Ukraine's 2025 Drone Strike Results List Top Units, Detail Russian Destruction

It was a very productive year for Ukrainian drone units. In a new report, the country's defense minister detailed the catalog of destruction against Russian forces, as well as naming the Ukrainian units with the best records. 

Almost as important as the damage and death delivered, Ukraine is now demonstrating that its forces can verify the data they receive, thereby allowing for optimal resource allocation and strategic decision-making. JL

Stefan Korshak reports in the Kyiv Post:

Ukrainian FPV and bomber drones carried out 819,737 successful strikes in 2025, all confirmed by drone video. Ukrainian drone pilots killed or seriously injured 240,000 Russian soldiers and knocked out or severely damaged 29,000 heavy weapons, 62,000 automobiles and ammunition storage sites. The highest-scoring drone unit of the year was 414th Unmanned Systems Brigade “Birds of of Madyar” Other high-scoring units included the national intelligence agency’s Strike Unit Alpha, the National Guard’s Lasar Group, and the drone unit of the 3rd Assault Brigade. "For the first time, a fighting army is receiving real, verified data from the battlefield that can be used for management decisions.”

Russia "Declining Major Power" Given "Extraordinary Price, Minimal Gain" In Ukraine

A new assessment, released yesterday by the Center for Strategic and International Studies, a leading US military think tank, reports that Russia's anemic performance against Ukraine has produced the slowest advance of any major campaign in the past century, including the infamous WWI Battle of the Somme, long considered the benchmark for modern military futility.

The extraordinary price Russia has paid in casualties and equipment lost is greater than any other major power in the past 80 years. And for all that, it has achieved none of its strategic goals while appearing unlikely to be capable of doing so. As a result, the Kremlin's prospects of conquering Ukraine are now considered nil. To make matters worse, the effort required to deliver even these pathetic results have virtually bankrupted Russia while causing as many as one million Russians to emigrate - many of them younger, technologically savvy and the only demographic capable of driving the country into the AI age. As a result, the report concludes that Russia is in decline as a major power. JL

Seth Jones and Riley McCabe report in the Center For Strategic and International Studies:

The data show Russia is paying an extraordinary price for minimal gains and is in decline as a major power. Since February 2022, Russian forces have suffered 1.2 million casualties, more losses than any major power in any war since World War II. (Since) 2024, Russian forces have advanced at an average rate of between 15 and 70 meters per day in their offensives, slower than almost any major campaign in any war in the last century. Meanwhile, Russia’s war economy is under mounting strain, with manufacturing declining, slowing growth of 0.6 percent in 2025. Russia’s progress on the battlefield, especially in the last two years, falls decisively short of Moscow’s goal to militarily conquer Ukraine.

Billion-Dollar AI Startups With No Products, No Revenues, Eager Investors

We may have entered the final, truly absurd AI end-game. Think dotcom sock puppets. Dozens of startups with no product or even prospect of revenues are being expansively funded by investors - some of them ostensible professionals from the venture community - who believe that AI's biggest challenges can now be overcome by researchers in their early 20s because - and here's where we may agree - leading models like ChatGPT and Claude have hit a dead end. 

The new entities are called neolabs. Their proposition is that pure AI research is where the really big payoff is going to be delivered. And, who knows, maybe they are right. But it feels more like the AI wealth creation window is beginning to close, that both young techies and investors are realizing it and that they are all trying to cash in before the ridiculously easy money dries up. JL

Kate Clark reports in the Wall Street Journal:

A new wave of startups some have dubbed “neolabs,” which give priority to long-term research and developing new AI models over immediate profits. Interest in neolabs has skyrocketed as investors hunt for the next OpenAI, which began as a research lab and became one of the world’s most valuable startups. A large subset of top AI researchers believe that models like ChatGPT and Claude have hit a dead end and will never reach a level of intelligence that matches or exceeds humans. At the same time, researchers are realizing that the opportunity to raise big VC dollars quickly and easily may not last long. Some neolabs have seen their valuations soar into the tens of billions of dollars, prompting critics to suggest that most of them will have slim odds of turning a profit or launching a winning product.

