Public transcript

Marc Andreessen: The Old Jobs Are Never Coming Back

By: Joe Lonsdale Clips
Length: 37 min
Language: English
Text from: the video's captions
Transcribed: 13 October 2026

Read the full transcript Open on YouTube

Notes in English.

Summary

Marc Andreessen argues the US should abandon nostalgia for old manufacturing jobs and instead lead the coming robotics and AI-hardware revolution, framing it as a national security necessity against China. He traces protectionist industrial policy from Hamilton through McKinley's 'American system,' explains Silicon Valley's left-wing politics through the trait of openness, and diagnoses urban dysfunction as the middle class being squeezed out of knowledge-economy cities. He closes by attacking the precautionary principle and the AI alignment field, arguing they have drifted from genuine risk into social engineering and moral panic.

Key points

Andreessen argues the US should not try to bring back old manufacturing jobs but instead build the manufacturing jobs of the future around AI, robotics, drones, and self-driving cars.
He says AI is effectively a two-country race between the US and China, with the US currently far ahead while Europe regulates itself out of the game.
He traces a protectionist tradition from Alexander Hamilton through William McKinley — the 'American system' — noting McKinley later pivoted from high tariffs to reciprocity and free trade as exporters grew.
He attributes Silicon Valley's left-wing politics to the personality trait of openness, and argues illiberalism is the 'full realization' of liberalism via the paradox of tolerance.
He describes cities as magnets for elite knowledge workers and a low-paid service underclass, squeezing out the middle class, citing Bay Area cops commuting three hours each way.
He frames current dysfunction as policy choices, citing Nixon's 1971 Project Independence for a thousand nuclear plants being killed by the EPA and the Nuclear Regulatory Commission.
He calls the precautionary principle — invented by the German Greens in the 1970s to stop nuclear power — a catastrophic mistake now being applied to AI, and warns against proposals like chip kill switches and even air strikes on data centers.
He argues housing, health care, and education prices explode because technology hasn't touched them and government restricts supply while subsidizing demand, contrasting with LASIK where prices fall like Moore's law.
He criticizes the shift from 'AI risk' to 'AI alignment' as turning a life-or-death question into a morality and politics question — social engineering.

Questions it answers

00:00Old jobs vs. future manufacturing

Can the US bring back the old manufacturing jobs?

He says those jobs aren't cost-effective to bring back; instead the US should lean into designing and building the new things — robots, drones, cars — in highly automated factories, which could generate tens or hundreds of millions of jobs.

AI is a US-versus-China race; Europe has regulated itself out.
Robotics could be the biggest industry in history.
Build 'alien dreadnought' factories, not human screw-driving lines.

03:54McKinley and the American system

What is the historical precedent for protectionist industrial policy?

The American system, originating with Hamilton, used protectionism to build an industrial ecosystem; McKinley ran on high tariffs then pivoted to reciprocity to open foreign markets for exporters — a playbook Germany, Japan, Korea, and China later copied.

Hamilton beat Jefferson's vision of a rural America.
The Second Industrial Revolution (1870–1930) created the modern world.
McKinley used tariffs as leverage toward free trade.

08:56Silicon Valley's politics

Why did Silicon Valley become so left-wing and illiberal?

He attributes it to the personality trait of openness, which maps to the political left, and says illiberalism is liberalism taken to its extreme via the paradox of tolerance.

The Valley is ~99% Democrat by donations and voting.
Maximal tolerance requires intolerance of the intolerant.

11:01Dysfunctional cities

What went wrong with American cities?

Knowledge work concentrates growth in cities, attracting elite professionals and a low-paid underclass while squeezing out the middle class, who face long commutes and unaffordable housing.

Palo Alto cops can commute three hours each way.
The middle class is ejected to a deindustrialized countryside.

15:43Policy choices and optimism

Are these economic outcomes inevitable?

He cites Nixon's 1971 Project Independence — a thousand nuclear plants by 2000 — killed by the EPA and Nuclear Regulatory Commission, showing these are choices, and says current opportunities exist in nuclear, minerals, rare earths, and new cities like the Solano County project.

Nuclear would have sped up electric cars and cut Middle East dependence.
California Forever bought four times Manhattan's acreage in Solano County.

17:44Europe and China

Where does Europe stand on AI, and how does China change US policy?

