Michael Kofman

1. Key Themes

The Kill Zone Is a Layered System, Not a "Drone War"

The popular framing of a pure drone war misses how the pieces interact. Drones cause most casualties, but they depend on minefields and artillery to function. Kofman: "the statistics out there do show that drones inflict 80% of losses. However, it is really a combination of things." 00:20:33 He explains that artillery "forces the enemy forces to disperse... it makes it hard for troops to concentrate or suppresses them, it slows them down in an assault, making them easy targets for drones" 00:20:56, and that drones fail in rain, wind, and the spring and fall transitions. His conclusion: "if you just had drones, you didn't have anything else, no minefields, no artillery, I don't think Ukraine would be able to hold the terrain this effectively with the low density of forces." 00:21:34

The Real Contest Is Drone Superiority, Not Ground Mass

Both sides fight over who can suppress the other's drone units and push them back from the line. Kofman: "this war really began to be much more of a tug of war over who has superiority in this drone engagement zone, which forces locally call the kill zone. And the side that is able to suppress the other side's drone units... is then able to push the drone support further away from the front line, and then, by displacing it, then be able to actually advance." 00:34:30 Successful Ukrainian offensives follow a recipe: "begin to suppress Russian drone units, begin to interdict logistics behind the drone units so that they don't get sufficient supplies... then having a coordinated strike plan for how you're going to suppress them during the assault." 00:35:10

Drone Economics: Videos Wildly Overstate Hit Rates

Footage creates an anchoring bias because only successes are published. Kofman: "the videos are all videos of successes. There are no videos of misses." 00:30:16 Real ratios: "most... is about 30 to 40" drones per tank, "about 15, maybe 15 to 20" per infantry fighting vehicle, and for a heavily protected Russian tank "last year was 79" 00:31:01. Preston Stewart adds that the one-$300-drone-kills-a-tank perception is false: "we're seeing 10, 20, I think upwards of 30 of these small drones at times being utilized to actually eliminate a tank." 00:29:43 Throughput is enormous: "for FPVs, it's probably somewhere 8 to 10,000 per day... and that would be for one side of the war." 00:24:57

The War Is Still Human-Centric; Autonomy Is Overhyped

Despite headlines, Kofman says: "Most of this war are systems that are remotely operated, but they are operated by people. It is actually quite manpower-intensive." 00:25:34 A typical team needs "two or three people. You have a pilot, you have a copilot-navigator" plus a technician 00:25:56. Autonomy "has been overhyped in terms of the percentage of the fight that's actually being run that way" 00:27:20, limited mostly to terminal guidance that helps beat jamming.

Innovation Cycles of Three to Four Months: Ukraine Innovates, Russia Scales

Kofman: "there's been about three- to four-month cycle of evolution of technology and tactics in this war between Russian and Ukrainian forces" 00:46:38. Ukrainian units behave like startups: "they're kind of all defense startups, they all have kind of their own R&D section, they have their own defense shop... building drones for them." 00:47:27 Russia is slower but "tended to be better at scaling a solution once they identified it... allocate capital, industrial capacity, scale it, deploy it across the force, in a standardized way." 00:47:52 Fiber-optic drones illustrate the pattern: Ukraine experimented first and didn't adopt, Russia scaled them after Kursk, Ukraine then copied them, and they are now roughly "15% or less" of tactical strike drones but impossible to jam or detect by radio.

Russia Has Hit Diminishing Returns on Ground Offense; Pivot to Strategic Bombardment

Kofman: "The Russian military is doing quite poorly this year in the war. For them, this is probably their worst year of combat performance since winter of 2023... the way they've been fighting has reached diminishing returns" 00:09:28, and they "increasingly will face some manpower issues, and their effort now fixates more on strategic bombardment." He warns: "strategic bombardment does not have a great track record... of achieving maximalist political aims in a war, especially if a ground campaign is unsuccessful." 00:09:47 The target set has broadened to "warehouses, data centers, schools, medical facilities, malls... trains." 00:14:02

Ukraine's Advantage Is Real but May Be a Window, Not a Trend

Kofman: "this year for Ukraine is the best year of combat performance for them on the ground since probably about the summer 2023 offensive" 00:08:38, with two localized offensives and gains that in places exceed 2023's. But he cautions: "The debate we have as analysts is is that going to be a lasting advantage, as a secular trend, or is that a window of opportunity for Ukraine?" 01:02:17 Winter infrastructure damage, the Russian strike campaign, and outside shocks like the US war with Iran (which "had a material and political impact on this war" 01:03:52) all remain wildcards.

