Elon Musk laid out a sweeping forecast for artificial intelligence in a lengthy interview with The Economist’s editor-in-chief, Zanny Minton Beddoes, recorded on July 20, 2026. He predicted that AI will surpass the combined intelligence of humanity within roughly a decade, that humanoid robots will carry that intelligence into the physical economy, and that money itself could lose much of its relevance by 2036. The claims are his personal predictions, not established outcomes, but they carry weight because of the companies he leads in AI models, robotics, chips and large-scale computing.
Intelligence beyond humans
Musk said AI could exceed the intelligence of any individual human within about five years and, within ten, surpass the collective mental capacity of all humans combined. At that point, he suggested, there may be no significant cognitive task that AI cannot do better than people. Analysis, planning, coding, research, writing and decision-making would all become routine for machines.
He drew a distinction between today’s AI, which he described as largely confined to the digital world, and what comes next. Current systems can understand information, write code and operate software tools, but they do not move through the physical world. The real shift, in his view, arrives when that digital intelligence is paired with humanoid robots and other machines.
Robots as the hands of AI
In Musk’s framing, robots become the physical limbs of AI, with the core intelligence living in large models and data centers and the robots carrying out its instructions. AI would then not only produce reports but move goods, run factories, do household work, take part in construction and deliver services. He argued that production would no longer be limited by human labor, fundamentally changing the nature of the economy.
From there he reached his most striking claim: that if large numbers of robots continuously produced goods and services, the world could move toward extraordinary abundance, and the current importance of money might not survive to 2036. His logic is that money exists to acquire goods and services; if AI and robots can produce those in near-unlimited quantities, scarcity shrinks and prices lose their meaning. The interview offered no practical account of how ownership, access or distribution would work, or how such abundance would reach everyone equally. Economists and commentators have since questioned the claim on exactly those grounds.
Control and values
On whether humans could remain in control, Musk was blunt. If AI became as far ahead of humans as humans are of chimpanzees, he said, it would be hard to imagine people staying fully in charge, and believing otherwise might be self-deception.
Even so, he argued the goal should not be to stop AI but to instill values that benefit humanity, naming truth-seeking, curiosity and an interest in human flourishing as the most important traits. He acknowledged that the risk is not zero and said he sometimes feels both excitement and fear about AI in the same day. But he described any attempt to halt AI and robotics as unrealistic, making risk reduction and preparation the better strategy. That marks a softer tone than his earlier warnings about runaway self-improving systems.
A proposal for safety
Musk offered one concrete suggestion: leading AI companies should stay in regular contact and give rivals a window of one to two weeks to test new frontier models for safety problems, cyber risks and potential misuse before release. Government, in his view, should act as a backstop, intervening only if several major labs judged a model dangerous and the developer refused to address the concerns. Competing labs, he argued, are better placed than a general regulator to spot risks quickly because they build similar systems themselves.
He said such a system would need to include Chinese AI companies to be effective globally, and suggested the U.S. and Chinese governments are the only powers that could realistically restrict the release of a truly dangerous model. Musk described Chinese labs as performing well despite limited computing resources, singling out the Kimi model, and said they could lead the field if given access to more advanced chips. He identified electricity and chips as the main physical constraints on AI progress, noting that in some regions power and cooling capacity have become a bigger bottleneck than silicon.
Jobs
For most people, Musk’s employment forecast is the most consequential part. He said nearly all work done on a computer or phone could soon fall within AI’s reach, claimed AI already writes better code than many professional programmers, and said the same would apply to writing, analysis, design and office tasks. Screen-based jobs would be affected first; physical jobs would follow as humanoid robots mature.
He used gardening as an illustration of work in that future: people can buy vegetables but still grow them for pleasure, and work could similarly shift from economic necessity to personal choice. He conceded the transition would not be smooth. Job disruption from AI, he said, could move far faster than past industrial or computing revolutions, leaving less time for new jobs to replace old ones and raising the prospect of serious social and economic strain in the interim.
Weighing the claims
The throughline of the conversation is that Musk now treats AI as a force that cannot be stopped, making the central question how human values, safety systems, international cooperation and economic distribution are arranged around it. His timelines for superhuman intelligence and a post-money economy are unproven and, in the case of 2036, widely disputed. But the pace of AI development is already fast enough that decisions about employment, education, law and safety carry real cost if deferred, whether or not his dates prove correct.
