other competent person with a keyboard and a Mocha Cloud Macchiato from Star-bizzle: Just fine.

After Russia’s full-scale invasion of Ukraine in 2022, that changed.

I stopped writing like someone trying to prove he could write, and I started writing in a conversational voice. That shift took me from a few hundred followers to hundreds of thousands across several platforms.

And it all started at Medium. This Friday, I want to talk about what changed for me.

Medium Day is Medium’s free, live, virtual gathering for writers, readers, and people who still believe the internet can occasionally be used for something better than doomscrolling, the inexplicable antagonization of Canada, and Google Gemini summaries.

This year, it runs on Friday, September 18, from 8 a.m. to 8 p.m. EDT, entirely on Zoom.

It’s FREE

You’ll need Zoom, but you don’t need a paid Zoom account. The free version works fine.

Wait Wes, why should I listen to you?

By 2015, I was trying to take writing seriously. I had opinions, experience, and a dangerous amount of [over]confidence. What I didn’t have yet was a fully formed voice.

I was still writing toward credibility, which is different from writing from it.

When I started writing on Medium in 2018, I treated it like a proving ground. I wanted the work to be smart. I wanted it to be clean. I wanted it to be taken seriously.

That helped me become a better writer, sure. But it also taught me the limits of polish. Polish can remove mistakes. It can also remove your fingerprints.

Back in those days, Medium used to have roving editors stalking the platform like a pack of gray wolves and would occasionally ask to republish one of my better articles to larger in-house publications like GEN, OneZero, or An Injustice.

They would offer around $200 for this which, at the time, was the best feeling in the world.

I was big into cybersecurity writing back then and one of my early republished articles was a piece where I interviewed NASA to see if it were possible to hack into a spacesuit. NASA wisely declined to comment but several cybersecurity firms were confident they could do it.

Ugh, I’m getting off topic.

For the most part, that early work was bland with only rare glimpses of my personality coming through to the page. I should have picked up on it at the time, but looking back now, I see that it was only the articles with my real voice peeking through miniblinds that ever got picked up by Medium’s editors.

Take this completely respectable sentence from a blog post a VPN provider paid me real money to write back in 2020:

“For at least 12,000 years, humans have been living alongside cannabis. From the invention of agriculture thousands of years ago to today’s technological innovations, technology and cannabis are synergistic.”

Oh. My. God. I actually used the word ‘synergistic.’

Then February 2022 arrived.

I had been covering Ukraine before the full-scale invasion. In 2019, I interviewed five Ukrainian soldiers who were already fighting Russians in the Donbas. At the time, I called it “Europe’s forgotten war.”

I have close friends in the military who had traveled to Ukraine between Russia’s invasion of Crimea and the Donbas in 2014 and the full invasion in 2022 to train the Ukrainians in modern Western warfighting doctrine.

So, when Russia widened the war in 2022, I had faces, voices, and lives in my head.

Then the Russian war crimes started appearing in public view.

That hit something very specific in me. I’m a veteran. I served as a US Army infantryman and later in the US Air Force. Watching a larger military brutalize civilians because its leader believed history owed him an empire offended every American soldier instinct I had.

Of course, I know the US military isn’t perfect. But I also know from first-hand experience the steps that the US military takes (or at least used to) to avoid civilian casualties.

So, I wrote angry…

Here’s an example from 2022:

“President Putin has made a fatal mistake: he believed his own hype and drank his own vodka-spiked Kool-Aid. Now, his invasion is producing the exact opposite result from what he intended. Europe is unifying. Ukraine is holding. And the shirtless equine aficionado is about to learn what happens when his corrupt army of grifters and malingerers goes up against Ukrainian ingenuity.”

That’s a hell of a lot more interesting to me than the cannibis piece.

Some of those early pieces after the invasion were close to rants.

But they weren’t empty rants… They were rants with reporting, military context, legal caution, and enough tactical detail to pick apart the invasion force piece by piece.

That’s when I started to understand something about voice:

Emotion isn’t the enemy of serious writing.

Sure, untested, raw emotion can make writing sloppy. But disciplined emotion can make writing clear. Things like anger, grief, awe, disgust, hope, and even sarcasm can all tell the reader why the subject matters, as long as the writer still respects evidence.

For me, anger opened the door. Expertise kept the piece standing.

My work became sharper, more direct, more conversational, more trusted, and more recognizable. People knew they were reading a Wes O’Donnell piece without looking at the byline.

I doubled down on my knowledge of military technology and carried that voice from Medium into Substack and YouTube.

And the strange part is that once I stopped trying to sound like a writer, the more people started treating me like one.

So why attend Medium Day? Well, besides all of the other brilliant people who will be offering their insights for free, my session is more than just “come hear my story, bro.” It’s a Craft session. People will leave with something they can use.

Title of my talk:

I stopped writing like a writer. Then people started reading.

Core lesson:

Technical skill makes writing clean. Voice makes writing trustworthy.

In my 30-minute session, I’ll talk about the gap between technically good writing and writing that sounds unmistakably like you. I’ll use my own before-and-after as the case study, then walk attendees through a six-question voice audit they can use on their own drafts.

The goal is simple: find the sentences that are clean, polished, and forgettable, then revise them until they carry judgment, lived experience, and a real human pulse.

This session is for writers who know their sentences are fine and suspect that ‘fine’ is the problem. It’s for anyone who has ever reread a draft and thought, yes, this is clear, but I could’ve been replaced by AI.

I don’t think every writer needs a war to find their voice. That would be a terrible professional development plan.

But I do think every writer eventually faces the same choice: You can keep writing technically competent sentences that prove you know how writing works (in an age of AI where synthetic writing is getting more proficient, no less).

Or… you can let readers see how you think on the page.

That’s what I’ll be teaching at Medium Day. I hope to see you there.

Register at MediumDay.com. My particular talk is from 4:00–4:30 PM EDT on Friday, September 18, but be sure to check out all the amazing topics from other speakers.

Remember, you have life experiences that are completely unique to you. People want to read that personality on the page.

Stay frosty.

Wes

Read more

Hey friends,

When the Russian military first crossed into Ukraine in 2022, the utter lack of discipline and functioning logistics, combined with shoddy maintenance and widespread corruption surprised even some of the more seasoned intelligence professionals in Washington.

Hell, it surprised me too. At the time, I was already thinking ahead to how Ukraine could run a successful insurgency against the occupying Russian forces.

But after a few weeks, it became clear that Ukraine’s extraordinary will to fight combined with Russia’s broken military kept Russia pinned to the south and east.

On March 6, 2022, Andrei Kozyrev, Foreign Minister of Russia from 1990 to 1996, said, “The Kremlin spent the last 20 years trying to modernize its military. Much of that budget was stolen and spent on mega-yachts in Cyprus. But as a military advisor you cannot report that to the President. So, they reported lies to him instead. Potemkin military.”

Basically, endemic corruption within the defense sector meant that billions of dollars earmarked for military modernization were instead siphoned off by oligarchs and high-ranking military leaders. But Putin was told everything was going great; and he believed it.

That pattern, of being afraid to tell the boss the truth, is still going on today; only now, these lies are making their way straight to Trump and affecting ongoing negotiations.

Right now, it looks something like this:

Some Russian unit reports that it has advanced.

Perhaps it seized a treeline.

Maybe scouts entered the outskirts of a settlement.

Or nothing significant happened at all.

The report goes upward.

The battalion improves it with beautiful prose that would make Alexander Pushkin himself weep.

The brigade reports substantial progress.

Eventually headquarters says the settlement has effectively fallen.

Then Vladimir Putin announces it, speaks to Trump on the phone, and Trump believes it.

That’s essentially what appears to have happened around Svyatohirsk.

On September 1, Putin claimed Russian forces had captured Svyatohirsk in Donetsk Oblast. Available evidence didn’t support the claim. On September 7, Ukrainian Third Army Corps commander Andriy Biletskyi filmed himself standing openly in the town and described it as comfortably inside Ukraine’s rear area.

He said “I want to appeal to the Russian military and political leadership. Take off your clown noses and colorful wigs. Don’t look like fools. The 60th Mechanized Brigade of the 3rd Army Corps is in full control of the city. This is actually our rear area.”

Even Kremlin-aligned Russian milbloggers acknowledged that the Ukrainian footage demonstrated that Russia didn’t control Svyatohirsk. ISW said the nearest confirmed Russian troops had previously been roughly 13 kilometers away, and more recent Ukrainian counterattacks may have pushed them farther back.

Two days later, Russian milbloggers were openly complaining that the Lyman sector had become distorted by what they call “beautiful reports,” flattering battlefield assessments that exaggerated Russian advances.

Read more

Hey friends.

Last week, I started working on a YUGE explainer of Ukraine’s veritable zoo of an air defense stack; the different systems, their ideal targets, how each system finds and tracks targets… the whole shebang.

I had just finished when Ukraine revealed the Koral yesterday! So, I was able to insert that system here at the last minute and actually use that news to kick off the whole piece.

On September 8, at Poland’s MSPO defense exhibition, Ukraine’s state defense conglomerate Ukroboronprom put a photograph on a screen that most of us scrolled past without a second thought: A missile launching from a boxy self-propelled system riding on a Czech Tatra truck chassis.

That’s Koral, Ukraine’s homegrown surface-to-air missile, and the launcher had never been shown publicly before.

