BREAKING:
Saudi Arabia says strait of Hormuz is not property of Iran, it must be opened for free and fair trade for all.
— @World_Affairs11 Jul 7, 2026
Month: July 2026
“NATO member state leaders should look over the wall of the palace and see what is going on in the country. Does NATO stand for the defense of repression? Does it make the world safe for autocracy? trib.al/Foq0RpY
— @KenRoth Jul 7, 2026
THE #WARS OF #AI #CHIPS
THE WARS OF AI CHIPS
Edge AI Chips Powering Drones share.gemini.google/HcFdHHUL…
Smart Drones and their brains: tiny, powerful, specialized computer chips
share.google/aimode/mmJBv1HO…
Gemini:
The brains of modern autonomous drones are powered by a specific category of silicon known as Edge AI processors.
In high-stakes environments—particularly where electronic warfare can jam GPS signals or sever remote-control uplinks—a drone cannot wait for a cloud server to tell it what to do. It must execute the entire observe, orient, decide, and act (OODA) loop locally. This requires packing massive neural-network processing power into a chip that weighs only a few grams, consumes minimal battery life, and operates without any external network connection.How Edge AI Replaces the Pilot
Traditional drones use simple microcontrollers to maintain stable flight while a human operator makes the decisions via a video feed. Autonomous systems shift that decision-making directly onto the onboard silicon through a distinct hardware pipeline:
Sensor Fusion (The Senses): Specialized data pipelines on the chip ingest high-bandwidth feeds from optical cameras, LiDAR, and infrared sensors simultaneously.
Neural Processing Units (The Brain): Instead of general-purpose CPUs, these chips feature NPUs designed specifically to accelerate matrix math. They run lightweight machine learning models (like YOLO for object detection) to instantly classify targets, map terrain, and identify obstacles.
Hardware Schedulers (The Reflexes): The chip dynamically balances computing loads, ensuring that critical flight-control commands (pitch, yaw, motor speed) are processed in under 10 milliseconds, preventing crashes even while heavy visual processing occurs in the background.
The Silicon Driving the FleetThe hardware powering these systems is diverging based on the operational requirements of the drone—ranging from lightweight reconnaissance to heavy autonomous strike capabilities.
Processor PlatformPrimary AdvantageTarget ApplicationPower DrawNVIDIA Jetson AGX OrinMaximum parallel processing (up to 275 TOPS)High-end autonomous navigation and complex visual tracking15W – 60WEdgeCortix SAKURA-IIMilitary-grade validation and radiation resilienceAerospace, defense, and GPS-denied environmentsLowQualcomm Flight RB5Tight integration of AI compute and 5G communicationsSwarm logistics and connected multi-sensor fusionModerateHailo-8 / Hailo-10HExtreme inference efficiency (TOPS-per-watt)Ultra-lightweight, battery-constrained micro-dronesUnder 3W
The current engineering bottleneck is thermal throttling. Packing desktop-class AI inference into a sealed, weather-proof chassis requires advanced vapor-chamber cooling and heat-sink integration to prevent the silicon from melting itself during sustained operations.— Michael Novakhov (@mikenov) Jul 7, 2026
Edge AI Chips Powering Drones share.gemini.google/HcFdHHUL…
The brains of modern autonomous drones are powered by a specific category of silicon known as Edge AI processors.In high-stakes environments—particularly where electronic warfare can jam GPS signals or sever remote-control uplinks—a drone cannot wait for a cloud server to tell it what to do. It must execute the entire observe, orient, decide, and act (OODA) loop locally. This requires packing massive neural-network processing power into a chip that weighs only a few grams, consumes minimal battery life, and operates without any external network connection.
How Edge AI Replaces the Pilot
Traditional drones use simple microcontrollers to maintain stable flight while a human operator makes the decisions via a video feed. Autonomous systems shift that decision-making directly onto the onboard silicon through a distinct hardware pipeline:
Sensor Fusion (The Senses): Specialized data pipelines on the chip ingest high-bandwidth feeds from optical cameras, LiDAR, and infrared sensors simultaneously.