Decentralized Command as Decisive Culture Edge

Stewart contrasts Western mission command with Soviet-legacy top-down control. In the first weeks, "There was a shocking number of senior Russian officers killed in the opening few weeks of the war, because they kept having to go closer and closer to the front to essentially move traffic jams." 01:23:45 He notes Ukraine's innovation is "driven from the bottom up, which is a very Western military type thing, whereas on the Russian side, it's still, to today, is still top-down driven." 01:24:35

Don't Abstract Military Power; Scenario Context Decides

Kofman warns against rankings: "military power by itself cannot be abstracted... There's no such thing. Military power requires a context to express itself." 01:15:49 Lessons from the war's latter positional phase may not describe how the next war opens, and "knowing something about militaries is not the same as knowing how the two of them will interact in a particular fight." 01:17:04

2. Contrarian Perspectives

Russia's Infiltration Tactic Works, and That's Exactly Why It's Failing Strategically

Most assume a tactic that gains ground is succeeding. Stewart relays a US Marine volunteer's assessment: "He described this as an incredible tactic if you don't care about losing soldiers... You're just going to lose 90, 95% of your guys along the way." 00:33:48 Kofman adds that light motorized attacks "had proven more successful than traditional mechanized attacks" because Ukraine's forward posts are "three-man posts... with considerable separation between them of several hundred meters" 00:38:57, yet he calls it "a grim calculus" that has hit diminishing returns.

The Front Is Not a Stalemate and Isn't Static

Public perception is positional deadlock. Kofman: "There's often a perception that it's just a positional stalemate, but in fact, even more recently, some of Ukraine's offensive operations had gained more terrain than Ukrainian forces were able to do during the summer 2023 offensive, which was planned to be a much bigger operation." 00:13:03 The "line" is a gray zone of pickets: "Ukraine does not hold cohesive defensive lines." 00:19:04

Don't Overlearn from This War's Drone Dynamic

Most commentators rush to declare the end of tanks, aircraft, and the old order. Stewart: "This war represents Russia and Ukraine. It is not representative of what every war in the world would look like... The lack of air superiority on Russia's part plays a major role." 00:50:13 Kofman notes drones "really came into their own more in these latter phases" of attritional war, and that "the next opening phase of a war is often different than kind of the middle phase." 00:55:58

Asking "Who Is Winning?" Is the Wrong Question

Stewart: "It's a question I generally avoid... Nazi Germany was winning that war for the majority of it until they weren't." 00:59:32 Kofman adds that "if you say someone's winning, it often sets for people and certainly any audience expectations. Most people think if you say someone's winning, that that side's going to win." 01:03:04 Analysts also routinely worst-case adversaries, which can mean overrating them.

Drone Mass Doesn't Replace Infantry; It Increases Manpower Needs

Counter to the "robots reduce soldiers" narrative, Kofman says both militaries "had to build out substantial numbers of drone units and channel manpower into the drone component of the fight." 00:25:34

3. Companies Identified

Anthropic

AI lab. Named in the sponsor read as a customer of Fin. Lex Fridman: "trusted by over 6,000 customer service leaders and top companies including Anthropic. Yes, Anthropic." 00:03:12 Also: "You know when an AI company trusts another company to do an AI thing for it, they're legit." 00:03:28

Fin

AI customer service agent company. Fridman: "the number one AI agent for customer service, 65% average resolution rate," built "to handle complex, multi-step queries like returns, exchanges, disputes." 00:03:39

GiveDirectly

Nonprofit sending cash directly to people in need. Fridman: "in Kenya, for example, every $1 that GiveDirectly delivered grew the local economy by $2.50." 00:02:39

WhisperFlow

AI voice dictation app. Fridman: "makes like an order of magnitude less mistakes than any system that I've tried before." 00:05:55

Shop (Shop skill, shop.md)

Lets AI agents buy things for users. Fridman: "your AI agent to buy stuff for you." 00:06:38

Perplexity

Knowledge exploration tool; Fridman used it live for the Prigozhin mutiny summary. 01:26:35

LMNT

Zero-sugar electrolyte mix, a daily staple for Fridman. 00:04:08

Wagner Group

Russian private military company that fought at Popasna and Bakhmut, recruiting from prisons. Kofman: "Prigozhin's given the authority by the Russian government to go recruit from Russian prisons. And he starts hiring convicts to build out special, convict-staffed assault formations." 01:27:52

CODE 9.2 and 1st Separate Assault Regiment (Ukraine)

Advanced Ukrainian drone assault units. Kofman: "using a combination of drones and very well-equipped and adapted mechanized formations to try to return maneuver to the battlefield." 00:36:17

Ukrainian Air Assault Brigades

Running a creeping offensive in Zaporizhzhia. Kofman: "three-man fire teams with drone support, systematically clearing tree line by tree line by tree line, slow and steady, but keeping their casualties low." 00:36:56

Ukraine's 3rd Corps

Conducted a successful offensive operation near Lyman. Kofman: "Ukraine's 3rd Corps having conducted a successful offensive operation." 00:18:46

1st National Guard Corps Azov

Ukrainian formation in the Dobropillya direction; the US Marine volunteer Stewart cited fought with it. 00:33:48

DJI

Chinese maker of the Mavic drones still used by both sides. Stewart: "relatively expensive compared to some other systems that Ukraine has. And those are not manufactured in Ukraine." 00:41:59

Tsukorok Drone Detector (Ukraine)

Cheap, handheld, multi-antenna detector that intercepts analog drone video. Stewart: "it looks like an old Game Boy, it has a screen on it... it is the enemy's view of the drone." 00:48:31

4. People Identified

Preston Stewart

West Point graduate and Afghanistan veteran; runs a YouTube channel, "War and Conflict" podcast, and the Defense Tech News newsletter. Fridman: "covers the war extensively." 00:00:06 He is the source of the front-line, tactical and Western-doctrine perspective.

