Abstract:
People have long been accustomed to Elon Musk making judgments, and this time, he intended to show that he was right all along. The CEO of SpaceX and Tesla agreed with the long article published by Anthropic CEO Dario Amodei.
Amodei's core point is: Artificial intelligence with the ability to recursively self-iterate will reach a speed beyond the scope of human understanding and control; even if such technology is to be developed, extreme caution must be exercised.
To prevent future risks, he proposed that cutting-edge AI companies should deploy third-party evaluators internally to test AI models and supervise laboratories to implement safety guidelines and commitments; at the same time, report safety incidents, not only to evaluate the risks of finished models, but also to review the alignment security of the entire training process and technical pipelines.
January 22, 2026 Elon Musk attended the World Economic Forum Annual Meeting in Davos, Switzerland
Musk responded by posting on the He wrote at the time: "In my opinion, the risk of general artificial intelligence (AGI) is much higher than that of nuclear weapons. No matter how smart a person is, it is difficult to imagine an existence that far exceeds his or her own intelligence."
What is intriguing is that while Musk is issuing risk warnings, he is betting heavily on the AI track: SpaceX's AI department is aggressively building data centers and launching the Grok chat robot; Tesla is also advancing "physical artificial intelligence" projects, including self-driving taxis and the humanoid robot Optimus, which is scheduled to be put into production at the end of this year.
At present, the largest investment is concentrated in the data center business. SpaceX Chief Financial Officer Brett Johansen recently revealed that as of December last year, the company’s AI computing power business’s annual recurring revenue (ARR) had reached $45.7 billion; he is more confident in reaching the $100 billion ARR target by the end of this year.
Musk isn’t the only one to suffer revenue losses from slowing AI investment. Amodei, as well as his rival OpenAI CEO Sam Altman, are likely to shrink their business and postpone their plans to go public.
On February 19, 2026, Anthropic CEO Dario Amodei speaks at the AI Impact Summit in New Delhi, India
Anthropic's IPO prospectus S-1 may be submitted at any time, and the document may indicate a slowdown in product iteration and infrastructure expansion, which will drag down profitability indicators. However, the Financial Times reported that after excluding some costs, the company can still maintain a profit margin of 80%.
Ultraman himself has announced that OpenAI will be postponed to 2027. The external rhetoric is to solve AI security risks.
Are the giants collectively putting the brakes on AI research and development out of consideration for the long-term safety of mankind, or is it just a utilitarian means to raise the valuation of their own AI assets, as Michael Burley questioned?
One thing is for sure: Musk’s bets tend to pay off, which not only benefits his listed companies, but also makes him steadily move towards becoming the world’s first trillionaire.
Comments