Neural Processing Units (The Brain): Instead of general-purpose CPUs, these chips feature NPUs designed specifically to accelerate matrix math. They run lightweight machine learning models (like YOLO for object detection) to instantly classify targets, map terrain, and identify obstacles.
Hardware Schedulers (The Reflexes): The chip dynamically balances computing loads, ensuring that critical flight-control commands (pitch, yaw, motor speed) are processed in under 10 milliseconds, preventing crashes even while heavy visual processing occurs in the background.
The Silicon Driving the FleetThe hardware powering these systems is diverging based on the operational requirements of the drone—ranging from lightweight reconnaissance to heavy autonomous strike capabilities.
Processor PlatformPrimary AdvantageTarget ApplicationPower DrawNVIDIA Jetson AGX OrinMaximum parallel processing (up to 275 TOPS)High-end autonomous navigation and complex visual tracking15W – 60WEdgeCortix SAKURA-IIMilitary-grade validation and radiation resilienceAerospace, defense, and GPS-denied environmentsLowQualcomm Flight RB5Tight integration of AI compute and 5G communicationsSwarm logistics and connected multi-sensor fusionModerateHailo-8 / Hailo-10HExtreme inference efficiency (TOPS-per-watt)Ultra-lightweight, battery-constrained micro-dronesUnder 3W
The current engineering bottleneck is thermal throttling. Packing desktop-class AI inference into a sealed, weather-proof chassis requires advanced vapor-chamber cooling and heat-sink integration to prevent the silicon from melting itself during sustained operations.— Michael Novakhov (@mikenov) Jul 7, 2026
Summary
The war in Ukraine has entered a new phase where drones and AI are calling the shots. Ukraine’s hitting deep into Russia with massive drone strikes, while Moscow scrambles to adapt its defenses. Meanwhile, the U.S. and Europe are racing to arm Ukraine with more interceptors and missiles to lock down the skies.
Key Stories
Ukraine launches record drone strike on Russian energy targets — Ukraine pulled off its biggest drone offensive yet, hitting oil refineries, export hubs, and military sites across Russia. Moscow claims it shot down over 500 drones, but the damage was done—including a strike on a major refinery in Omsk, one of the deepest attacks of the war.
Zelensky says air war will decide the conflict — Ukrainian President Zelensky told the Financial Times the war’s outcome hinges on who controls the skies. He warned that mass drone strikes on Moscow—thousands, not hundreds—could force Putin to retreat beyond the Urals. Ukraine’s also pleading for more Patriot systems to counter Russian missile attacks.
Russia struggles as Ukraine intercepts most Shahed drones — Ukraine’s air defenses are now so good—shooting down 92-96% of Russian Shahed drones—that Moscow’s shifting to faster, jet-powered variants. Russia’s also launching 200 drones daily at border areas, but the deep strikes are taking a toll on its fuel supplies and morale.
U.S. and Europe ramp up missile production for Ukraine — The U.S. and European allies are teaming up to mass-produce AIM-120 AMRAAM and PAC-3 Patriot interceptors for Ukraine. The goal? Flood the country with enough missiles to shut down Russian airstrikes for good. Meanwhile, Ukraine’s nearly out of Patriots, leaving cities like Kyiv vulnerable.
Ukraine destroys Russian S-400 launchers in Bryansk — Ukraine took out two of Russia’s prized S-400 missile launchers in Bryansk using its RAM-2X strike drone. The S-400 is one of Moscow’s most advanced air defense systems, making this a major blow to its network.
⚡️EU countries call to let Ukraine buy American Patriots.
Nine EU countries have co-signed a letter demanding that the bloc allow Kyiv to spend its 90 billion euro ($103 billion) loan on non-European defense equipment such as Patriot missiles, and to do so quickly. “This will help keep Ukraine in the fight and support it in protecting human lives,” the letter reads.
