Public transcript

Eric Schmidt on AI, the Battle with China, and the Future of America

By: Zenvora Productions
Length: 29 min
Language: English
Text from: the video's captions
Transcribed: 14 October 2026

Read the full transcript Open on YouTube

Notes in English.

Summary

Eric Schmidt, former Google CEO and executive chairman, joins a wide-ranging conversation covering the AI revolution, competition with China, drone warfare, space investing, and American strength. He argues the AI revolution is underhyped, that China is competing through open weights and applied AI rather than AGI, and that cheap drones are transforming warfare. He rejects near-term superintelligence timelines, estimating AGI-type breakthroughs six or seven years out, and closes by urging America to use its capital markets, universities, and entrepreneurs to win.

Key points

Eric Schmidt believes the AI revolution is underhyped and that agents will become powerful and start working together.
China is not pursuing AGI strategies but instead applying AI to everything, competing with open weights and open training data while the US focuses on closed models.
DeepSeek's R1 reasoning model achieved reinforcement learning with lower numeric precision (eight and even four bit) than typical American 16-bit models.
In warfare, cheap drones (a $4,000-$5,000 drone destroying a $30 million tank) make traditional weapons like tanks obsolete, and future conflict becomes drone versus drone.
AI-driven battle planning via reinforcement learning could create deterrence because neither side can count or see the other's plan.
Declining birth rates (China about 1.0 per two parents, Korea 0.78) mean societies are choosing to depopulate, which Schmidt calls a legitimate concern.
Schmidt disagrees with the San Francisco narrative of superintelligence in three years, estimating six or seven years instead, and notes no evidence of AI setting its own objective function.
AI is 'middle to middle' while humans are 'end to end,' so AI will be synergistic with humans rather than replacing them.

Questions it answers

02:00Work culture and China

Why does Eric Schmidt oppose remote work for young technologists?

He says he is not in favor of working from home for young people, because 20-somethings learn how the world works by listening to older colleagues argue in person. He contrasts this with China's 996 work culture, which is illegal but widely practiced.

Young employees learn from in-person mentorship.
China's 996 schedule is the real competition.

03:40China's AI strategy

How is China approaching AI differently from the US?

China isn't pursuing AGI due to chip restrictions and shallow capital markets; instead it applies AI broadly and competes with open weights and open training data. Schmidt warns most of the world will use Chinese models unless the US proliferates its own, and notes DeepSeek's R1 achieved reinforcement learning at lower numeric precision.

China competes via open weights and open training data.
DeepSeek R1 uses eight and four bit precision versus American FP16.

07:50Relativity Space

Why did Schmidt invest in Relativity Space?

He says rockets are "really cool and really hard," an art and a science, with 80% of weight being propellant. Relativity competes in low Earth orbit satellite launches and its order book is full.

Rockets are far less mature than jet engines.
Relativity's LEO launch order book is full.

10:10Drone warfare

How does Schmidt think drones will change warfare?

Cheap drones (a $4,000-$5,000 drone destroying a $30 million tank) make traditional weapons obsolete. Future war is drone versus drone, with drones forward and people behind; AI-generated battle plans that neither side can see could create deterrence, though Schmidt stresses there are no winners in war.

The kill ratio favors cheap drones over tanks.
Unseen AI battle plans may deter conflict.
Ukraine used USVs to destroy Russia's Black Sea fleet.

17:40America and birth rates

What does Schmidt think about Western decline and demographics?

He calls declining birth rates a legitimate concern — China at about 1.0 per two parents, Korea at 0.78 — and favors immigration as part of the solution. He wants America to win using its deep capital markets, universities, and entrepreneurs.

Humans are collectively choosing to depopulate.
Immigration can help address the problem.
"Don't screw it up, guys."

22:10AGI and superintelligence

Where does Schmidt stand on AGI timelines?

He rejects the three-year superintelligence timeline, suggesting six or seven years, and says general intelligence requires setting one's own objective function, for which there's no evidence. AI is "middle to middle" while humans are "end to end," so AI will be synergistic with humans.

No evidence yet of recursive self-improvement.
Non-stationarity of proofs is a key unsolved problem.
AI will assist rather than replace humans for now.

Notes

Overview

This conversation features Eric Schmidt, former Google executive chairman and CEO, discussing AI, competition with China, drone warfare, space investing, and America's future.

AI and the Competition with China

Schmidt believes the AI revolution is underhyped; agents will become powerful and start working together.
China is not pursuing AGI strategies, partly due to US chip restrictions and shallow capital markets; instead it applies AI to everything, including consumer apps and robotics.
China competes with open weights and open training data, while the US focuses on closed weights and closed data, meaning much of the world (the "belt and road" countries) may adopt Chinese models.
DeepSeek's R1 reasoning model did reinforcement learning with lower numeric precision (eight and four bit) versus the typical American 16-bit (FP16) training.
Sam Altman indicated a small open-weights model, much smaller than the o3 model, that can fit on a phone.

Work Culture

Schmidt is against working from home for young people, arguing 20-somethings learn from older colleagues in person; he jokes he favors work-life balance "for people who work for the government."
China's "996" culture (9am to 9pm, 6 days a week) is technically illegal but widely practiced.

Space and Relativity

Schmidt invested in Relativity Space because "rockets are really cool and really hard"; roughly 2% of a rocket's weight is payload, 18% the rocket, 80% propellant.
Relativity competes in low Earth orbit satellite launches with a full order book.

Drone Warfare

Ukraine, with no navy or air force, innovated spectacularly; a $4,000-$5,000 drone can destroy a $30 million tank.
Future war: drones forward, people behind; each side with a million drones using reinforcement-learning battle plans neither side can see, creating deterrence.
Ukrainians used USVs to destroy the Russian Black Sea fleet, protecting grain exports worth 6-10% of their economy.

AGI and Superintelligence

The "San Francisco narrative" predicts superintelligence in about three years; Schmidt says six or seven years.
General intelligence means setting one's own objective function; there's no evidence of that yet, and non-stationarity of proofs is a hard technical problem.
AI is "middle to middle" while humans are "end to end," so AI will be synergistic with humans.

America and Demographics

Birth rates: China about 1.0 per two parents, Korea 0.78; Schmidt favors immigration to help address depopulation.
America's strengths: deep capital markets, universities, entrepreneurs — "Don't screw it up, guys."

Text from the video's own captions. Summary and notes written by AI from the transcript, so check anything important against the recording.