Mark Johnston retweeted
‼️‼️ Now this is something! 🔽 This technical thread suggests (explains) that the entire Starlink/shield slowdown issue is not a consequence of a political decision in the US or by Musk but the result of Ukrainian engineers hacking the system to bypass it's controls and do something they are not permitted to do - and when Starlink took steps to prevent the bypassing and ensure the controls remained effective, suddenly the system stopped working in 'bypass mode' - and Zelensky then set out to politicise matters. The explanation makes the bits of the puzzle fit together and fits with the New York Times article in which Ukrainians explained that although the system is formally limited to 100 mph they have managed to exceed it.
~~~~
"What was really going on with the Starlink scandal in Ukraine? An analysis of the technical modifications, hacker boards, Elon Musk’s restrictions, and why drones started falling at the end of 2026 and Zelensky staged a pre-election Trump scandal. Big thread 👇 1/15
Let’s start with the tariffs. Inside the country, only civilian Starlink has always worked: the basic tariff ($100), commercial in-motion ($400), and priority headquarters ($2–4k). Aviation tariffs cost up to $1K–20K–100K+, but they are unavailable in the country. 2/15
Because of the high load, the network was overloaded. SpaceX had to keep prices high for the military in order to build an excess of ground stations around Ukraine for minimal signal latency. By the way, these sites with antenna domes recently began to be burned on orders from the darknet. 3/15
The main problem at the front is electronic warfare (EW). Because of GPS jamming, civilian Starlink chips froze, and with false location the terminal was blocked. The solution was a GPS hack board. Engineers soldered it inside to manually set the required coordinates. 4/15
The hack boards also allowed bypassing Musk’s geographic zones. For a long time Starlink did not work in the gray zone under Russian control. But if you entered the coordinates of a neighboring region of Ukraine, the connection came back to life. The Russians used this too, until Kyiv received exclusive whitelists. 5/15
By 2025 Musk began tightening the screws and lowering speed limits for commercial tariffs. But military engineers discovered a loophole: technically the satellite-terminal communication session is not rigidly tied to GPS. The protection was only on the terminal module itself. 6/15
If the terminal thought via the hack board that it was static, it itself sent a radio request to the satellite and received the necessary data for the session. There are already about 10,000 satellites, and the signal was picked up even with a spread of 150+ km. The drone was flying, but the Starlink server considered it stationary. 7/15
This was a shared secret. Before being sent to the front, the cases of new Starlinks were massively opened in order to solder in the GPS hack and install durable moisture-proof connectors. This way the ordinary tariff allowed drones to fly at high speeds. 8/15
At the beginning of 2026 Musk lowered the speed of the commercial tariff to 160 km/h across the entire planet. This is below the speed of many simple UAVs with a tailwind. But because of the mass use of hack boards in Ukraine, no one noticed this. The terminals reported that they were supposedly standing still. 9/15
At the END of 2026 SpaceX responded harshly. A server-side check was introduced: if according to GPS the terminal is stationary, but at the same time it switches from satellite to satellite too quickly (which reveals a real speed above 160 km/h), the server instantly drops the session. 10/15
The GPS hack stopped working. Ukrainian military began having problems with long-range UAVs. Now flights must be strictly planned so as not to exceed the 160 km/h limit. There is not always money for aviation tariffs, and SpaceX does not sell them because of the attacking nature of the use. 11/15
Against this background Zelensky is trying to get access to StarShield (the US military Starlink) without restrictions in order to fly over the Russian Federation. But Trump, Musk, and the Pentagon refuse him. The Ukrainians presented the restrictions as a political blockade, remaining silent about the fact that the Starlink speed limit has already been in place for a year. 12/15
Now there is behind-the-scenes ping-pong: Musk says that he changed nothing these days, Rubio states that StarShield is unavailable to Ukraine. The bottom line: Starlink is gradually being squeezed into the framework of cars and stationary use. Bypassing the system no longer works. 13/15
Whether the speed check was turned off or access was given to special aviation tariffs is still unclear. The failure affected a narrow segment of drones. However, satellite communication on super-high-speed drones was already a compromise because of the signal delays of 1000 km to orbit and back. 14/15
Russian UAV specialists are in no hurry to rejoice. The simplicity of using Starlink slowed down the development of other technologies in Ukraine. Now Kyiv can, on an emergency priority basis, refine more complex autonomous mesh-AI systems that will fly deep into the Russian Federation. End of thread. 15/15
Written based on materials from last year’s newspapers, most likely not true and total nonsense. No one and nothing is right on the internet. Author — a boring schoolboy from Georgia."