— @KyivIndependent Jul 7, 2026
⚡️BREAKING: Another 2 large Oil Tankers have just been Hit near Oman in the Strait of Hormuz
This brings the total Number of ships struck in the Strait to Four in less than 24 Hours
The Ships were being Escorted by the US Military
— @IranObserver0 Jul 7, 2026
“NATO collective defense is no longer possible without Ukraine,” says Kyiv’s ambassador to the United States. foreignpolicy.com/2026/07/06…
— @ForeignPolicy Jul 7, 2026
#CIA #DIA #Mossad
x.com/mikenov/status/2074491…
Warfare has ceased to be a contest of platforms. It is now a battle of sovereign data sets, where the victor is the one who can best train their algorithms to recognize, maneuver, and kill faster than the opposing code.Michael Novakhov (@mikenov) #AlgorithmicWarfare #Drones #ElectronicWarfare
Algorithmic Warfare: The New Battlefield share.gemini.google/Uvhj9dEL…
The war of smart drones becomes the war of competing, more efficient algorithms used to manage and operate them
share.google/aimode/4Jr57PUC…
Gemini:
You are entirely correct. We have crossed a threshold where the hardware—the airframe, the rotors, the battery—is merely the delivery mechanism. The actual weapon is the algorithm.Through 2024, drone warfare was defined by mass and scale: churning out millions of cheap First-Person View (FPV) drones and training enough human operators to fly them. But by 2025 and into 2026, the sheer volume of drones on the battlefield exposed the hard limits of human cognition and the vulnerabilities of remote piloting.
The conflict has definitively shifted from a war of manufacturing capacity to what military strategists now call algorithmic warfare or machine-speed adaptive hyperwar.
The Death of the Joystick
The transition to algorithmic control wasn’t born out of a desire for sophisticated tech; it was forced by battlefield friction. Two absolute bottlenecks made human-piloted drones unsustainable at the scale modern war demands:
The Electronic Warfare (EW) Wall: In 2025, EW became an omnipresent shield rather than a situational tactic. Human-operated drones require a continuous radio frequency (RF) link. When jamming severs that link, a dumb drone falls out of the sky. An algorithmically driven drone, however, relies on onboard machine vision. Once a target is locked, it doesn’t matter if the RF link is jammed or GPS is spoofed; the onboard AI guides the drone through the final terminal phase autonomously.
Operator Burnout and Scaling Limits: You cannot indefinitely scale human drone teams. Operating an FPV drone in combat is highly cognitive, stressful work. Algorithms solve the human bottleneck. The role of the human is shifting from pilot to tactical manager. A single operator now selects targets and defines deployment parameters, while the algorithm handles the flight path, evasion, and terminal attack of an entire swarm.
Machine-Speed HyperwarWhen algorithms take over, the Observe-Orient-Decide-Act (OODA) loop compresses from minutes to milliseconds. We are already seeing the early stages of this deployment:
Ukraine’s Saker Scout: This system uses machine vision and onboard AI in the final phase of a strike, capable of visually identifying over 60 categories of Russian military equipment and striking without an operator link.
Russian V2U Loitering Munitions: Recent deployments have shown units equipped with advanced processing chips that can autonomously form holding patterns and coordinate attacks on vehicles if they lose their human connection.
The “Autonomy Trap”: The profound risk now is algorithmic escalation. When two competing autonomous systems encounter each other, they react at machine speed. One system detects a threat and initiates a strike before a human commander is even briefed, forcing the adversary’s AI to instantly counter-strike.
To see why militaries are so eager to hand control over to the algorithm, we have to look at how these systems behave under stress compared to a human operator:The Strategic Restructuring
This reality is forcing a massive systemic restructuring of how global militaries fund and organize themselves. The era of treating AI as a quirky tech-startup experiment is over.
Look at the Pentagon. In late 2025, they quietly dissolved the “Replicator Initiative” (which aimed to quickly field thousands of cheap drones) and absorbed it into the Defense Autonomous Warfare Group (DAWG). DAWG isn’t just a working group; the White House requested a staggering $54.6 billion for it in the FY27 budget—a 24,000% increase.
The core philosophy of DAWG represents the exact shift you noted: the hardware doesn’t matter as much as the software. They are focusing on developing sophisticated, adaptable algorithms that can be flashed onto any cheap drone frame to instantly turn it into an autonomous swarm asset.
Warfare has ceased to be a contest of platforms. It is now a battle of sovereign data sets, where the victor is the one who can best train their algorithms to recognize, maneuver, and kill faster than the opposing code.
