On October 6 local time, former US Treasury Secretary Robert Rubin, speaking in conversation with a moderator at the Greenwich Economic Forum, warned that there are "circular risks" behind large-scale AI investment: some large AI companies make massive commitments, and suppliers borrow based on these commitments; once the commitments cannot be fulfilled, the companies involved could be hit.

"Some very large AI companies have made massive commitments, and many suppliers have borrowed money based on these commitments," Rubin said.

He then asked: "What happens if they can't fulfill their commitments to the companies that borrowed based on those commitments?"

"This is the so-called 'circular risk,' and I think this circular risk really does exist right now," Rubin said.

He did not name specific companies or give a specific probability of the risk occurring, but said: "I don't know what the specific probability is... I think it's not even close to zero."

Regarding the massive scale of AI investment, Rubin is similarly cautious about the ultimate returns. "Will these investments pay off? I don't know," Rubin said.

Rubin believes the US economy still has growth momentum, part of which is driven by AI investment. AI is expected to significantly boost productivity, but it may also come with job losses and career transitions, so the final returns cannot be judged solely by investment scale and growth performance.

He also warned that the current strong performance of the economy and financial markets does not mean potential risks have disappeared. "Markets and economic conditions can remain disconnected from reality for a very long time, and then suddenly undergo very violent changes."

He cited the 1987 US stock market crash and the European sovereign debt crisis as examples of this, while stressing, "I'm not saying this will happen, and I'm not making a prediction, but it could happen."

On employment, Rubin believes AI will bring potential job losses, including knowledge workers such as lawyers, accountants and consultants, who will face career transitions and pressure from declining incomes. He acknowledged that AI is currently increasing employment at some companies, but even if overall employment does not fall sharply, the transition problems could still be very severe.

On US fiscal policy, Rubin believes that one cannot simply hope AI-driven growth will solve the debt problem; the longer fiscal adjustment is delayed, the harder it becomes.

Despite expressing many concerns, when asked whether he still believed the United States was the best place to invest, he replied: "I still think so."