Mark Johnston retweeted
This makes sense. (1) Ukraine had been hacking Starlink to get around flight restrictions for years. (2) When Starlink tightened their security, the hacks failed. (3) For political reasons, Zelenskyy accused Musk of shutting down access.
‼️‼️ Now this is something! 🔽 This technical thread suggests (explains) that the entire Starlink/shield slowdown issue is not a consequence of a political decision in the US or by Musk but the result of Ukrainian engineers hacking the system to bypass it's controls and do something they are not permitted to do - and when Starlink took steps to prevent the bypassing and ensure the controls remained effective, suddenly the system stopped working in 'bypass mode' - and Zelensky then set out to politicise matters. The explanation makes the bits of the puzzle fit together and fits with the New York Times article in which Ukrainians explained that although the system is formally limited to 100 mph they have managed to exceed it.
~~~~
"What was really going on with the Starlink scandal in Ukraine? An analysis of the technical modifications, hacker boards, Elon Musk’s restrictions, and why drones started falling at the end of 2026 and Zelensky staged a pre-election Trump scandal. Big thread 👇 1/15
Let’s start with the tariffs. Inside the country, only civilian Starlink has always worked: the basic tariff ($100), commercial in-motion ($400), and priority headquarters ($2–4k). Aviation tariffs cost up to $1K–20K–100K+, but they are unavailable in the country. 2/15
Because of the high load, the network was overloaded. SpaceX had to keep prices high for the military in order to build an excess of ground stations around Ukraine for minimal signal latency. By the way, these sites with antenna domes recently began to be burned on orders from the darknet. 3/15
The main problem at the front is electronic warfare (EW). Because of GPS jamming, civilian Starlink chips froze, and with false location the terminal was blocked. The solution was a GPS hack board. Engineers soldered it inside to manually set the required coordinates. 4/15
The hack boards also allowed bypassing Musk’s geographic zones. For a long time Starlink did not work in the gray zone under Russian control. But if you entered the coordinates of a neighboring region of Ukraine, the connection came back to life. The Russians used this too, until Kyiv received exclusive whitelists. 5/15
By 2025 Musk began tightening the screws and lowering speed limits for commercial tariffs. But military engineers discovered a loophole: technically the satellite-terminal communication session is not rigidly tied to GPS. The protection was only on the terminal module itself. 6/15
If the terminal thought via the hack board that it was static, it itself sent a radio request to the satellite and received the necessary data for the session. There are already about 10,000 satellites, and the signal was picked up even with a spread of 150+ km. The drone was flying, but the Starlink server considered it stationary. 7/15
This was a shared secret. Before being sent to the front, the cases of new Starlinks were massively opened in order to solder in the GPS hack and install durable moisture-proof connectors. This way the ordinary tariff allowed drones to fly at high speeds. 8/15
At the beginning of 2026 Musk lowered the speed of the commercial tariff to 160 km/h across the entire planet. This is below the speed of many simple UAVs with a tailwind. But because of the mass use of hack boards in Ukraine, no one noticed this. The terminals reported that they were supposedly standing still. 9/15
At the END of 2026 SpaceX responded harshly. A server-side check was introduced: if according to GPS the terminal is stationary, but at the same time it switches from satellite to satellite too quickly (which reveals a real speed above 160 km/h), the server instantly drops the session. 10/15
The GPS hack stopped working. Ukrainian military began having problems with long-range UAVs. Now flights must be strictly planned so as not to exceed the 160 km/h limit. There is not always money for aviation tariffs, and SpaceX does not sell them because of the attacking nature of the use. 11/15
Against this background Zelensky is trying to get access to StarShield (the US military Starlink) without restrictions in order to fly over the Russian Federation. But Trump, Musk, and the Pentagon refuse him. The Ukrainians presented the restrictions as a political blockade, remaining silent about the fact that the Starlink speed limit has already been in place for a year. 12/15
Now there is behind-the-scenes ping-pong: Musk says that he changed nothing these days, Rubio states that StarShield is unavailable to Ukraine. The bottom line: Starlink is gradually being squeezed into the framework of cars and stationary use. Bypassing the system no longer works. 13/15
Whether the speed check was turned off or access was given to special aviation tariffs is still unclear. The failure affected a narrow segment of drones. However, satellite communication on super-high-speed drones was already a compromise because of the signal delays of 1000 km to orbit and back. 14/15
Russian UAV specialists are in no hurry to rejoice. The simplicity of using Starlink slowed down the development of other technologies in Ukraine. Now Kyiv can, on an emergency priority basis, refine more complex autonomous mesh-AI systems that will fly deep into the Russian Federation. End of thread. 15/15
Written based on materials from last year’s newspapers, most likely not true and total nonsense. No one and nothing is right on the internet. Author — a boring schoolboy from Georgia."
