Elon Musk Gave Money an Expiration Date. Does His 2036 Forecast Add Up?

Elon Musk has given money an expiration date.

During a July 2026 interview with The Economist, editor-in-chief Zanny Minton Beddoes asked how Musk's companies would make money in the future he was describing. His answer was unusually specific: "Money won't matter in 2036."

This was not a prediction that dollar bills will be replaced by an app. Musk was proposing something much stranger: an economy in which artificial intelligence supplies the brains, humanoid robots supply the bodies, and the two together produce more goods and services than humans can consume. Work becomes a hobby. Many prices fall. Governments issue everyone a "universal high income." Eventually, currency loses its job.

It is an astonishing idea. It is not complete nonsense. But it contains a hole large enough to drive several million robot trucks through.

Musk has described a possible destination. He has not explained how society gets there, or who owns the road.

The logic of a moneyless robot economy

Musk's argument begins with a real distinction. Today's most capable AI systems largely operate in the digital world. They can manipulate language, images and software, but they cannot replace a pipe, harvest a field or build a house by themselves.

For that, AI needs what Musk calls an "end effector": a physical machine capable of acting in the world. In his preferred version, that machine is a general-purpose humanoid robot.

Combine extremely capable AI with vast numbers of inexpensive robots and the result, Musk argues, is a nearly unlimited supply of labor. Machines could mine materials, operate factories, move freight, construct buildings and provide services around the clock. Human labor would no longer be the constraint it is today.

The next part of the argument is conventional economics taken to an unconventional extreme. If the supply of goods and services grows much faster than the supply of money, prices can fall. The familiar quantity equation is written MV = PQ: money multiplied by the rate at which it circulates equals prices multiplied by real output. As the Federal Reserve Bank of St. Louis explains, the relationship depends on assumptions about velocity and other conditions, but the basic direction is sound. More output relative to spending can mean lower prices.

If robots make food, transportation, manufactured goods and many services extraordinarily cheap, a modest income could buy a great deal. Musk says governments could issue checks to people who no longer receive wages. Eventually, he believes production would become so abundant that people would have little reason to use money at all.

There is evidence for the first step. Generative AI is already making some workers faster. A field study of 5,179 customer-support agents found that an AI assistant increased productivity by 14 percent on average. The gain reached 34 percent for novice and lower-skilled agents, while experienced workers benefited much less.

That is meaningful. It is also a very long way from an economy in which machines can do virtually everything.

Today's robots are not tomorrow's tireless workforce

The physical side of Musk's forecast is less developed than the digital side.

The world already contains millions of robots, but most are specialized industrial machines: arms that weld car bodies, move components or perform other carefully engineered tasks in controlled environments. The International Federation of Robotics reported that 4,663,773 industrial robots were operating worldwide in 2024. Factories installed 542,076 that year.

Those are impressive numbers. They are not evidence of millions of general-purpose humanoids preparing to leave the factory. A machine designed to repeat a precise motion behind a safety barrier is fundamentally different from an affordable robot that can enter an unfamiliar kitchen, notice a leaking pipe, find the shutoff valve and repair the damage without hurting anyone.

Tesla is moving from prototypes toward a production program, but its latest disclosure also shows how early that program remains. In its July 22, 2026 investor update, Tesla said it was installing first-generation Optimus production lines and expected production to start soon. The company said the initial builds would be used to collect training data and develop additional functions.

That is more concrete than a robot presentation. It is still a company plan for early machines, not independent proof that useful humanoids can be produced reliably by the millions, or billions.

The labor evidence is similarly more cautious than Musk. A 2025 International Labour Organization and NASK study estimated that one-quarter of global employment was in occupations with some generative-AI exposure. In high-income countries, the figure was 34 percent. But exposure does not mean that a job will disappear. Because most occupations still contain tasks requiring human input, the researchers identified transformation as the more likely outcome.

The economic forecasts are all over the map

Economists do disagree sharply about how much AI could increase overall production, but the disagreement is not a simple contest between optimism and pessimism. Different models begin with very different evidence and assumptions.

MIT economist Daron Acemoglu estimated in The Simple Macroeconomics of AI that current task-level evidence implied no more than a 0.66 percent cumulative increase in total factor productivity over ten years, and less than 0.53 percent after accounting for tasks that are harder for AI to learn.

At the other end of the range, Jessica and Jonathan Wachter used the extraordinary level of AI investment to calibrate rare productivity-boom scenarios. Their June 2026 NBER paper produced outcomes ranging from 5 to 58 additional percentage points of cumulative GDP growth by 2030.