Koral’s guidance system combines three completely different methods of finding its target: Inertial navigation, radio correction, and active radar homing. So, why in God’s name does one missile need three different ways to find the same object in the sky? Because each one solves a different piece of a very hard problem.

Let’s build this thing from the ground up.

By the end, you’ll understand not just what Koral is, but why Ukraine operates roughly two dozen fundamentally different ways to kill something in its own sky.

How a missile finds anything at all

Before strolling through the zoo, let me explain the three-part dance almost every guided weapon performs, because Koral’s specific combination is really just the modern standard dressed up in Ukrainian components.

Start with inertial navigation, or INS.

Picture a tiny set of gyroscopes and accelerometers riding inside the missile, constantly measuring how the missile itself is accelerating and rotating after launch.

It doesn’t need GPS, radar, and critically, it isn’t looking at the target at all. The missile launches toward a calculated intercept point in the sky, a spot where the fire-control computer predicted the target would be and then flies there using nothing but its own sense of motion, the same basic principle a submarine uses to navigate underwater without ever surfacing to check a star.

That works fine for the first stretch of an engagement. The problem is that targets have the extremely rude habit of moving after you’ve already committed to a flight path; plus tiny positional errors accumulate the longer the missile flies blind.

So Koral needs a second layer. That’s radio correction, AKA the midcourse update.

A ground radar keeps tracking the real target the entire time the missile is in flight. The fire-control system recalculates a fresh predicted intercept point as the target maneuvers, and beams that update to the missile over a radio datalink.

Then the missile adjusts its course accordingly.

I want to be detailed about what’s happening here, because it’s easy to misread:

The missile isn’t searching for anything during this phase. Someone else already found the target. The missile is simply receiving updated directions, the way an Amazon delivery driver gets a text saying the customer actually wants this delivered on the back porch, while the driver was already en route.

AMRAAM works on this exact principle during long-range engagements, flying inertially at first while soaking up datalink updates the whole way.

Then, somewhere near the endgame, the third layer wakes up.

Active radar homing means the missile itself starts transmitting radar energy from its own nose, listening for the reflections bouncing back, and using that information to calculate the target’s exact range, bearing, and closing geometry, then steering itself into the intercept without anyone’s further help.

That’s what “active” actually means in the phrase ‘active radar seeker.’ (As opposed to semi-active. Hang with me. We’ll get there).

The radar transmitter rides inside the missile, not on the ground. AMRAAM uses this. So does the French Aster 30. So does the terminal phase of Patriot’s PAC-3 MSE, running an active Ka-band seeker in the final seconds of flight.

Put all three stages together and you get the full Koral engagement sequence:

Ground radar detects the target, then

Fire control computes an intercept, then

Koral launches, then

Inertial navigation flies it toward the intercept area, then

Radio updates correct the course along the way (midcourse guidance), then

The active radar seeker finally wakes up, acquires the target, and takes over for terminal homing, then

Intercept, then

Boom!

The six ways a missile can actually be told where to go

Active radar homing is only one branch of a much bigger family tree.

Semi-active radar homing is the older cousin of active radar.

Here, the missile carries a receiver but no transmitter of its own capable of illuminating the target. A ground radar shines energy directly at the target instead, and the missile simply follows the reflected signal home, like a moth chasing a beam of light somebody else is holding steady.

HAWK works this way. So does the legacy Soviet Buk 9M38 family during its terminal phase.

It’s a simpler, cheaper missile to build, but it comes with an obvious weak point. Somebody has to keep that ground radar shining on the target the entire time, and killing or jamming that radar breaks the whole engagement.

Track-via-missile is a clever middle path, and Patriot’s older PAC-2 GEM-T uses this architecture. The missile catches radar reflections bouncing off the target and relays what it’s seeing back down to the ground station, which does the heavy computational lifting and radios steering commands back up to the missile.

The missile is functioning partly as a remote sensor while the expensive brain stays safely on the ground.

Notice something important here… That’s a completely different guidance philosophy running inside the exact same Patriot battery that fires PAC-3 MSE, which hits its target through pure kinetic collision rather than a blast-fragmentation warhead.

Same launcher. Two entirely different weapons underneath the same delicious brand name. Also, it’s part of a balanced breakfast…

Command guidance strips the seeker out of the missile entirely. A ground system tracks both the missile and the target simultaneously, works out the geometry, and continuously radios steering commands the missile simply obeys, no thinking required on the missile’s part.

Crotale, which I’ve written about before, uses exactly this command-to-line-of-sight approach, and plenty of older Soviet systems lean on it too.

The advantage is a cheap, dumb missile, since all the expensive intelligence lives on the ground rather than getting thrown away every time you fire a round.

Infrared homing turns radar off completely and looks for heat instead.

Stinger works this way. So does the IRIS-T family, and the modern SLM variant is particularly interesting because it blends in-flight datalink updates with a somewhat sophisticated imaging infrared seeker for the terminal phase, one that builds something closer to a real thermal picture than a crude heat blob, letting it pick a real aircraft out of flares and background clutter far more intelligently.

No radar emissions from the seeker at all. No need to illuminate anything. Jamming an enemy’s radar does nothing to an infrared seeker.

And then there’s Starstreak, one of my faves, because apparently British engineers looked at every existing guidance method and decided none of them has a cool enough name.

Starstreak uses semi-automatic laser beam riding. The operator keeps a sight fixed on the target, the launcher projects guidance information through a laser field, and three separate darts ride that beam toward the target at speeds beyond Mach 3.

No radar seeker to jam.

No infrared seeker to fool with a flare.

The tradeoff is that a human has to keep the sight steady on the target for the whole engagement, which is a very different kind of vulnerability than an electronic one.

Okay, friendos, hold that whole taxonomy in your head, because it’s the key that unlocks everything below.

The Ukrainian air defense stack:

Layer one: where Patriot earns its terrifying price tag

Let’s start with the hardest problem Ukraine faces. Russian Iskander-M ballistic missiles and air-launched Kinzhals come in fast, at speeds and angles nothing cheap can touch, and you cannot send a machine-gun truck after one.

This is where Patriot lives, and specifically PAC-3 MSE, the missile I’ve written about extensively as Ukraine’s supply of it keeps dwindling toward crisis levels. Ground radar detects and tracks. Fire control predicts an intercept point. The missile launches, flies inertially at first, gets corrected over an RF datalink, and near the intercept point its active Ka-band seeker wakes up and locks onto the actual incoming missile.

Then, unlike almost every other weapon on this list, it doesn’t explode near the threat. It physically rams into it, body against body, what Lockheed calls hit-to-kill. For a ballistic missile potentially carrying a large warhead, smashing the incoming object directly beats spraying fragments nearby and hoping.

Here’s the twist:

The same Patriot battery can also fire PAC-2 GEM-T, using entirely different track-via-missile guidance and a traditional blast-fragmentation warhead instead of a kinetic hit.

I wouldn’t call that a downgrade, it’s a different tool sitting in the same toolbox, useful against aircraft and cruise missiles while carrying some capability against certain ballistic threats too.

Even a single Patriot system, in other words, secretly contains two separate air-defense philosophies wearing the same uniform.

Honorable mention: FREYJA

FREYJA is being designed specifically around the ballistic-missile problem. Fire Point says the FP-7.X is intended to intercept ballistic missiles below roughly 25 km altitude, with an engagement radius around 80 km in the current development concept. The missile has already undergone several flight tests, with Fire Point saying the current version reaches about 1,800 m/s, roughly Mach 5.

The goal is eventually around 2,200 m/s, roughly Mach 6.5.

But, FREYJA is still a program; not operational.

Layer two: SAMP/T, Europe’s other answer to a screaming missile

Aster 30, the missile running inside France and Italy’s SAMP/T system, gives us almost a mirror image of the Koral guidance chain, which is exactly why Koral is worth paying attention to.

Aster launches vertically, rides inertial guidance through the midcourse phase, receives refreshed target data over a datalink, and then hands off to an active electromagnetic seeker for terminal homing.

What makes Aster sexy is a system called PIF-PAF, [Pilotage en Force (pilotage / control by force) and Pilotage Aérodynamique Fort (strong aerodynamic control)] combining conventional aerodynamic fins with small lateral thrusters that let the missile make brutally abrupt last-second course corrections, closer to Patriot’s attitude-control philosophy than to a normal missile’s fins alone.

Unlike PAC-3, though, Aster still uses a proximity fuse and a fragmentation warhead rather than a direct hit.

Ukraine is already flying SAMP/T, and as I’ve reported, it’s slated to become the first combat operator of the newer SAMP/T NG variant, with licensed Aster 30 production coming to Ukrainian soil.

That gives you a clean comparison worth remembering. PAC-3 MSE hits the threat directly. Aster 30 maneuvers close enough that a precisely timed fragmentation burst does the job instead.

Both demand extraordinarily accurate terminal guidance to pull off either trick.

Layer three: the Soviet heavyweights Ukraine inherited

Ukraine walked into this war already carrying a large Soviet-era air-defense architecture built around systems like the S-300 and Buk. There’s no single “S-300 missile,” there are multiple generations running different missiles with different guidance arrangements entirely, so let’s keep the S-300 focused on its operational role rather than its internals.

Long-range area defense. Denying Russian aircraft the sky. A serious deterrent against cruise missiles. It’s a big part of why Russian pilots remain so reluctant to fly deep over Ukrainian-held territory even now.