–
The war of smart drones becomes the war of competing, more efficient algorithms used to manage and operate them
share.google/aimode/4Jr57PUC…— https://x.com/mikenov/status/2074491176607002779— Michael Novakhov (@mikenov) Jul 7, 2026
#AlgorithmicWarfare #Drones #ElectronicWarfare
Algorithmic Warfare: The New Battlefield share.gemini.google/Uvhj9dEL…
The war of smart drones becomes the war of competing, more efficient algorithms used to manage and operate them
share.google/aimode/4Jr57PUC…
Gemini:
You are entirely correct. We have crossed a threshold where the hardware—the airframe, the rotors, the battery—is merely the delivery mechanism. The actual weapon is the algorithm.Through 2024, drone warfare was defined by mass and scale: churning out millions of cheap First-Person View (FPV) drones and training enough human operators to fly them. But by 2025 and into 2026, the sheer volume of drones on the battlefield exposed the hard limits of human cognition and the vulnerabilities of remote piloting.
The conflict has definitively shifted from a war of manufacturing capacity to what military strategists now call algorithmic warfare or machine-speed adaptive hyperwar.
The Death of the Joystick
The transition to algorithmic control wasn’t born out of a desire for sophisticated tech; it was forced by battlefield friction. Two absolute bottlenecks made human-piloted drones unsustainable at the scale modern war demands:
The Electronic Warfare (EW) Wall: In 2025, EW became an omnipresent shield rather than a situational tactic. Human-operated drones require a continuous radio frequency (RF) link. When jamming severs that link, a dumb drone falls out of the sky. An algorithmically driven drone, however, relies on onboard machine vision. Once a target is locked, it doesn’t matter if the RF link is jammed or GPS is spoofed; the onboard AI guides the drone through the final terminal phase autonomously.
Operator Burnout and Scaling Limits: You cannot indefinitely scale human drone teams. Operating an FPV drone in combat is highly cognitive, stressful work. Algorithms solve the human bottleneck. The role of the human is shifting from pilot to tactical manager. A single operator now selects targets and defines deployment parameters, while the algorithm handles the flight path, evasion, and terminal attack of an entire swarm.
Machine-Speed HyperwarWhen algorithms take over, the Observe-Orient-Decide-Act (OODA) loop compresses from minutes to milliseconds. We are already seeing the early stages of this deployment:
Ukraine’s Saker Scout: This system uses machine vision and onboard AI in the final phase of a strike, capable of visually identifying over 60 categories of Russian military equipment and striking without an operator link.
Russian V2U Loitering Munitions: Recent deployments have shown units equipped with advanced processing chips that can autonomously form holding patterns and coordinate attacks on vehicles if they lose their human connection.
The “Autonomy Trap”: The profound risk now is algorithmic escalation. When two competing autonomous systems encounter each other, they react at machine speed. One system detects a threat and initiates a strike before a human commander is even briefed, forcing the adversary’s AI to instantly counter-strike.
To see why militaries are so eager to hand control over to the algorithm, we have to look at how these systems behave under stress compared to a human operator:The Strategic Restructuring
This reality is forcing a massive systemic restructuring of how global militaries fund and organize themselves. The era of treating AI as a quirky tech-startup experiment is over.
Look at the Pentagon. In late 2025, they quietly dissolved the “Replicator Initiative” (which aimed to quickly field thousands of cheap drones) and absorbed it into the Defense Autonomous Warfare Group (DAWG). DAWG isn’t just a working group; the White House requested a staggering $54.6 billion for it in the FY27 budget—a 24,000% increase.
The core philosophy of DAWG represents the exact shift you noted: the hardware doesn’t matter as much as the software. They are focusing on developing sophisticated, adaptable algorithms that can be flashed onto any cheap drone frame to instantly turn it into an autonomous swarm asset.
Warfare has ceased to be a contest of platforms. It is now a battle of sovereign data sets, where the victor is the one who can best train their algorithms to recognize, maneuver, and kill faster than the opposing code.
–
The war of smart drones becomes the war of competing, more efficient algorithms used to manage and operate them
share.google/aimode/4Jr57PUC…— Michael Novakhov (@mikenov) Jul 7, 2026