Eric Schmidt on the war in Ukraine, AI and how Russia is innovating

AI summary

What is Schmidt’s overall argument about AI and adaptation in Ukraine? In an interview with Richard Engel, who is reporting from Ukraine, Eric Schmidt argues that the war has become a fast-moving technological contest in which adaptation matters as much as invention. His central claim is that Ukraine’s earlier advantage as a small, fast innovator is being threatened because Russia is now adapting and scaling military technology much faster than before. Schmidt says the battlefield is increasingly defined by drones, anti-drones, glide bombs, ballistic missiles, communications networks, and automation. He frames the conflict as a race of “innovation to scale,” where the side that improves and mass-produces new systems fastest gains the advantage. - Relevant source moment: [34:04] Schmidt says Ukraine used to be “small, really smart, really fast,” while Russia was “big, slow,” but that has changed. How does Schmidt describe Russia adapting its military technology? Schmidt says Russia has adapted by upgrading Iranian-designed Shahed drones with small jet engines, making them faster than current Ukrainian interceptors. His point is not just that Russia has a new weapon, but that each successful Ukrainian defense creates a Russian countermeasure, forcing Ukraine to adapt again. He also describes Russia as building drone-to-drone mesh communications, allowing longer-range control or guidance. He cites a strike on an SBU building in Kyiv as an example of what he believes this kind of network may enable, though he presents some details as inference rather than independently proven fact. - Relevant source moment: [05:31] Schmidt explains that Russia has added jet engines to Shahed drones, forcing Ukraine to develop faster interceptors. What role does AI actually play, according to Schmidt? Schmidt draws a sharp distinction between automation, computer vision, and what he considers true AI. He says many systems being called “AI weapons” are really automated targeting tools: a camera recognizes a target, locks on, and guides the weapon in the final phase.

In his account, true AI would involve learning from battlefield data and improving tactics or algorithms over time. He says he has not yet seen that fully on the battlefield, but expects it soon because Russia has video feeds, mesh networks, and performance data that could be used to train better systems. - Relevant source moment: [22:31] Schmidt says AI means learning and prediction, and that what has appeared so far is mostly automation. What limitations or uncertainties does he identify for AI in warfare? Schmidt says the strongest current examples are not necessarily AI in the technical sense, but computer vision and automated final targeting. He emphasizes that a human-in-the-loop or human-on-the-loop model still matters legally and morally, especially under U.S. and expected NATO rules. He also repeatedly marks uncertainty. He says he cannot validate some public reports about Russian use of large language models or Chinese assistance, and he avoids confirming specific battlefield performance claims about drones supplied by his company. - Relevant source moment: [26:50] Schmidt says battlefield learning systems have not yet been seen clearly, but he expects them soon. What evidence and examples does Schmidt use to support his warning? Schmidt describes a layered battlefield: cheap FPV drones at the front, KAB glide bombs, Shahed drones, jet-powered Shaheds, and ballistic missiles. He says FPV drones may cost hundreds of dollars, while ballistic missiles can cost millions, creating a hierarchy of range, destructive power, and scarcity. He cites several specific claims and examples: Russia allegedly producing around 1,300 ballistic missiles a year; KAB glide bombs carrying up to 3,000 kilograms; a public claim that Russia may have readied 80,000 drones or related systems between now and December; and a Russian Starlink-like system called Rusvet that he says may become operational next year. He is careful to say some figures come from open reports and that he does not know whether all are true.

- Relevant source moment: [10:57] Schmidt lays out five reasons he thinks the war may get worse, including missiles, jet drones, glide bombs, mobilization, and satellite-like communications. What distinctive insight does he offer about adaptation? Schmidt’s most important insight is that the war is not simply a story of new weapons, but of rapid counter-adaptation. He says innovations that work for a few months are then blunted, copied, or bypassed by the other side. He argues that this cycle is now happening in days and months, not years. In his view, AI could make the cycle faster by accelerating design, targeting, learning, and operational planning. - Relevant source moment: [29:58] Schmidt says this is a war of innovation “in days, not years,” with each side adapting within months. What does he predict about automated warfare? Schmidt predicts that automated drone-on-drone warfare will become common on both sides. He expects systems to roam, identify enemy drones or other targets, and attack them with less direct human authorization, though he distinguishes this from fully learning AI. He also predicts that humans will not be able to control large drone swarms directly, while computers can. His more speculative forecast is that reinforcement-learning systems could eventually generate swarm war plans, perhaps creating a form of AI deterrence if both sides fear the other’s unknown algorithms. - Relevant source moment: [23:34] Schmidt says humans cannot run a swarm of a thousand drones, but computers can. What practical implications does Schmidt draw? Schmidt says Ukraine needs money and technological help quickly to counter the next waves of Russian adaptation.

He also argues that Western countries and NATO are not absorbing enough battlefield knowledge from Ukraine, even though Russia may be a future opponent. He suggests that the West could especially help Ukraine with the AI side, because Ukraine is fighting a war while also trying to understand a difficult technical field. His practical message is that delay matters because the technological balance is changing quickly. - Relevant source moment: [14:04] Schmidt says Ukraine can hold if it receives Western money and technological help. What caveats and open questions remain? The interview cannot independently verify Schmidt’s claims about Russian production numbers, Chinese model support, Rusvet’s timeline, or the exact effectiveness of particular Ukrainian interceptor drones. Schmidt himself declines to discuss operational specifics and says some public figures may not be true. There is also an unresolved legal and moral question about how “human-on-the-loop” systems will be governed once weapons act faster than human decision-making. Schmidt believes NATO will preserve limits, but he doubts Russia and China will follow similar rules. Key takeaways Schmidt says Ukraine is still highly capable, but its early innovation edge is under pressure because Russia is adapting faster and scaling more effectively. He sees the war as a preview of future conflict: drones, counter-drones, automation, mesh communications, and eventually learning systems. His answer on AI is precise: much of what is now called AI warfare is automated computer vision, while the more dangerous next phase would be systems that learn, coordinate swarms, and improve tactics from battlefield data. His warning is that Ukraine and its supporters must adapt as fast as the technology does.