Two Heavy Hitters, Same Conclusion

Goldman Sachs CEO David Solomon published an op-ed in The New York Times calling the 'AI job apocalypse and mass unemployment' narrative 'overblown.' Around the same time, David George, General Partner at Andreessen Horowitz (a16z), published a detailed essay on May 6, 2026 calling the entire AI apocalypse framing 'a complete fantasy' — and backing that up with actual economic reasoning.

The Actual Argument

George's a16z essay — reported on by Fortune on May 7, 2026 — is built around a concept economists have been explaining for well over a century: the lump-of-labor fallacy.

The fallacy assumes there is a fixed amount of work to be done. If AI does more of it, humans must do less. Zero-sum. Game over.

The problem: that's NOT how economies work. Human wants and needs are not fixed. They expand.

George invokes John Maynard Keynes, who predicted in the 1930s that automation would produce a 15-hour work week. Keynes was wrong. Automation created more productivity, which created more wealth, which created entirely new categories of work that didn't exist before.

According to the a16z essay, the same dynamic applies here. When the cost of a powerful input — like cognition — falls, 'costs fall, quality rises, speed rises, new products become viable, and demand moves outward.' That's Jevons Paradox. When energy got cheap, we didn't just fire whalers. We invented plastics.

The Marc Andreessen Point Nobody Is Covering

ZeroHedge noted that a16z co-founder Marc Andreessen has been making a related argument for months: AI is arriving precisely when demographic collapse makes it necessary.

His quote: 'We're going to have AI and robots precisely when we actually need them — with populations shrinking — to keep the economy from actually shrinking.'

This angle is almost entirely absent from mainstream coverage. Most outlets frame AI job fears in isolation, ignoring the demographic winter playing out across the developed world. Birth rates are collapsing. Labor forces are shrinking. Japan, South Korea, Germany, and yes, the United States are all staring at serious workforce shortfalls in the coming decades.

AI isn't the threat to the labor market. A shrinking labor force is the threat to the labor market. AI might be the answer.

What the Doomers Are Actually Saying

The other side of this debate includes Pope Leo XIV, who on Monday warned that AI and the digital economy could produce 'new forms of slavery' and mass job losses. Senator Bernie Sanders has pushed for halting data center buildouts — a policy that, as ZeroHedge pointed out, would effectively cede compute power advantage to China.

The strategy being floated to 'protect' American workers would hand Beijing a structural lead in the defining technology of the next century.

What Solomon and a16z Both Acknowledge

Neither Solomon nor George are claiming AI disrupts NOTHING. They're not that naive.

George explicitly writes that 'AI will absolutely eliminate some tasks and compress some roles' and that 'there's some evidence that may already be happening.' Solomon, per ZeroHedge's summary, acknowledges that transitions will require workers to shift to higher-value tasks and new roles managing, implementing, validating, and regulating AI systems.

The argument is NOT 'AI changes nothing.' The argument is: economy-wide, permanent unemployment is not what history predicts, and not what the data shows.

Ben Horowitz, a16z co-founder, pointed out on the Invest Like the Best podcast earlier this year that AI technologies have been advancing since at least 2012 — when ImageNet changed computer vision — and the catastrophic job destruction still hasn't arrived. That's 13-plus years of waiting for the apocalypse.

Framing the Debate

Most mainstream outlets frame this debate as 'tech bros vs. worried workers.' That framing is lazy and dishonest.

The lump-of-labor fallacy is not a tech-bro invention. It's a foundational concept in academic economics. The historical track record of 'this technology will end work' predictions is uniformly bad — steam engines, the printing press, the tractor, the personal computer. Every single time, new categories of work emerged.

Coverage also largely ignores the demographic context. Reporters treating AI job fears as self-contained miss that the U.S. and its allies face a labor shortage problem, not a labor surplus problem, over the next 30 years.

And few voices in mainstream coverage are seriously engaging with the geopolitical cost of restricting AI infrastructure. Slowing U.S. data center buildout while China scales its compute capacity is a national security issue, not just an economic one.

The Stakes

The AI job apocalypse narrative is emotionally compelling and economically weak. The historical record, the demographic math, and 100 years of economic theory all push in the same direction: AI will change the shape of work, not end it.

That doesn't mean the transition will be painless. It won't be. Some roles will shrink. Real people will have to adapt. Policy should help them do that.

But the 'halt all AI development to save jobs' argument would leave American workers less productive, American companies less competitive, and Beijing laughing all the way to the compute cluster.

Fear is easy. Facts are harder. Right now, the facts favor the optimists.