That enormous range does not prove Musk right. It shows how violently the result changes when a model allows rare technological booms and assumes that investment signals future profits and productivity. Acemoglu's calculation is anchored more closely to measured gains in automatable tasks; the Wachters ask what growth could look like if current investment is rationally anticipating a major boom. Neither approach can observe breakthroughs that have not happened.

The 2036 prediction is therefore not a straightforward extrapolation from established data. It is a bet that multiple breakthroughs will arrive, scale and work together within a decade.

Money is not merely a paycheck delivery system

Suppose Musk wins that bet. AI becomes extraordinarily capable. Humanoid robots become cheap and reliable. Production soars.

Money could matter much less. That still would not necessarily make it obsolete.

The Bank of England describes money as a medium of exchange, a store of value and a common way of measuring value. We use it because people want different things at different times, and because many of those things are limited.

Automation can reduce scarcity. It cannot automatically eliminate it.

Robots might build a billion excellent chairs, making chairs almost free. They cannot manufacture another plot of land beside Central Park. They cannot give every person the same beachfront view, the same seat at a concert or the attention of the same surgeon at the same moment. Energy, computer chips, raw materials and factory capacity can expand, but not infinitely or instantly.

Musk has acknowledged this problem himself. At the November 19, 2025 U.S.-Saudi Investment Forum, captured in official event video, he said electricity, mass and fundamental physics would remain constraints even if currency eventually became irrelevant.

But a constraint is precisely where an allocation system becomes useful. If ten people want one scarce thing, society needs some way to decide who gets it. Money is one such mechanism. Remove it and the competition does not disappear; it may reappear as a queue, lottery, quota, favor, political decision or privilege attached to ownership.

A highly automated future could therefore contain large islands of abundance surrounded by stubborn pockets of scarcity. Many digital services, manufactured products and routine tasks might become nearly free. Housing in desirable places, limited natural resources and uniquely human services might remain expensive.

That would make money less important in daily life without making it irrelevant.

The missing question is ownership

There is an even larger problem. Producing an abundance of goods does not tell us who is allowed to use them.

Imagine one company owns the robots, the AI models, the factories and the energy supplying them. The machines can produce astonishing wealth, but their output still belongs to someone. Unless ownership is broad, prices fall far enough or governments redistribute access, people who have lost their wages could be surrounded by abundance they cannot claim.

The OECD has examined how AI automation could shift income from labor toward capital. Because ownership of capital is concentrated, the technology could increase income and wealth inequality without policies that spread its gains. The same machines that make society richer in total could leave many individuals with less bargaining power.

Musk's answer is universal high income. When Beddoes asked how people would live after displacement, he suggested that the Treasury issue checks. If real production were expanding faster than the money supply, he argued, additional money could be created without producing inflation.

That is possible in a narrow, conditional sense: a rapidly growing economy has room for more nominal spending. But it assumes the very outcome in dispute, an extraordinary increase in real output, and does not settle how much each person receives, which goods remain privately owned or how the system survives the transition.

Universal high income is not a feature that switches on when a robot learns to fold laundry. It is a political arrangement. Governments would have to create it, citizens would have to accept it, and institutions would have to maintain it while employment and tax systems changed underneath them.

The period when machines can replace large amounts of labor but have not yet created broadly shared abundance could be the most consequential part of the entire story.

So, is Musk right?

He could be right about the direction. AI and robotics may make intelligence, labor and many products dramatically cheaper. A future person may receive services that would cost a fortune today while spending very little. For basic necessities, sufficiently high productivity combined with public provision or income guarantees could make money feel almost optional.

There is no solid evidence that this will happen by 2036. Current AI has produced useful but bounded gains. General-purpose humanoid robotics remains in an early production and development phase. Economic forecasts vary from modest gains to historic booms because they rely on radically different assumptions. The physical inputs to production remain limited. And no technology, however capable, decides by itself who owns its output.

The strongest version of Musk's forecast is therefore conditional: if AI can perform almost all cognitive work, if robots can perform almost all physical work, if energy and material production scale with them, if society distributes the resulting output broadly, and if the things people most want are no longer meaningfully scarce, then money could lose much of its purpose.

That is a formidable collection of ifs.

Musk has offered a compelling picture of abundance. But abundance is not the same as access, and cheap labor is not the same as infinite resources. Technology may one day make money matter far less. It will not, by itself, decide that everyone has enough.