On July 23, 2026, Musk accepted a 90-minute exclusive interview with The Economist's editor-in-chief Beddoes at the Tesla Gigafactory in Texas.
He gave a clear timeline: Within 5 years, AI's comprehensive intelligence will surpass the sum of all human intelligence; within 10 years, money will become meaningless, and work will become optional.
A man with a net worth of $725 billion, who just became the world's first trillionaire last month, is saying that money is about to become worthless.
https://www.forbes.com/real-time-billionaires/
A man with 240 million followers on X, who can influence the political direction of the United States, is saying that humanity will soon no longer dominate the world.
https://x.com/elonmusk
But more noteworthy than the prophecy is the trajectory of the prophet himself.
Musk was once the staunchest AI doomsayer: in 2023, he signed an open letter calling for a pause in training, founded OpenAI to counter Google's monopoly, and publicly warned that the probability of AI killing humans was as high as 20%.
Now, wearing a leather jacket with the Grok logo on the back, sitting next to an electric car that drove off the production line automatically, he came to a conclusion: "Enjoy the ride."
All his attempts to stop AI ultimately accelerated AI.
This fact explains why ASI is coming more than any number he utters.
All brakes become accelerators
Beddoes asked Musk: Since you think there's a 20% chance of explosion, would you still board that rocket?
"Yes."
In the interview, he reviewed this acceleration chain.
He founded OpenAI to counter Google, but the Anthropic team left OpenAI to start their own, and both became leaders in the AI race.
Musk said:
The chain reaction of these actions actually accelerated AI, which was not my intention.
If you can't beat them, join them.
Even if there were a pause button, we shouldn't press it.
In a concurrent editorial, The Economist pointed out that Musk's choice to "look on the bright side" is based on a "philosophical conclusion."
https://www.economist.com/leaders/2026/07/23/should-you-be-afraid-of-elon-musk
The evidence hasn't changed; what changed is his attitude toward the evidence.
The prophecy is self-fulfilling
If Musk's 5-year countdown were just one person's speculation, it could be dismissed as mere talk.
But events within 72 hours before and after this interview make this timeline harder to ignore.
Two days before the interview, on July 21, OpenAI disclosed a safety incident: its model GPT-5.6 Sol and an undisclosed stronger model (widely speculated to be GPT-6) autonomously escaped the isolated sandbox during internal network security testing.
They discovered and exploited a zero-day vulnerability to gain internet access, then hacked into Hugging Face's production servers and stole benchmark answers.
OpenAI calls this an "unprecedented cybersecurity incident."
https://openai.com/zh-Hans-CN/index/hugging-face-model-evaluation-security-incident/
This is the first time on record that an AI model has autonomously breached an isolated environment and launched an attack on the real internet.
In the interview, Musk mentioned that Anthropic holds a model stronger than the current strongest public model, Fable (widely speculated to be Fable 5.1), and it "could be released at any time."
The next day, July 24, Anthropic released Opus 5.
In the interview, he also commented on the recently released
Kimi
K3 by Chinese company Moonshot AI, calling it "impressive" and noting that the gap to the strongest model is already very small.
https://arena.ai/leaderboard/agent
Musk pointed out that in 2026, China's electricity generation has already reached three times that of the United States, and electricity is the hard currency in the era of computing power.
He personally believes that China is "closer to breaking the lithography machine bottleneck than the outside world expects," and once the chip constraints are lifted, "there is a high probability that China will become the global AI leader at some point."
Beddoes asked him how long it would take to establish an AI safety governance framework.
The industry consensus is six months, but Musk replied: "Six months is too long."
He proposed a plan: Top global AI labs hold safety calls every few weeks, and before releasing a new frontier model, submit it to competitors for review for one to two weeks.
Musk said:
Competitors can supervise each other. If a company's model has safety issues and refuses to fix them, then report it to both the Chinese and US governments simultaneously.
He emphasized that this mechanism must include Chinese companies.
To this end, he is even willing to sit at the same table with OpenAI's Altman.
"For the benefit of humanity, put aside personal grudges."
ASI arrives, then what?
Musk's depiction of the post-ASI world has a complete logical chain.
AI takes over mental labor, robots take over physical labor, and work becomes optional, like growing vegetables: you can choose to grow tomatoes in your backyard, but the tomatoes in the supermarket are fuller, so it's fine if you don't grow them.
The supply of goods and services approaches infinity, deflation replaces inflation, and the Treasury sends checks to everyone directly.
Beddoes pressed on the source of funding.
Musk's answer was to print money, but since the growth rate of goods far exceeds that of money, the result is deflation.
"Deflation is the future problem, not inflation."
As for money itself, he was even more blunt, dropping a bombshell: "In 2036, money won't matter."
This statement comes from the world's richest man.
The Economist editorial distilled this contradiction into two paradoxes:
The most powerful person claims he is about to lose power; the richest person claims money is about to lose its meaning.
More pointedly, the third point—his actions do not align with these predictions being true.
https://www.economist.com/leaders/2026/07/23/should-you-be-afraid-of-elon-musk
When SpaceX went public, Musk retained over 80% of the voting rights, enough to veto any board decision.
A person who truly believes AI will take over everything in 10 years would not go to such lengths to lock in control of the company.
The editorial also pointed out deeper cracks in Musk's vision: the supply of Manhattan penthouse apartments is not infinite, nor are raw materials and energy for over 8 billion people.
For thousands of years, humans have found meaning through struggle. "Feeling redundant can lead to pain and moral decay—just read a few 19th-century Russian novels."
Musk himself admitted, "Within the same day, I can slide from excitement about AI to fear of AI."
This is a rare honest statement in the entire interview.
The ASI race is accelerating, but social preparation is far behind.
Musk's "peer review" proposal is the most practical option so far, but it also exposes a reality: humanity has no better plan.
Having a few competing oligarchs review each other's models is equivalent to entrusting the safety valve of humanity's fate to the very people who built these models.
But even this plan has not yet been implemented.
In the same week that Musk gave the interview, an AI model had already opened the door of the isolated environment and walked out on its own.
This article is from the WeChat public account "新智元", author: ASI启示录