Rotating my retirement account money from 100% S&P (kind of my default state for that account) to 60% small cap / 40% S&P...
Like the rotation potential here after a lot of names have been bombed out - and there have been hints of it waking up on leaderboard this past week
Mark Johnston retweeted
Gave you $LPTH at $6.81 look at all that alpha somebody doesn't want you to see. 😝 7.8% ROI just in a few hours while I have a family ER crisis ongoing. Clearly taking advantage of all of you - womp, womp.
I really can’t wait till Anduril goes public. @PalmerLuckey gosh another amazingly smart guy. I wish someone had offered me a preIPO placement. Just feeling super duper interested in Nat’l defense contractors and tech. Like $LPTH $LMT (this one carries a really really special place in my heart). I guess SNL is not publicly traded but I bet things are super interesting there too especially for a tech nerd such as myself.
I rode the metro to AUSA with a couple of fine young gentlemen who spent a bit of time in Ukraine recently
I asked them what their FAVORITE drone was for the front lines.. they both agreed it was one of the PDW drones (they said a model number that I don't remember)
NEWS: PDW + @BoozAllen are bringing @shieldaitech powered autonomy to a new drone platform
PDW (Performance Drone Works) is integrating Booz Allen Hamilton's mission-autonomy software into its Attritable Multirotor drone
Booz invested in PDW in April to help scale U.S. drone manufacturing. Its Huntsville factory is designed for up to 100,000 drones annually
Booz licenses Shield AI's Hivemind software and will begin to scale it across PWD's fleet
This is evidence of a partner-led distribution model extending from a 2025 partnership
Shield AI is known as a drone manufacturer but they should be known as a autonomy software/AI company that also makes badass drones
Chills... we are living in a new world
Throw out the old assumptions
The US military gets over 80% of its dogs from Europe, and fewer than half of them turn out fit for service.
South Korea hit the same wall back in 2007, and their fix sounds like a movie plot. They took their best drug-sniffing dog, a Lab named Chase, and cloned him seven times.
All seven puppies got the same name, Toppy, short for "tomorrow's puppy." They went through the customs service's detector dog program like any other recruit while researchers tracked how they did. Regular puppies made the cut about 30% of the time. The clones made it 86% of the time, and by 2009 they were working airports and harbors across the country.
That's the business case for what Palmer Luckey and Colossal's Ben Lamm just launched.
Training one of these dogs can run past $60,000. When half your dogs wash out, every dog that graduates is secretly paying for the one that didn't, so each graduate really costs you something closer to $120,000. Get the pass rate up to Korea's 86% and that drops to about $70,000. You'd knock roughly 40% off every working dog the military buys before the dog does a single thing better.
And Warhound wants the dog to do things better. Korea could only copy a great dog. Lamm's team is banking cells from top German Shepherds and Belgian Malinois and then editing the genes. According to Lamm, the dogs that fail usually fail on stuff they were born with, like bad eyesight, cancer risk, hip dysplasia and muscle wasting. So the plan is to copy the best dog they can find and fix whatever would have ended its career.
The cloning runs through ViaGen, a Texas company Colossal owns that has spent years cloning people's pets for around $50,000 a dog. Anduril builds what goes on top of the dog, like vests with sensors that feed data back to the handler. For the record, Palmer doesn't own any of it. Colossal holds the equity, and the CIA's venture fund In-Q-Tel is in the $200 million round with Sequoia.
First puppies are due early next year. Korea's clones needed 16 months of training to prove themselves. If Warhound's dogs put up anything close to those numbers, the next generation of America's war dogs gets born in Texas instead of bought from Europe.
Gold...
"You're probably right as a group, maybe 60, 70% of the time, but because everyone is on one side of the boat, when you're right, you're not paid as much as you think, because you need to sell to pay and then everyone's selling. But when you're wrong, maybe 20% of the time, you really get screwed because everyone's getting out."
Gold @huaxying
Defense wins with the uniparty in power
It's been a silly narrative that either party cuts defense spending... the primes have played this game for far too long to let pesky things like elections matter
$LMT $LPTH $RTX
$LMT must be filling Democrats pockets by the bushel the past 2 days!