Buk is easier to explain, and Ukraine’s Ground Forces still publicly list Buk-M1 among their active systems. The 9M38 missile family flies inertially at first, then switches to semi-active radar homing for the terminal phase, meaning a ground radar has to keep illuminating the target the entire way in.

And that requirement set up one of the most Ukrainian moments of this entire war: FrankenSAM.

Layer four: FrankenSAM, because adapters apparently work on missiles too

Eventually Ukraine started running low on Soviet-standard Buk missiles, and somebody asked a question that could only come from a country fighting for its survival with whatever was lying around.

Could we bolt an American naval missile onto a Soviet launcher instead?

Ukraine and the United States modified Buk-M1 systems to fire RIM-7 Sea Sparrow missiles, and Ukraine’s own Ministry of Defense openly identifies this as FrankenSAM. The RIM-7 itself is semi-active radar homing, chasing continuous-wave or pulse-Doppler energy reflected off the target, with later variants able to pick up midcourse datalink updates too.

This is the moment Ukraine stopped simply accumulating disconnected weapons and started actively breaking down the walls between them.

Soviet launcher. Western missile. Modified fire control. Ukrainian engineering gluing the whole thing together. Keep that theme in your back pocket, because it’s about to show up again, twice.

Layer five: NASAMS, the battery that isn’t really one battery

NASAMS flips the traditional monolithic SAM battery inside out.

A typical fire unit combines an AN/MPQ-64 Sentinel radar, a Fire Distribution Center, electro-optical and infrared sensors, and a cluster of physically separate launchers, all tied together over a network rather than bolted onto one vehicle.

Its primary missile, AIM-120 AMRAAM, originally built for fighter aircraft, runs the same three-stage philosophy Koral does: Inertial guidance, datalink updates, then an active radar seeker takes over near the target.

NASAMS can also fire AIM-9X depending on the configuration, which means the launcher itself no longer really defines the weapon… Software and networking do.

Layer six: IRIS-T, the missile that skips radar for the kill

Ukraine runs both the longer-range IRIS-T SLM, reaching roughly 40 kilometers, and the shorter SLS variant out to around 12, with nine systems reportedly guarding Ukrainian skies by the end of 2025.

The interesting part is the seeker. A traditional heat-seeking missile just hunts for a hot blob. IRIS-T’s modern imaging infrared seeker builds something much closer to an actual infrared picture, letting it discriminate the real target from clutter and countermeasures with real intelligence rather than chasing whatever’s hottest.

That gives Ukraine a way to kill things without ever depending on terminal radar homing, and on a battlefield this heavily jammed, that redundancy is worth its weight in gold.

Layer seven: HAWK, the Eisenhower-era system that refuses to retire

MIM-23 HAWK entered American service under Dwight Eisenhower and is still shooting at Russian weapons in 2026.

Ukraine’s authorities continue listing it among the Western systems in active service. Its missile runs classic semi-active radar homing, riding the reflected energy of a ground radar illuminating the target the entire way in, which makes it the perfect direct comparison to Koral.

HAWK needs an outside radar shining on the target through the whole endgame. Koral’s active seeker is designed to eventually find and lock the target entirely on its own. Same basic problem both weapons are solving.

Sixty-plus years of electronics separate the two solutions.

Layer eight: the command-guided family

Let’s group a handful of systems together by philosophy rather than giving each their own section otherwise this piece would be 10k words and I have videos to edit.

Crotale, which I’ve covered in detail before, uses command-to-line-of-sight guidance, with the fire-control system tracking both missile and target and radioing continuous steering corrections.

Ukraine’s Ground Forces also list Osa-AKM, Strela-10, and Tunguska.

Osa runs a radar-centric short-range approach with radio-command missile guidance. Strela-10 leans on a passive electro-optical and infrared family depending on the specific missile variant loaded. Tunguska combines radar and command-guided missiles with a pair of 30mm cannons bolted right onto the same vehicle.

All three were built to move with ground forces rather than defend a fixed strategic asset. A Patriot battery protects a huge swath of territory. A Tunguska protects the specific battalion currently driving down a muddy road.

Layer nine: Starstreak, the British system nobody else copied

Britain’s Stormer vehicle carries Starstreak HVM, and the whole concept is delightfully strange once you understand it. As I mentioned above, an operator keeps a sight tracking the target while a laser guidance field gets projected toward it.

The missile accelerates past Mach 3, then three separate darts split off and ride that laser beam the rest of the way in.

Read more

Think of this as your weekly strategic weather report exclusively for paid subscribers. Five things to watch and what could break next in war, defense tech, and geopolitics. Just the pressure points most likely to shape the next seven days.

I try to keep this a no-nonsense, high-level brief to start your week… But snark occasionally creeps in.

Happy Labor Day in the States. I’m still working today (off and on), but for those enjoying the symbolic end of summer, enjoy your day off. Personally, I’m looking forward to the cooler Fall weather. Thanks to El Nino, winter may skip Michigan altogether this year which means my Nissan Zed doesn’t have to spend four months garaged.

Source links in the text. Let’s get to it:

BLUF:

Iran is building a toll booth in the Strait of Hormuz and calling it a shipping corridor.

North Korea commissioned a nuclear destroyer the day before trilateral exercises began and gave an eight-month warning.

The AfD got 43.8% in Saxony-Anhalt and the CDU lost two-thirds of its seats.

Chinese state enterprises are supplying the carbon fiber for Russian drones.

And US envoys went to Kyiv for the first time.

Stay sharp.

1. Iran’s New Hormuz Corridor Plan Is a Toll Booth

Iran is set to unveil a new restricted maritime zone in the Gulf alongside maps for a new shipping corridor through the Strait of Hormuz, with Mohsen Rezaei, secretary of Iran’s Supreme National Security Council, confirming the announcement is days away.

Read more

Oj då!

Hey friends.

Europe can build spectacular jet engines:

Rolls-Royce builds the EJ200 that powers the Typhoon.

Safran builds the M88 that powers the Rafale.

GKN Aerospace helps manufacture components flying inside some of the most advanced military aircraft on the continent.

But none of that expertise, it turns out, was solving the problem that 2026 may eventually become known for: we need more disposable jet engines!

In April, reporters interviewed more than a dozen weapons experts, company executives, and government officials about Europe’s supply of a much less glamorous piece of hardware: the tiny turbojet engines that power expendable long-range attack drones.

The consensus was consistent.

A Ukrainian defense-industry source told reporters the shortage was probably the main factor limiting Ukraine’s missile-drone production. Fabian Hoffmann, a senior research fellow at the Norwegian Defense University College and someone I’ve quoted numerous times, said, “Europe has a massive bottleneck in mini-jet engine production,” adding that solving it mattered for reducing the continent’s reliance on the United States too.

It’s not Europe’s fault. Before 2022, nobody on earth needed tens of thousands of disposable jet engines.

The big aerospace primes chased lucrative aircraft propulsion contracts while a tiny handful of specialist firms serviced whatever small market existed for miniature turbojets.

Then a war broke out that suddenly wanted thousands of them a year, and Europe discovered it had maybe three companies capable of building them at any real scale.

Four months later, on August 21, GKN Aerospace showed up at Malmen Air Base in Sweden with something unusual. A full-scale mockup of a long-range unmanned aircraft demonstrator, and sitting right beside it, a completely new 3D-printed turbojet engine, built from scratch, purpose-designed to power it.

Both had gone from a blank sheet of paper to physical hardware in roughly nine months.

Sweden’s answer to the disposable turbojet shortage was characteristically Swedish: “Fine. We’ll make one ourselves.”

Read more

Hey friends.

Let me be up front here: I’m a Skyranger stan.

Read more

Hey friends,

Russia spent four years jamming, jailing, and complaining about Starlink. Then in February, it lost access to it, and discovered it couldn’t fight without it. If you remember, this was the turning point earlier this year where it finally felt like Ukraine has the upper hand.

Now Moscow is trying to build its own Starlink and the first satellite already fell out of the sky.

The day Russia lost its own warfighting network

SpaceX doesn’t operate in Russia. Never did.

What Russian units did instead was quietly acquire Starlink terminals through intermediaries and third countries and fold them straight into frontline operations, eventually mounting some on long-range attack drones so operators could steer them by video feed in real time instead of trusting a preprogrammed flight path.

At the time, several outlets reported that Russian forces had been running thousands of these unauthorized connections.

Ukraine noticed, built a registration system through Kyiv’s Diia app to whitelist its own legitimate terminals, and worked with SpaceX to shut off everything that wasn’t on the list.

On February 5, Elon Musk confirmed the unauthorized access had been cut. What happened next was comical.

Serhii “Flash” Beskrestnov, now electronic-warfare adviser to Zelensky, said, “All command of the troops has collapsed. Assault operations have been stopped in many areas.”

Three Ukrainian frontline sources told Western outlets the disruption looked significant. Russian milbloggers, who rarely miss a chance to complain, spent that week complaining about exactly this.

Now, “collapse” is a bit of strong word here… Russia still had VHF and HF radios, fiber lines, mesh networks, couriers, the whole analog toolkit a military has been falling back on since before satellites existed.

Russian Deputy Defense Minister Aleksei Krivoruchko told state television two weeks later that the shutdown hadn’t affected drone effectiveness at all, but that’s also what you expect a Russian deputy defense minister to say regardless of what was actually happening on the ground.

What’s not in dispute is the scale of Russia’s new problem.

A voice radio can tell a commander that somebody spotted a Ukrainian vehicle. Starlink could send the drone footage, the coordinates, the map update, and the targeting handoff between a reconnaissance unit and a strike unit, all at once, with low enough latency that the video was still useful by the time somebody looked at it.