A Ukrainian Ground Drone Captures 3 Russian Soldiers

Russian soldiers have already surrendered to Ukrainian aerial drones, so seeing them do so to a ground drone is an evolutionary rather than revolutionary event. 

That said, facing a tracked vehicle mounted with a machine gun and demanding your surrender or death must be an intimidating experience. JL

Vladyslav Khomenko reports in Militarnyi:

The Ukrainian Droid TW-7.62 ground robotic system, based on the NUMO platform, has captured three Russian soldiers. Droid TW-7.62 is a reconnaissance and strike ground robotic system adapted for the KT-7.62 (PKT) machine gun. The footage shows three Russian soldiers approaching one by one, removing their equipment, and lying down next to the UGV. It is equipped with a ballistic computer that ensures high accuracy and effective target engagement. Additionally, it has artificial intelligence elements for autonomous detection, capture, and tracking of targets.

Even In the Drone War, Winter Favors Ukrainian Defenders

Drones, especially those with thermal cameras, are particularly effective in winter because they can human and vehicle tracks in the snow. 

Attacking is generally harder in winter, as soldiers have learned over millennia, but drones make it even harder to assault in the snow and cold. JL

Cassandra Vinograd and Oleksandr Chubko report in the New York Times:

Winter tends to favor defenders, a view echoed by Ukrainian soldiers. It makes troops more exposed to drones and renders any movement dangerous. Footprints in the snow are easy to spot from above. Lower temperatures make thermal cameras on drones more effective. “We can clearly see tracks in the snow, where they lead, and identify positions where the enemy is hiding.” Propellers on some drones are prone to freezing, but  rubbing simple meat fat on the propellers has proved most effective, providing a layer of protection.

Russia's Kostiantynivka Offensive Ends Before It Begins As Ammo Depots Destroyed

Knowing that the Russians were planning a new offensive against Kostiantynivka, Ukrainian forces targeted the nearest Kremlin ammunition storage depots for each prong of the attack, using missiles and drones, as well as the strategic reserve depot 100 kilometers behind the front. 

As a result of those Ukrainian attacks, 200,000 tons of Russian ammunition went up in explosions and fires so devastating that a number of civilian evacuations were required. The result of that loss to Russian forces is a delay of approximately three months until stocks can be resupplied. JL

RFU News reports:

The Russians launched an offensive on Kostiantynivka, which was meant to be the next major effort i Donetsk. But the Russians' plan fall apart before the battle had even started. Ukrainian forces struck an ammunition depot in Russian-controlled Debaltseve. A powerful explosion was followed by large-scale fires and secondary detonations throughout the night. Ukraine also struck Russian ammunition at a strategic depth, targeting the Arsenal of Russia’s Missile and Artillery Directorate in Kostroma. The strike caused a fire significant enough to trigger civilian evacuations.

AI Savant Yann LeCun's New Firm Is Betting Against Large Language Models

Yann LeCun is considered one of the true intellectual and business leaders in tech generally and AI specifically. So when he recently quit as the head of Meta/Facebook's AI effort and announced the formation of a new company dedicated to advancing AI but not focusing on large language models, it understandably attracted a lot of attention. 

His point is that LLMs are limited and that the truly forward-looking and thinking AI models need a different approach in order to effectively perceive and navigate. To what degree he is correct remains to be seen, but the larger implication is that concerns about the current strategy for AI are well founded in the core of their science and math, so cannot be glossed over by the VCs and others who remain so quick to dismiss critics. JL

Caiwei Chen reports in MIT Technology Review:

Yann LeCun, a Turing Award winner and top AI researcher, has long been a contrarian in tech. He believes that the industry’s obsession with large language models is wrong. He thinks we should be betting on world models—a different type of AI. He is also a staunch advocate for open-source AI and criticizes the closed approach of  OpenAI and Anthropic. "The truly difficult part is understanding the real world. What’s easy for us, like perception and navigation, is hard for computers. LLMs are limited to text. People have this delusion that it is a matter of time until we can scale them up to having human-level intelligence, and that is false. They can’t reason or plan. They can’t predict the consequences of their actions."