BLUE WAVE*
*Unless you fill my personal PAC to the top of the jar
Shield.AI was one of the better booth presentations at AUSA
National Defense article by Shield AI co-founder @brandontseng2
"Training people and training machines draw on common principles, and training an AI pilot is a data problem.
Autonomy software is trained in simulation and refined through real-world operations. What it learns depends on the world it is put in — terrain, weather, sensor behavior and adversary reactions.
AI models need enormous volumes of data, and the military has very little to give them.
A self-driving car company can draw on millions of hours of dashboard camera footage, street imagery and public transit records.
No equivalent archive exists for military aviation.
Sensor video from a fighter stays with the unit that flew the mission. Nations do not hand that material to each other, and services within the same country often keep it from one another, for reasons of classification that will not change.
The data that would teach an AI pilot the most is the data it will never receive.
Simulation is how we close that gap. Image generators and sensor models produce the annotated imagery and video that operational archives cannot supply: the same valley in fog, at night, in winter, with a threat radar in a different position each run.
They also produce the rare events a squadron might see once in a decade, and a machine can explore 10,000 variations of an engagement without risk to an airframe or a life."
Shield AI acquired Aechelon earlier this year to integrate high-fidelity simulation with real-world autonomy. 🔑
Mark Johnston retweeted
National Defense article by Shield AI co-founder @brandontseng2
"Training people and training machines draw on common principles, and training an AI pilot is a data problem.
Autonomy software is trained in simulation and refined through real-world operations. What it learns depends on the world it is put in — terrain, weather, sensor behavior and adversary reactions.
AI models need enormous volumes of data, and the military has very little to give them.
A self-driving car company can draw on millions of hours of dashboard camera footage, street imagery and public transit records.
No equivalent archive exists for military aviation.
Sensor video from a fighter stays with the unit that flew the mission. Nations do not hand that material to each other, and services within the same country often keep it from one another, for reasons of classification that will not change.
The data that would teach an AI pilot the most is the data it will never receive.
Simulation is how we close that gap. Image generators and sensor models produce the annotated imagery and video that operational archives cannot supply: the same valley in fog, at night, in winter, with a threat radar in a different position each run.
They also produce the rare events a squadron might see once in a decade, and a machine can explore 10,000 variations of an engagement without risk to an airframe or a life."
Shield AI acquired Aechelon earlier this year to integrate high-fidelity simulation with real-world autonomy. 🔑
NDIA POLICY POINTS: Simulation Makes Both Personnel, AI Ready | By Brandon Tseng | @NDIAToday @USNavy @shieldaitech #AI
Read now: nationaldefensemagazine.org/…
Mark Johnston retweeted
I really can’t wait till Anduril goes public. @PalmerLuckey gosh another amazingly smart guy. I wish someone had offered me a preIPO placement. Just feeling super duper interested in Nat’l defense contractors and tech. Like $LPTH $LMT (this one carries a really really special place in my heart). I guess SNL is not publicly traded but I bet things are super interesting there too especially for a tech nerd such as myself.
Mark Johnston retweeted
Today was just the 13th trading day since 1990 where the S&P 500 fell 1%+ on positive breadth (more stocks up than down on the day).
Mark Johnston retweeted
Brutal day following a gut-wrenching couple months for uranium equities. Let's check some historical data.
The table below outlines historical forward performance across these episodes when the sector reached equivalent oversold extremes:
Mark Johnston retweeted
#Nuclear Power Now Has Bipartisan Support
Majorities of Both Major Parties now recognize its potential as a More Clean & Stable #Energy Source.
LINK: yahoo.com/news/politics/arti…
#EnergySecurity #Uranium
Someone getting blown out of $LPTH position today if my tape reading skills are anything close to what they used to be...
See about 20 posts in my history this year on being careful with margin... and you have my genuine sympathy if you are reading this post
Mark Johnston retweeted
>"So basically the entire Internet runs on PHP and today they found some critical secu-"
>"Honey, please shut the fuck up."
Big congratulations to North Run on their exit prices and to those shorting $LPTH
Whatever you saw... I did not... and you executed extremely well. I'm genuinely impressed - always learning
I sure as heck thought I was being somewhat careful with a covered call strategy combined with the underlying revenue growth - but even that cannot account for a 62% decline from the high
--Sincerely
$LPTH bag holder





