Russia lost the bandwidth that let it fight the way a modern, drone-saturated army needs to fight. Beskrestnov later told Defense Express that the shutdown quietly wrecked Russia’s plans to expand its own Black Sea drone fleet in 2026, because it turned out Starlink was the only communications link that actually worked for naval drones operating off the Ukrainian coast.

Whoopsie. The lesson here? Don’t rely on your fucking adversary’s communications infrastructure.

Every domestic alternative Russian engineers tried either failed outright or proved too unreliable to trust with an attack boat.

Meet Rassvet

Russia’s answer to Starlink is a company called Bureau 1440, employing several thousand engineers, and building a constellation named Rassvet, “daybreak,” after the number of orbits Sputnik made on its first flight.

It’s actually an ambitious system. Bureau 1440 advertises an 800-kilometer operating altitude, gigabit-class speeds, sub-70-millisecond latency, laser links between satellites, and direct 5G connections to ordinary handsets, essentially each satellite acting as a cell tower in orbit.

The company’s long-term goal is somewhere north of 900 satellites by 2035, with an initial operational threshold of a few hundred targeted for next year.

Read more

I break down why American intelligence believes Russia may be considering a limited, deniable test of NATO’s Baltic flank this fall, why sending the CIA director in person carries a different weight than a phone call, and what to actually watch for next.

VIDEO TRANSCRIPT:

Three weeks ago, I told you American intelligence had changed its assessment of Vladimir Putin. The new concern was that Russia might test NATO as early as this fall, probably through something limited, ambiguous, and deliberately difficult for the alliance to answer cleanly.

This week, CIA Director John Ratcliffe boarded an American military aircraft, flew through Latvia into Moscow, met Russia’s foreign intelligence chief, and multiple US news organizations now report his message was very simple.

Do not attack NATO.

Holy shit. How did we get here?

Before Ratcliffe arrived, Washington reportedly told Ukraine a senior American delegation would be inside Russia and asked Kyiv to pause strikes on Moscow, St. Petersburg, and other northern targets. Ukraine complied. Ratcliffe never met Putin personally, though the Kremlin says Putin was briefed immediately afterward. Trump calls the visit “semi-routine.”

A warning doesn’t mean an attack is imminent. It means American intelligence appears concerned enough that the CIA director physically flew into Moscow to deliver a message the Russians could have received over the telephone.

Why wasn’t the telephone enough?

Hey friends, Wes O’Donnell here. Defense journalist, Army and Air Force veteran. Let me ask you a question…

Knowing what you know about President Donald J. Trump’s transactional nature and let’s say, distaste for alliances, do you think he would commit US troops to fight and die for Latvia in the event of a Russian incursion?

No? Me neither…

I think he would ask the question, “What’s in it for the United States?”

Then, after some bipartisan uproar coming from Congress demanding we honor our alliances, I think he would ask, “But why send US soldiers to fight and die for a small country on NATO’s frontier?”

Then, after having a nightmare involving the ghost of Lindsey Graham and the realization that several of the fourteen McDonalds restaurants in Latvia are under threat, he might still resist and ask, “Why risk a nuclear war with Russia for a country with fewer people than Detroit, Michigan?”

Unless… (ahem) Latvia had some sort of mineral rights they would like to negotiate?

That’s right, the United States has now become a caricature of the Italian mafia where our so-called leaders say, in a bad New York accent, “Hey, uh, maybe you sell us some rare earths at a discount for some of the Don’s protection, huh? Hey, fat Tony, fuhgeddaboudit, bada bing.”

Ratcliffe left Washington on a US government aircraft, stopped in Riga, and flew into Moscow aboard an Air Force C-17 on August 25th. The trip was unannounced. The Kremlin confirmed the meeting; Russia’s foreign intelligence chief Sergei Naryshkin confirmed he personally held what he called a working-level meeting with Ratcliffe. Kremlin spokesman Dmitry Peskov said Putin was informed without delay.

This was the first known visit to Moscow by a serving CIA director since William Burns went in November 2021, three months before the full-scale invasion. Burns went to tell Moscow that Washington saw the buildup and understood where it was heading. Putin ignored the warning. That parallel isn’t comforting.

Here’s what makes this stranger than it first appears: Ratcliffe and Naryshkin already had a direct line. They spoke publicly about intelligence cooperation in March 2025, and by June, Naryshkin was telling Russian state television they’d agreed to call each other anytime. Ratcliffe didn’t fly to Moscow because America’s spy chief discovered Russia’s phone number. The channel was already open. When two intelligence chiefs who can talk whenever they want decide one of them needs to physically walk into the room, the visit itself becomes part of the message. Burns did the same thing in 2021, for the same reason: to put the CIA director across the table and make clear the warning carried the weight of the American intelligence community and the president behind it.

The Wall Street Journal first reported he went to warn Moscow against attacking NATO. CBS, ABC, and RFE/RL have since reported the same basic purpose through their own sources, with particular concern focused on the Baltic states. Trump publicly denied that explanation. The CIA isn’t commenting. Moscow calls it a working-level meeting.

We don’t have an official transcript. What we have is a confirmed trip and meeting, followed by multiple independent reports describing a NATO warning, against the backdrop of an intelligence assessment that Putin may be considering exactly that kind of test. That’s considerably more than internet speculation. It’s also less than proof he’s already decided to do it.

When I covered that assessment three weeks ago, the scenario was never Russian tanks charging toward Warsaw. The concern was something smaller: cyberattack, sabotage, a drone crossing a border, a limited incursion by unmarked forces, something calibrated to create an argument inside NATO about whether it qualifies as the kind of attack Article 5 was designed to answer. Putin doesn’t need to defeat NATO militarily for that strategy to work. He needs one moment of hesitation. If Latvia gets hit and Washington spends forty-eight hours arguing definitions while Brussels schedules an emergency meeting, Moscow learns something useful. If Europe responds immediately and Washington makes clear a small deliberate attack produces a major response, Moscow learns something very different. Ratcliffe’s trip may have been designed to remove that uncertainty before Putin runs the experiment.

I used to participate in intelligence forecasting competitions, the kind where you spend an unhealthy amount of time assigning actual probabilities to terrible things happening, and one lesson from that world applies directly here. Who delivers a warning is data, independent of what the warning says.

A diplomat delivering this message tells Moscow that Washington is concerned. An intelligence chief delivering it tells Moscow something else entirely, that the concern is being read through the exact apparatus that watches Russia, feeds Ukraine’s targeting, and would be the first to know if Moscow tried something quiet. That’s not a stylistic detail. In forecasting terms, it changes the base rate Moscow should assign to getting caught.

Yuri Baranchik, a Kremlin-friendly analyst, made a point worth taking seriously, not as evidence of what Washington intended, but as a signal of how Moscow might read the messenger. Ratcliffe isn’t Steve Witkoff. He isn’t Jared Kushner. He represents the American intelligence apparatus, and that matters enormously to Ukraine’s war. Recent reporting says American and French intelligence has helped Ukrainian planners understand Russian air-defense coverage, choose deep-strike flight routes, and identify which parts of Russian refineries to hit to keep them offline longest.

Ratcliffe doesn’t personally sit at Langley with a lever marked “Ukraine targeting intelligence.” But everyone in that Moscow meeting understands the United States possesses capabilities that could make Ukraine’s strike campaign substantially more dangerous. That makes Ratcliffe more than a courier. If Russia escalates against NATO, Washington has options to make Ukraine better informed and more capable of hitting the Russian war economy. If Washington chooses differently, it can constrain that relationship instead. I’m not saying Ratcliffe offered either option in Moscow, we have no evidence of that. I’m saying the person delivering the warning embodied a capability Russia cares about in a way an ordinary envoy simply can’t.

Then there’s the strike pause. The straightforward explanation is operational safety, and it’s a good one: you don’t want an American C-17 anywhere near a Ukrainian strike Russia is trying to intercept. That alone justifies the request. But from Moscow’s side, something else happened simultaneously. Washington asked Kyiv to alter its strike behavior for a specific window, and got a yes. That doesn’t prove theater. It does demonstrate something the Russians could see for themselves. Ratcliffe may have gone primarily to protect NATO; Trump also said the trip might contribute to ending the Ukraine war, and ABC reports Iran came up too. Intelligence chiefs don’t fly halfway around the planet to discuss one subject. Ukraine has every reason to wonder what else got discussed while it was outside the room.

NATO’s eastern flank currently runs nine multinational battlegroups from Bulgaria to Estonia, deliberately structured so that an attack on any one of them kills soldiers from several allied nations simultaneously, not by accident, but by design. The UK leads the battlegroup in Estonia. Canada leads Latvia’s, now scaled up to a full brigade with up to 2,200 Canadian troops alone, backed by contributors from Albania to Spain. Germany is standing up a permanent brigade in Lithuania, reaching 5,000 troops by 2027. None of that includes American forces, which lead the separate battlegroup in Poland. For the specific scenario this video is about, not a conventional invasion, but a limited, deniable probe, those forces, combined with each country’s own standing military and territorial defense units, are not undersized for the job. They’re sized to guarantee that if Russia tests one of them, the casualties are instantly multinational, which is exactly what makes the political response fast rather than optional.