Europe seems determined to be left behind — only Spotify remains among top tech companies — while in Washington, China's presence is the only thing that makes tech policy conversations rational.

The EU would happily shut down AI, social media, and the internet.
China as an external threat glues America together.

21:28Precautionary principle and AI doom

What's wrong with the precautionary principle applied to AI?

The German Greens invented it in the 1970s to stop nuclear power — a catastrophic mistake — and applying it to AI leads to proposals like chip kill switches, monitoring agents, air strikes on data centers, and risking nuclear war, i.e., a totalitarian superstate.

Stopping nuclear power links to Europe funding Russia's Ukraine invasion.
We're in a moral panic; rationality has left the room.

29:34The American dream's cost disease

Why do housing, health care, and education keep getting more expensive?

Technology hasn't touched these sectors and government restricts supply while subsidizing demand, so prices explode; LASIK shows the alternative — outside the system, prices fall like Moore's law while quality rises.

Health care is 20% of the economy, heading toward 50%.
Subsidizing demand with fixed supply drives prices up.

Notes

AI and the Jobs of the Future

Andreessen rejects the traditionalist view of bringing back old manufacturing jobs; they aren't cost-effective to return.
The US is the clear leader in AI, though many forces — including over a thousand proposed state laws and a near-passed California ban — threaten that position.
AI today is software (ChatGPT as an app), but the next turn is embodied physical AI: drones already fly autonomously, cars are becoming rolling computers (he recommends a Waymo ride in San Francisco), and Elon's Optimus points to general-purpose humanoid robots.
He believes robotics could be the biggest industry in history — billions of robots — and the US should build 'alien dreadnought' factories rather than human screw-driving lines, generating tens or hundreds of millions of jobs and a third or fourth industrial revolution.

McKinley and the American System

The American system originated with Alexander Hamilton against the British Empire's industrial dominance; Hamilton won his debate with Jefferson over an industrial versus rural America.
The Second Industrial Revolution (roughly 1870–1930) invented nearly everything we consider the modern world; McKinley's presidency catalyzed America becoming the world's industrial superpower, enabling victory in the World Wars.
McKinley ran on high, permanent tariffs, then switched to 'reciprocity' — using tariffs as leverage to get others to lower theirs. Germany, Japan, Korea, and now China have run the same playbook.

Silicon Valley Politics and Cities

He explains the Valley's left-wing politics via the Big Five trait 'openness' — highly open people cluster on the political left; the Valley is ~99% Democrat by donations and voting.
Illiberalism follows from the paradox of tolerance: to maximize tolerance you must be intolerant of the intolerant.
Knowledge work slams economic growth into cities, creating a top-plus-bottom political structure — elite professionals plus a clientele underclass — that squeezes out the middle class (e.g., Palo Alto cops commuting three hours each way).

Policy Choices, Europe, and China

These outcomes are choices, not impersonal forces: Nixon's 1971 Project Independence proposed a thousand nuclear plants by 2000, but the EPA and NRC he created prevented any from being built.
Europe seems 'enthusiastic about being left behind' — of the top 50 tech companies, only Spotify remains European; the EU would happily shut down AI, social media, and the internet.
China is the 'dog that didn't bark': the same Washington people who loathe US tech become rational the moment China is on the table.

Precautionary Principle and AI Doom

The German Greens invented the precautionary principle in the 1970s to stop nuclear power — a catastrophic mistake linked to Europe funding Russia's Ukraine invasion and spiking coal emissions.
Applying it to AI leads to proposals like registered chips with monitoring agents, government kill switches, Time magazine's call for air strikes on rogue data centers, and risking nuclear war — a path toward a 'global totalitarian superstate.'

Housing, Health Care, Education

The American dream has three components — house, health care, education — and all three have exploding prices while everything tech touches collapses in price.
Cause: technology hasn't touched these sectors, and government restricts supply (zoning, certification) while subsidizing demand.
LASIK is the counterexample: outside the system, paid by consumers, prices fall like Moore's law while quality rises.

AI Alignment

The field began as 'AI risk' (killer robots) and shifted to 'alignment' ~10 years ago — alignment with whose values? It became social engineering and politics; the original risk researchers propose renaming the field 'AI not kill everyoneism.'

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