Senior fellow at the Carnegie Endowment, host of The Russia Contingency podcast. Fridman: "a world-class military analysis expert specializing in the Russian and Ukrainian militaries." 00:00:22 He candidly flags what he got wrong: "we overrated them, which is a tendency when looking at adversaries." 01:19:09

Vladimir Putin

Kofman ties the invasion's flawed assumptions to him: "Vladimir Putin did not see Ukraine as a real independent country with a separate national identity, and really had not learned or updated his views at all." 01:14:08 He cites Putin's July 2021 article as the clear statement of those assumptions.

Yevgeny Prigozhin

Wagner chief who led the June 2023 mutiny. Kofman: "rose to prominence... from outspokenness and public criticism and the conflict that he developed with the Minister of Defense at the time, Sergei Shoigu, and the Chief of General Staff, Valery Gerasimov." 01:26:56 Stewart sees the march on Moscow as a possible trigger for "a Russian inability to continue to project power into Ukraine." 01:01:26

Sergei Shoigu and Valery Gerasimov

Russian Defense Minister and Chief of General Staff at the time, targets of Prigozhin's criticism. 01:27:24

Kim Jong Un

Stewart mentions a photo of him "walking through a hallway of stacked FPVs." 00:24:43

Volodymyr Zelenskyy

Mentioned in the episode title; not discussed in the portion of the transcript provided.

Unnamed US Marine Volunteer (Azov)

Provided the 90–95% loss-rate assessment of Russian infiltration. 00:34:08

William Gibson

Kofman quoted him: "as Gibson used to say, 'The future is already here, it's not evenly distributed.'" 00:59:00

James Holland

Historian of WWII Fridman referenced when discussing levels of analysis. 01:04:16

5. Operating Insights

Let the Field Experiment, Then Standardize the Winner

Ukraine's decentralized units each ran their own R&D and tactical experiments, producing faster innovation; Russia was slower but scaled what worked. Kofman: Russia "would then allocate capital, industrial capacity, scale it, deploy it across the force, in a standardized way." 00:47:52 The takeaway for operators is to pair bottom-up experimentation with a strong central capability to scale proven solutions.

Push Decision Rights Down by Making Intent Explicit

Stewart's core Western-military lesson: communicate "This is the mission, and this is what we're trying to accomplish, so they have a full understanding. So, then when something inevitably goes sideways, they're not stuck." 01:23:22 Decentralization only works with training and shared intent; otherwise senior leaders become traffic cops, as with the Russian officers killed moving traffic jams.

Beware Survivorship Bias in Your Metrics

Public drone videos made hit rates look like 1:1 when reality is closer to 5:1 or far worse. Kofman: "you don't know what percentage of hits or percentage of drones launched that really is." 00:30:16 Operators should ask what the denominator is before trusting a vivid success story.

Build Strategy Around Explicit Assumptions and Revisit Them

Kofman: "any good strategy must have core planning assumptions, and if those assumptions are proven wrong, you should expect to adjust. Actually, a lot of times strategy goes poorly is when people are trying to execute a strategy even though their assumptions are clearly wrong." 01:09:42 Russia's 2022 plan had "no hedge" or Plan B.

Strategy Without Tactical Feasibility Is Just a Staff Exercise

Kofman: "If strategy is divorced from tactics, then it's a great staff-planning exercise, it's maybe very well-written, but the force can't execute that strategy... alternatively, if you just have tactics... you haven't thought through how you're going to win it." 01:10:05

6. Overlooked Insights

Cheap Detectors Turn Enemy Drone Feeds into Early Warning, and Fiber-Optic Drones Defeat It

The Tsukorok-style detector exploits the analog video signal strength once an enemy drone is closer to you than to its operator. Kofman: "with a fiber-optic cable drone, you have nothing in terms of warning cuz you can't receive that signal." 00:46:38 This means the shift to fiber optics doesn't just beat jamming; it also removes the passive detection layer nearly every soldier carries, a quiet threat to the entire counter-drone stack and a market signal for new detection approaches.

Scaling Complexity Is the Hidden US Military Advantage, and It's Untested Against Drone-Saturated Battlefields

Kofman's offhand business analogy captures the real moat: "The ability to scale operations, complex operations, is a big limiter for other militaries... it's one thing to run a business with 10 employees. It's another thing to run a corporation with 10- or 20,000." 00:57:40 He pairs it with the unresolved question of whether "our traditional advantages will offer the same benefits in a future war if that war features these technologies," noting that opponents like Russia already plan to fight this way. That uncertainty about whether expeditionary airpower-centric combined-arms advantages survive in a drone-dense environment is where much of the future defense-tech opportunity sits.