Estonia has already taken that logic to its cleanest possible conclusion. I covered this in an earlier video: Estonian officials have stated plainly that unmarked Russians crossing the border get treated as armed intruders, full stop, no waiting to confirm insignia or chain of command before responding. That’s the direct answer to the exact gray-zone strategy Russia ran in Crimea in 2014, where unmarked soldiers existed specifically to create just enough legal ambiguity to slow down a response. Estonia removed the ambiguity by rule, before the incident happens, rather than negotiating it in real time while soldiers are already on the ground.

So if Putin is running the calculation Baranchik and others suggest, betting that a small test produces Western hesitation, he’s betting against a part of NATO that has spent years deliberately engineering itself to remove exactly that hesitation. The forces are multinational on purpose. The rules of engagement in at least one frontline state are unambiguous on purpose. The physical capability to stop a limited incursion was never the open question here.

The open question was always what Moscow believed about NATO’s willingness to use it. That’s the belief Ratcliffe may have just flown to Moscow to correct.

Why did this warning need delivering right now? Three reasons line up with the earlier intelligence assessment, and none of them require assuming an attack is imminent.

There’s the battlefield: Russia has spent four years grinding through Ukraine for small territorial gains at enormous cost. A conventional NATO attack mid-war would be military insanity. A limited political test is a different proposition entirely, and Russia already has years of gray-zone experience running exactly that kind of operation without an armored division in sight.

Then there’s the timing: Europe is rearming, but most of that money hasn’t yet become finished missiles and full magazines. The US burned through major munitions stockpiles during the Iran conflict. If Moscow believes NATO will be materially stronger in 2029 than in 2026, waiting makes the problem worse from Russia’s perspective.

Then there’s the politics: Putin watched the West argue publicly over every escalation this war, tanks, F-16s, ATACMS, long-range strikes inside Russia. Moscow may have concluded Western fear of escalation can itself be weaponized. That could be a catastrophic misreading, since NATO’s caution over Ukraine, a non-member, tells you nothing about how the alliance responds to an attack on actual member territory, where multinational forces are deliberately stationed so a small Russian attack starts killing troops from several allied countries immediately. Still, if Putin believes the alliance is fractured, the temptation to test that belief gets easier to understand.

Some Kremlin-adjacent commentary is already floating scenarios about shifting pressure toward a weak point in Europe, and others have dragged nuclear doctrine into discussions of Ukraine’s energy-infrastructure strikes. That’s not evidence of an operational decision. It’s threat signaling aimed at Western capitals, designed to make policymakers imagine the worst consequence before approving the next escalation step. This is exactly why intelligence services watch rhetoric alongside hardware: rhetoric tells you which ideas the political system is normalizing, hardware tells you whether anyone’s preparing to act on them. Right now, I haven’t seen public evidence of the kind of major Russian military buildup that would precede a conventional NATO attack. That’s reassuring. A gray-zone test is harder to see coming precisely because it doesn’t require divisions moving toward a border.

So what should we actually watch? Unusual Russian movements around the Baltics, Kaliningrad, and Belarus. Snap exercises that turn more serious than planned. Escalating sabotage, drone incidents, cyber operations, border provocations. Then watch the American side: does Washington increase intelligence sharing with frontline NATO states, does Ukraine see new restrictions or increased targeting support, does Ratcliffe get back on that “routine” phone line with Naryshkin in the coming weeks. Those changes tell us more than another Kremlin spokesperson insisting Russia has no intention of doing the thing Washington apparently just sent its spy chief to warn against.

One more detail worth filing rather than building on: a Kremlin-linked analyst claimed a Russian presidential-fleet aircraft headed west roughly an hour after Ratcliffe’s C-17 departed, speculating it carried Naryshkin. No confirmation of who, or whether anyone relevant, was aboard. Interesting if somebody later proves it. Nothing more than that today.

Three weeks ago, US intelligence was reportedly warning Putin might test NATO sooner than expected. This week, America’s intelligence chief personally flew to Moscow and multiple independent sources say he told Russia not to do exactly that. Washington apparently believed the risk of miscalculation had gotten serious enough that a phone call was no longer sufficient.

The last CIA director to make a trip like this was William Burns in 2021, warning Moscow that Washington saw the invasion preparations. Putin invaded anyway. So I don’t take comfort merely from the fact that the message was delivered. The reassuring read is that Washington saw enough warning signs to intervene early, and the Kremlin now knows a NATO test won’t arrive as a surprise. The darker read is that American intelligence has seen enough behind closed doors that a warning from the CIA director suddenly felt necessary. We don’t know which explanation deserves more weight yet.

Ratcliffe walked into that room representing the system helping America understand Russia, helping NATO watch Russia, and helping Ukraine penetrate parts of Russia’s defenses. The message Moscow may have heard went beyond “don’t do this.” It likely carried an implicit second sentence.

We can see you thinking about it.

That’s deterrence at its most basic. You remove the adversary’s hope that he can do something quickly, ambiguously, or quietly enough to split the coalition before anyone understands what happened.

The Kremlin has spent years trying to convince the West that escalation is something only Russia gets to control. This week, the head of the CIA flew into Moscow to remind them it isn’t.

That’s it for today, my friends. If this kind of intelligence analysis is useful, hit the thumbs up and subscribe. I cover Ukraine, NATO, Russia, and the increasingly strange ways governments communicate when normal diplomatic channels stop doing the job, four times a week.

And as always, Glory to Ukraine, glory to the heroes, Crimea is Ukraine.

Think of this as your weekly strategic weather report exclusively for paid subscribers. Five things to watch and what could break next in war, defense tech, and geopolitics. Just the pressure points most likely to shape the next seven days.

I try to keep this a no-nonsense, high-level brief to start your week… But snark occasionally creeps in.

If you want to understand how I build my OSINT dashboard, I wrote about my workflow here:

Source links in the text. Let’s jump in:

BLUF:

The month-long pause ended overnight with missiles over Jordan. One person is dead at the Grand Canyon and fifteen more are missing. Iceland voted “no” for the EU. The White House declared a national emergency over the power grid on the last day of August. And China’s factories contracted for the second straight month while Brent crude jumped to $90. Stay sharp.

Read more

Hey friends, anytime I can turn a bit of Ukraine news into a radar story, you know I’m all over it.

In Ukraine, an interceptor drone launches, climbs, and gets vectored toward the predicted intercept point with beautiful precision.

Then the pilot looks through the camera and sees a wall of gray.

The Shahed might be two hundred meters away. Ground radar knows it’s there. The interceptor is close enough to reach out and smack it upside the head.

But the pilot can’t see a damn thing.

Serhii “Flash” Beskrestnov, who now advises President Zelenskyy, described this exact sequence when he wrote late last month that fog and cloud can cut visibility down to a few dozen meters, and that interceptor success rates fall off a cliff once winter or weather sets in.

So, Ukraine built one of the most effective air defense innovations of this war, a fleet of $2,000 drones that hunt $40,000 Shaheds, and the thing that’s currently beating it is just bad weather.

One way around this is to put a small radar seeker in the nose of the interceptor drone, but miniaturizing radar is a kind of a hard engineering problem.

One California company thinks they have it solved… and, are already working with Ukraine.

How the “thrift-shop-kill chain” works (and I say that as a term of endearment)

A Shahed is small, made mostly of composite fiberglass and foam. It flies low and moves somewhere around 180 km/h for the propeller-driven versions. That’s close to the worst possible target for a traditional military radar, which is generally built and tuned to find fast, high-flying jets and missiles with a big metal radar cross-section.

Add to that the fact that Russia’s opening missile strikes in 2022 wiped out a meaningful chunk of Ukraine’s Soviet-era radar coverage, and Ukraine found itself needing a ground-based detection network almost from scratch, against a target type conventional radar was never optimized to see.

Ukraine’s answer was to stop relying on radar alone.

The country built out two parallel acoustic sensor networks, Sky Fortress and Zvook, that now total somewhere north of 24,000 individual listening posts for a combined cost under $5 million.

Each unit is a cheap microphone rig, sometimes not much more than a parabolic concentrator and a repurposed phone processor mounted on a radio tower ten or twelve meters up, tuned to recognize the distinctive moped-engine drone of a Shahed’s two-stroke motor.

A Shahed can dodge radar by flying low, but can’t fly without making noise, and it turns out a $500 microphone array is exceptional at hearing a go cart crossing the sky at three in the morning.

These systems typically pick up a Shahed around 5 kilometers out and a cruise missile even farther, and Zvook in particular runs silent because it doesn’t emit anything for Russian electronic warfare to detect and target back.

Radar still carries a lot of the load, especially against faster or higher targets, and it’s a mixed fleet.

Germany’s Hensoldt has been shipping Ukraine the TRML-4D since the fall of 2022, an X-band AESA radar (that’s active electronically scanned array, essentially a radar face built from thousands of tiny transmit-and-receive elements instead of one big spinning dish, letting it track many targets at once instead of sweeping past them one at a time). This system is capable of holding roughly 1,500 tracks simultaneously out to 250 kilometers.

It’s also the radar that normally cues Ukraine’s German-supplied IRIS-T batteries.

Sweden has sent over its Giraffe family, older G/H-band systems, think 4 to 8 GHz, that were originally designed for exactly this kind of gap-filling short-to-medium-range air defense role and can pick out small, slow, low-flying drones out to about 100 kilometers.

Purpose-built counter-drone radars like Robin Radars round out the mix at the shorter end, compact enough to set up in minutes, with a recent software upgrade pushing detection range on a 180 km/h target out to 12 kilometers.

None of these systems work alone.

The acoustic network hears a Shahed enter Ukrainian airspace near the border and starts a track. Radar picks it up and refines that track with speed, altitude, and heading. Both feed the same national air-defense picture.

That fused picture is what cues the nearest interceptor crew and gets a Sting or an Octopus into the air, headed toward a specific patch of sky with genuine confidence that something worth killing is actually in it.

Only then does the human pilot start doing the hard part, hunting visually through an FPV camera feed until the target appears, closing the gap, and finishing it.

Flash calls that original approach primitive, cheap, and effective, and the results back him up.

Wild Hornets’ Sting interceptor alone has claimed several thousand Shahed and Gerbera kills since its first confirmed hit went viral in the spring of 2025, at a unit cost hovering around $2,000.

Ukraine’s defense ministry reported more than 33,000 enemy drones destroyed by interceptors in March of this year alone, double the month before.

Newer Shahed and Geran variants carry onboard electronic warfare, fly evasive profiles, and in some cases mount rear-facing cameras so the drone can watch its own tail for an approaching interceptor.

Flash recently posted footage of an EW module pulled from a downed Shahed that detects the video signal an interceptor pilot’s camera transmits, then jams that exact frequency at higher power, blinding the pilot right at the moment he needs his eyes the most.

Ukraine’s answer has largely been AI and machine vision.

A Brave1-backed system tested in combat around Kharkiv this June automated roughly 95 percent of the engagement, letting a human operator select a target and authorize the attack while software handles acquisition, tracking, and the final approach. Other manufacturers are chasing similar terminal guidance.

But machine vision is still vision. A computer can spot a Shahed faster than a tired human ever could. It still can’t see one if the actual photons never make it through several hundred meters of cloud and hit the camera’s sensor in the first place.

That’s the hole a seeker radar is built to plug.

What a radar seeker does

Think about how a real guided missile finds its target... It’s not staring through a camera.

It carries a small transmitter that sends out radio energy, catches whatever bounces back, and lets an onboard computer calculate where the target sits relative to the missile itself.

None of that depends on visible light; clouds, smoke, darkness, and fog barely register.

The trick was always that a full radar seeker belonged to a different budget bracket entirely, the kind of six-figure subsystem that takes a couple hundred engineers at a major defense contractor years to build.

Like Lockheed and their Scrooge McDuck-sized money vault.

Now, a tiny San Francisco startup called Vernius Systems just unveiled something that tries to shrink that math down to size.

Its Archimedes 1A is an 80-millimeter active radar seeker, operating at 24 GHz, weighing little enough to bolt onto an existing interceptor airframe, and priced around $4,000.

The company has already demonstrated it mounted on Wild Hornets’ Sting.

Flash himself flagged the announcement, and to his credit he didn’t oversell it.

He called miniature radar seekers an extremely difficult engineering problem and said battlefield performance is still an open question. Let’s borrow some of that caution, because Vernius is a very young company making a lot of claims about its own hardware (it also says it landed $500 million in letters of intent and sold an initial batch of 45 units to a Ukrainian special-forces unit within eight weeks of starting).

And here’s the kicker: A radar seeker is normally the single most expensive part of any guided interceptor.

Compress that cost from six figures down to $4,000 and you can strap a serious sensor onto an airframe that itself costs about half as much as the sensor sitting on top of it.

Against a Shahed running tens of thousands of dollars, and against the alternative of burning a more expensive interceptor, like Aster 30 or AMRAAM, on a target Russia can mass-produce, that math still works heavily in Ukraine’s favor.

It’s not one radar doing everything

Vernius describes a layered approach.

The larger ground radar network still does the long-range detection and hands the interceptor a rough cue. Archimedes picks up from there, using its radar to search the small assigned slice of sky, find the target at around a kilometer out, and guide the interceptor closer.

There’s also a simpler mode worth mentioning here: Rather than flying the intercept itself, Archimedes can just feed range and bearing back to a human pilot, essentially handing him an instrument reading that tells him where the target is before his own eyes or camera ever could.

In this use case, the human pilot would be essentially flying by instrument only, the way a commercial airline pilot might fly in bad weather when she can’t see anything outside of the cockpit window.

Just trust the radar seeker to guide you to the target and fly by instrument only. In commercial aviation, this is the IFR equivalent. VFR (Visual Flight Rules) means flying by looking outside at the ground and horizon, while IFR (Instrument Flight Rules) means flying using only the cockpit instruments with help from air traffic control.

Flash has sketched out where this eventually goes: the network detects a Shahed, assigns the nearest interceptor, launches it, steers it with ground radar, and hands off to the onboard radar seeker for the kill.

And that, friendos, describes, you guessed it… a surface-to-air missile; just an incredibly cheap one. Detection, a targeting network, an effector that launches on cue, midcourse guidance, a terminal seeker, and an intercept. Bob’s your uncle.

Ukraine is building it for the price of a high-end bicycle instead of the price of a Dallas builder-grade McMansion.

Read more

Hey friends, if you care about Canadian sovereignty, you need to read this…

Canada just signed a C$2.3 billion contract for a sovereign milita…

Read more

Hey friends.

On August 25, standing next to Estonian President Alar Karis in Kyiv, Volodymyr Zelensky said something that managed to be both good news and a warning shot at the same time.

“To be honest, some Patriot missiles also arrived. I cannot say how many, but some. We need more.”

That’s the whole quote, and I want you to notice everything he didn’t say.

Not how many.

Not which country sent them.

Not which variant of interceptor.

Just: some arrived, and it isn’t enough.

Back on August 6, Zelensky met with NATO Secretary General Mark Rutte to secure more Patriot interceptors ahead of winter. It would be my guess that this first small batch is likely the result of that meeting.

Yesterday, he also confirmed Ukraine received a fresh batch of AIM air-to-air missiles and, in the same breath, said Kyiv had gotten confirmation that several French Crotale air-defense systems and their missiles would be delivered, again declining to say which country was actually sending them.

The Crotale is notable because I don’t think a lot of people are familiar with this system; so, I figured today we’d take an in-depth look at the Crotale.

By the way, I saw reporting that said Crotale was a French MANPAD.

A MANPAD is the thing most people picture when they hear “portable air defense”: a soldier with a Stinger or Igla on his shoulder.

Crotale NG is actually several weight classes above that.

We’re talking about a complete short-range surface-to-air missile system built around radar, electro-optical sensors, fire-control equipment, computers, and eight ready-to-fire VT-1 missiles.

Depending on the configuration, the whole package can sit on a vehicle or trailer. Nobody is throwing one over his shoulder unless France has quietly developed a super-soldier program I somehow missed. Sacre Bleu!

The name Crotale, incidentally, is French for rattlesnake - or so my French-speaking wife tells me. For once, the weapons marketing department got something right.

(I only speak English, Texan, bad Russian, and enough Spanish to bribe my way out of a jail in Ecuador.)

The rattlesnake at the middle of Ukraine’s air-defense stack

Crotale has been around in one form or another since the Cold War.

The original system emerged from a South African requirement in the 1960s, France adopted its own version in the following decade, and the much more capable Crotale NG, or “New Generation,” arrived around 1990.

That makes it old by defense-industry standards.

But, as Ukraine has repeatedly proven, old shit can be useful shit.

Read more

NOTE: I use two primary OSINT dashboards to scan for these topics; SITDECK and God’s Eye View open-source software. The issue was likely on my end, but both were down for me on Monday. Apologies for the late delivery.

Think of this as your weekly strategic weather report exclusively for paid subscribers. Five things to watch and what could break next in war, defense tech, and geopolitics. Just the pressure points most likely to shape the next seven days.

I try to keep this a no-nonsense, high-level brief to start your week… But snark occasionally creeps in.

If you want to understand how I build my OSINT dashboard, I wrote about my workflow here:

Source links in the text. Let’s jump in:

BLUF:

Iran caused the most expensive intelligence damage in American history and it’s just now being reported. Elbridge Colby is rewriting nuclear doctrine with no arms control treaty left to constrain it. Sudan bombed Chad. Two B-2s are flying low and slow near Japan. And gold is screaming danger while the VIX is taking a nap. Stay sharp.

Read more

Read more

But I didn’t want to leave my free subscribers hanging, so check out this video on the formation of Sweden’s newest spy agency to counter Russia. If you’re not into video, click on the transcript button to read it instead… But then you’ll miss me in a tux; a truly rare occurrence in the wild.

Sweden just confirmed it’s building a brand-new foreign intelligence agency modeled directly on Britain’s MI6, putting the final nail in the coffin of neutrality. The trigger sits on the seafloor: a sustained pattern of undersea cable sabotage across the Baltic, tied to Russia’s shadow fleet, that Sweden’s fragmented intelligence structure was never built to piece together.

I cover how the new agency, UND, will work, why NATO membership made the old setup untenable, and what’s still genuinely uncertain about the timeline.

Слава Україні!

EDIT: Transcript button is being funky, so here’s my script pasted in from word:

VIDEO TRANSCRIPT:

Since 2022, undersea cables connecting Sweden to Latvia, Finland to Estonia, and Sweden to Lithuania have all been damaged, in incidents that keep involving the same category of vessel: aging tankers and cargo ships tied to Russia’s so-called shadow fleet, dragging anchors across exactly the infrastructure that keeps Nordic countries talking to each other.

None of the individual investigations has definitively proven Russian state involvement. That’s exactly the point. Sweden’s own official review concluded its military intelligence service failed to predict Russia’s full-scale invasion in 2022, despite visible signs of a military buildup beforehand. Now the country that spent 200 years staying deliberately neutral, refusing to join alliances, refusing to run a foreign spy service of its own, just announced it’s building one from scratch.

Sweden is creating a new foreign intelligence agency. Officials are comparing it directly to Britain’s MI6. And the timing isn’t subtle. This is a country that watched cables get cut in its own territorial waters, watched its own intelligence services miss a war they should have seen coming, and decided the fix wasn’t better weapons.

It was better eyes.

If Britain has MI6, America has the CIA, and Russia has a collection of men falling out of windows, Sweden now wants a dedicated foreign intelligence service of its own.

Hey friends, Wes here. Defense journalist, law grad, Army and Air Force veteran. If you’re new here, welcome to my channel and if you’re not, welcome back, you are looking amazing today, I hope somebody has told you that. Whatever you’re doing, keep doing it because it’s working.

Sweden is building its own MI6. So let’s talk about why a neutral country for two centuries just decided espionage is now a national priority. Bond had an Aston Martin. Sweden has Saab. Personally, I’d take the Gripen.

Most people think of intelligence agencies as something countries either have or don’t, fixed institutions that exist quietly in the background regardless of the geopolitical weather. That’s not actually true for Sweden. Sweden has spent two centuries deliberately avoiding exactly this kind of institution, precisely because neutrality meant staying out of the alliance structures and intelligence-sharing arrangements that make a foreign spy service actually useful.

Neutrality is gone. Sweden joined NATO in 2024. And once you’re inside an alliance, staying without a foreign intelligence service isn’t caution anymore. It’s a liability, because every ally around you is sharing a level of intelligence product Sweden currently has no dedicated agency built to produce, receive, or reciprocate.

Let’s be clear about the mechanics, because this is a real institutional shift.

Sweden’s government confirmed on May 5th, 2026, that it’s moving forward with a new civilian foreign intelligence agency, officially called Sweden’s foreign intelligence service, abbreviated UND, set to begin operations at the start of 2027. Foreign Minister Maria Malmer Stenergard announced the plan, explicitly framing it as a rethink prompted directly by the war in Ukraine: “During the ongoing war in Ukraine, it has become very clear that an information advantage and the ability to rapidly and continuously adapt various technical systems are just as crucial as advanced weapons systems.” She compared the new service directly to the UK’s MI6.

Sweden currently runs three separate agencies handling pieces of this mission. MUST, the Military Intelligence and Security Service, covers foreign military threats. SAPO, the Swedish Security Service, handles domestic threats and sits within the police structure. FRA, the National Defense Radio Establishment, runs signals intelligence. UND will absorb some of MUST’s current responsibilities while working closely alongside all three, reporting directly to the government as a distinct civilian foreign intelligence body, comparable in structure to how the CIA or MI6 sit outside their countries’ military chains of command.

The cost is roughly 2.8 billion Swedish kronor, about $302 million, with resources primarily transferred from the existing armed forces budget rather than new spending layered on top.

If Sweden’s new foreign spy agency is watching Russia, I hope they budget extra storage. Russian incompetence generates a lot of footage.

Wait, why did Sweden never have a foreign spy agency?

Sweden maintained military non-alignment for roughly two centuries, a policy that shaped nearly every institution in the country’s security architecture. Non-aligned countries generally don’t build dedicated foreign human intelligence services in the mold of MI6 or the CIA, because that kind of service is built specifically to support alliance-level military planning, coordinated foreign policy, and the kind of intelligence-sharing relationships neutral countries deliberately avoid entangling themselves in. Sweden’s existing structure, military intelligence covering foreign military threats, domestic security handled through the police, and signals intelligence running separately, reflected exactly that posture. Watch, but don’t fully commit.

Russia’s full-scale invasion of Ukraine in February 2022 broke that calculus immediately. Sweden applied for NATO membership within months and formally joined the alliance in 2024. Stenergard was explicit about what that membership actually demands: “We also have new expectations of ourselves as a NATO ally, and as we now develop our intelligence structure, we will also better match the structures that exist within NATO and among our allies.”

That’s a direct admission. Sweden’s old intelligence architecture was built for neutrality. NATO membership requires a different one, and Sweden is now building it specifically to match what its allies already run.

This isn’t purely a forward-looking modernization project. It’s also an admission of a specific failure. Like many Western intelligence services, Sweden’s Military Intelligence and Security Service did not foresee the scale of Russia’s 2022 invasion, despite visible indicators of a major military buildup beforehand. That miss triggered an official government review and real political criticism, and the proposal for a dedicated foreign intelligence agency emerged directly from that review process, receiving initial government approval before this month’s formal announcement to proceed.

That sequence matters. Sweden isn’t building UND because Brussels asked nicely or because NATO membership comes with a mandatory checklist. It’s building UND because Swedish officials looked directly at their own intelligence failure around the most consequential European security event in decades, and concluded the existing structure, split across three separate agencies with different mandates and different chains of command, wasn’t producing the kind of coordinated foreign intelligence picture a NATO ally actually needs.

Now let’s connect this decision to what’s actually been happening in Nordic and Baltic waters, because the timing lines up perfectly.

Since Russia’s invasion began, Nordic and Baltic security services have documented a sustained pattern of incidents involving undersea infrastructure. In December 2024, Finland seized a Russian-linked oil tanker after suspected sabotage of an undersea fiber-optic cable, part of what Finnish investigators later said was a crew poised to cut additional cables when caught. That same period saw the Eagle S, flagged in the Cook Islands, drag its anchor and cut the Estlink-2 cable in the Gulf of Finland. The Chinese-flagged Yi Peng 3 is suspected of deliberately dragging its anchor to sever both the C-Lion1 cable connecting Finland and Germany and the Arelion cable linking Sweden and Lithuania.

Four additional incidents followed in late 2024 and early 2025, three of them near Sweden’s strategically significant island of Gotland, affecting telecommunications, power, and gas connections running between Nordic countries, the Baltic states, and Germany. On January 26th, 2025, Swedish coast guard vessels boarded a Bulgarian-flagged ship on suspicion of anchor dragging and cable sabotage, triggering NATO to deploy patrol ships under a mission specifically created for this pattern, called Baltic Sentry. A 2025 cyberattack on a heating plant in western Sweden was formally attributed by Swedish officials to a pro-Russian group with links to Russian security and intelligence services, a case Swedish authorities directly compared to a December 2025 cyberattack against Poland’s power grid.

None of these individual incidents has been definitively proven as a direct Kremlin order. That’s consistently true across nearly all of them, and it’s not an accident. It’s the entire operating logic of the campaign: deniable vessels, ambiguous flags, damage that could plausibly be accidental anchor dragging, repeated often enough and across enough infrastructure that the cumulative pattern becomes unmistakable even when each individual case stays legally unresolved.

A broader assessment counted more than 150 incidents of sabotage, cyberattacks, and influence operations linked to Russia across Europe since 2022, run primarily through the GRU, Russia’s military intelligence directorate. Norway’s own Police Security Service has documented similar patterns in its national threat assessments, and a joint 2023 investigation by Swedish, Danish, Norwegian, and Finnish broadcasters tracked Russian naval intelligence vessels, including one called the Admiral Vladimirsky, operating near Swedish training areas and wind farm infrastructure.

Look, in the military, patterns are everything.

One aircraft on an unusual heading could be weather, could be a navigation error, could be nothing. Ten aircraft on the same unusual heading across six months, near the same infrastructure, using the same evasive profile when challenged, stops being noise and becomes signal. That’s exactly the analytical discipline this undersea cable pattern requires, and it’s exactly what a fragmented intelligence structure struggles to produce.

When foreign military threats sit with one agency, domestic security sits with another, and signals intelligence sits with a third, connecting a Bulgarian-flagged tanker’s anchor drag near Gotland to a cyberattack on a Swedish heating plant six months later to a Russian naval intelligence vessel spotted near a wind farm the previous year requires those three agencies to actually talk to each other, consistently, in real time. A dedicated foreign intelligence service, built specifically to fuse exactly that kind of scattered evidence into one coherent picture, is the institutional fix for exactly the gap Sweden’s own 2022 failure exposed.

Here’s what’s uncertain about this announcement.

The legislation still has to clear Sweden’s Council on Legislation, the body that formally scrutinizes draft laws, before heading to parliament, originally targeted for June 2026. Standing up a functioning foreign intelligence service from nothing, recruiting case officers, building overseas infrastructure, establishing the kind of trusted liaison relationships with allied services that actually produce usable intelligence, takes considerably longer than the January 2027 operational start date might suggest to a casual reader. MI6 and the CIA took decades to build the institutional relationships and tradecraft that make them really effective. Sweden is compressing a process that historically unfolds over generations into a matter of months on paper.

Also, the 2.8 billion kronor budget is coming primarily from the existing armed forces budget, not new spending layered on top. That’s a real tradeoff, not a free upgrade, and it means UND’s creation is drawing resources away from other defense priorities at exactly the moment Sweden is also trying to scale up conventional military spending as a new NATO member facing the same Russian threat this agency is meant to help counter.

None of that undermines the logic behind the decision. It means the gap between announcing an MI6-equivalent and actually operating like one will likely take years to close, even after January 2027 arrives on the calendar.

So, Russia’s hybrid campaign against Nordic and Baltic infrastructure has succeeded because it operates in the gap between clearly military and clearly civilian, between obviously deliberate and plausibly accidental, between a single incident that means nothing and a pattern across years that means everything. Sweden’s old intelligence architecture, split across military, domestic, and signals agencies with different mandates, was structurally well-suited to a world of clear threats and clear categories. It was poorly suited to a world where a tanker’s anchor drag, a power plant cyberattack, and a naval intelligence vessel near a wind farm are actually three data points in the same campaign.

Sweden isn’t building UND because espionage suddenly became fashionable. It’s building UND because two centuries of deliberately staying outside alliance intelligence structures left the country structurally unequipped for exactly the kind of deniable, distributed, pattern-based threat Russia has been running against it since 2022. NATO membership didn’t just add Sweden to a defense pact. It exposed how much of Sweden’s intelligence architecture still belonged to the neutrality era.

UND doesn’t yet exist as an operating institution, and the January 2027 target remains a plan rather than a completed transition. Building genuine tradecraft, overseas networks, and the kind of trusted liaison relationships that make a foreign intelligence service actually effective is a multi-year process at minimum, regardless of how quickly the legislation moves through parliament. And Russia’s sabotage campaign has continued evolving specifically to stay ambiguous enough to avoid clean attribution, which means even a fully staffed and operational UND will likely face the same evidentiary challenges that have made most of the undersea cable incidents legally unresolved so far.

Sweden spent two centuries deciding that staying out of everyone else’s intelligence business was the safest possible position. Neutral. Watchful. Uncommitted.

Then a Bulgarian-flagged tanker dragged its anchor across a cable near Gotland, a heating plant in western Sweden got hit by a group with Russian intelligence links, and Sweden’s own military intelligence service admitted it hadn’t seen 2022 coming despite the warning signs sitting in plain view. Two hundred years of careful neutrality couldn’t survive contact with a war being fought partly on the seafloor and partly in cyberspace, using tools specifically designed to never quite cross the line into something a country could point at and call an act of war.

So Sweden is building MI6. In a tuxedo, metaphorically speaking, if not literally, the country that spent two centuries avoiding this exact kind of institution just decided it’s finally ready to dress the part.

If you found this useful, here’s the one thing that actually matters most: hitting subscribe. That’s the single strongest signal I can get that stories like this one, the quiet institutional shifts that don’t trend but actually reshape how a country defends itself, are worth making more of.

Glory to Ukraine, glory to the heroes, Crimea is Ukraine.

Hey friends.

Buried inside a long Atlantic piece by Simon Shuster this week sits what may be the single most consequential paragraph written about this war in months, and nobody is talking about the implications.

Ukrainian officials proposed a plan to swarm Moscow’s airports with up to a thousand drones a night, hoping to shut down international air travel in and out of the Russian capital and isolate Putin’s elite from the world.

It was to be called Operation M&Ms, named for the small, cheap, nearly identical delicious drones the mission would require.

Zelensky presumably heard the pitch, raised serious concerns about the risk to civilian aviation, then reportedly allowed preparations to continue anyway. The plan was suspended after Fedorov’s dismissal in July.

Now, it’s worth noting that Zelensky’s own presidential office called the entire account “total nonsense.”

But the bombshell, and the reason Zelensky rejected the program, is that these drones would have been 100% AI operated; as in, they would have been deciding, on their own, what and who to kill.

If executed, the plan would have marked a sobering milestone in the history of warfare; an attack in which swarms of autonomous drones equipped with AI-based guidance systems would seek out, pick, and strike their targets without a human pilot controlling their movements.

We’re talking autonomous drones that would independently locate and strike targets without human involvement in the engagement.

What the Atlantic reported

The Atlantic’s account comes from two people described as familiar with the plan. According to those sources, Ukrainian officials came to Zelensky in early 2026 with a proposal to expand the ongoing deep-strike campaign beyond refineries, military bases, and logistics infrastructure to include Moscow’s airports.

The concept called for launching waves of cheap AI-enabled drones at the airfields serving the capital, with the stated goal of closing Moscow’s airspace to civilian flights for an extended period, effectively cutting off the Russian elite from international travel.

The sources said the drones could identify targets despite heavy Russian GPS jamming, which is an important technical detail I’ll come back to.

One source described the idea as isolating Putin’s inner circle the way sanctions were supposed to but never fully did, by making Moscow’s airports operationally unusable rather than merely occasionally disrupted.

When the plan was presented to Zelensky, he expressed what the sources described as serious concerns. He feared a drone could hit a passenger aircraft. He worried about killing people in and around the airports. And then, according to The Atlantic’s sources, he allowed preparations to continue.

The operation never launched. Fedorov’s dismissal in July removed the official who had been assembling the development team, creating the AI guidance systems, and securing production funding.

One source told The Atlantic the project could be revived with a new team in roughly three months.

Zelensky’s presidential office flatly denied the entire account. A representative called it “complete nonsense,” saying Ukraine lacked long-range weapons reliable enough to strike airports without posing undue risk to civilians.

It doesn’t really matter if you put more weight behind the Atlantic’s two sources, or Zelensky’s refusal. This technology is here and will be used by someone in the very near future.

It’s quite possible it’s already been used and we don’t know about it yet. When the wreckage was examined, Ukraine’s Major General Vladyslav Klochkov called Russia’s MS001 drone a “digital predator” that “sees, analyzes, decides, and strikes without external commands.” So Russia may already be stumbling drunk toward this very capability.

This also opens a bit of a Pandora’s Box that I think we, as a society, should be discussing. China is likely very close, if not having already achieved, fully AI-driven drones where the AI will decide who to kill.

Do you think teh PRC is going to let a pesky thing like ethics stand in the way of a massive capability leap over the Americans?

The technology is moving far faster than our laws and treaties can keep up.

Every serious military power on earth is chasing autonomous weapons. What’s new here, or what would have been new, is deploying it at scale against a populated city with the explicit intent that the machines find and strike their own targets.

So how does a machine actually ‘decide’ to kill?

The brain of these systems, at least in 2025 and 2026, is an Nvidia Jetson Orin module. It’s a palm-sized computer roughly the size of a credit card, designed and marketed by Nvidia for robotics, autonomous vehicles, and industrial automation.

It was built to help warehouse robots avoid running into shelves and to let self-driving cars recognize a stop sign.

Before the drone ever launches, an operator or a planning system loads it with a target library: a database of images, thermal signatures, and shape profiles representing the things the drone is authorized to hit.

For an airport mission, that library would contain runway surfaces, radar installations, navigation antennas, taxiway infrastructure, terminal buildings, fuel facilities, whatever the mission planners decided counted as a valid target.

Each entry is essentially a digital flashcard the AI has been trained to recognize.

Once the drone is in the air, its onboard camera, either visible-light or thermal, feeds a continuous video stream directly into the Jetson Orin. The processor runs that video through a convolutional neural network, which is the same basic category of software that lets your phone recognize a face in a photograph, except this one has been trained on military targets.

The network compares what the camera sees against every entry in the target library, frame by frame, dozens of times per second.

When the neural network identifies something in the camera feed that matches a target profile above a preset confidence threshold, let’s say 92 percent certainty that the object below is a radar antenna rather than a water tower, the system generates what is functionally a kill recommendation.

In a human-in-the-loop system, that recommendation would go to an operator who approves or rejects it.

In a fully autonomous system, the recommendation goes straight to the flight controller.

The drone adjusts its trajectory, commits to a terminal dive, and strikes.

No human sees the target.

No human approves the engagement.

No human even knows the drone found something until after the detonation, if anyone checks at all.

The navigation side of this runs in parallel and solves the GPS-jamming problem that makes autonomy necessary in the first place.

Instead of relying on satellite signals Russia can jam or spoof, the drone uses computer vision to navigate, comparing what its camera sees on the ground against a library of preloaded terrain imagery and elevation data.

The Jetson Orin runs that comparison continuously, essentially asking “does the ground below me look like the ground I expected to see at this point in my flight?” and correcting its course whenever the answer drifts.

Ukrainian intelligence found that Russia’s V2U drone carried only a single GPS module, suggesting the designers had already accepted that GPS would be unreliable and had built the system to navigate primarily by sight instead. That’s the same approach The Atlantic’s sources described when they said M&Ms drones could “identify and strike targets despite jamming.”

Every piece of hardware I just described, the Jetson Orin, the camera, the neural network, the terrain-matching navigation, already exists.

Ukraine doesn’t need to invent autonomous targeting from scratch.

And by the way, the confidence threshold, (that 92 percent number I used as an example), is a setting somebody chooses before the drone launches.

Set it at 99 percent and the drone rejects most of what it sees, reducing false positives but also reducing the number of targets it engages.

Set it at 80 percent and the drone becomes an aggressive little bastard, striking more frequently but also more likely to misidentify a civilian structure as a military target.

That single number, buried inside a configuration file on a circuit board the size of a queen of hearts playing card, is where the entire moral weight of autonomous warfare collapses into an engineering decision… made by someone who will never personally see what the drone hits.

A thousand drones a night, each carrying its own copy of that threshold, each making its own decision over a city of twelve million people.

That’s what M&Ms would have looked like from the inside.

And remember, there is currently no binding international treaty prohibiting lethal autonomous weapons.

Read more

Hey friends.

I’ve done some weird things to keep my jets flying. I’ll keep the obscener and more sacrilegious things to myself, but here’s an example of how a security forces airman’s burrito saved…

Read more