Foreign media reports reveal that Anthropic's IPO prospectus suggests a potential valuation exceeding $2 trillion. However, analysts warn that if Anthropic does indeed IPO at a $2 trillion valuation, it could expose the biggest problem in the entire

2026-09-29

Foreign media reports reveal that Anthropic's IPO prospectus suggests a potential valuation exceeding $2 trillion. However, analysts warn that if Anthropic does indeed IPO at a $2 trillion valuation, it could expose the biggest problem in the entire AI boom: capital constantly circulates among the same group of companies. The logic is as follows: Amazon/Google invests in Anthropic, which then raises massive amounts of funding and uses these funds to purchase AWS/Google Cloud computing power. Amazon/Google earns cloud revenue, Anthropic raises further funding at an even higher valuation, Amazon/Google's stake in Anthropic appreciates, and more capital can flow into the AI ecosystem. In this process, investors and suppliers highly overlap. Amazon has invested $33 billion in Anthropic, while Anthropic has committed to spending over $100 billion on AWS over the next ten years; Google has committed approximately $43 billion, while Anthropic plans to spend $200 billion on Google Cloud over the next five years. Therefore, large tech companies are both Anthropic's shareholders and its largest infrastructure providers. The computing power provided by AWS and Google Cloud is real, and Anthropic is indeed using these resources; therefore, it cannot be simply stated that cloud revenue is fictitious. The real question is the source of demand: how much of the computing power demand ultimately comes from paying end-user enterprises and consumers, and how much is simply due to large tech companies initially providing capital to AI companies, which then reinvest that capital back into the tech companies? Ultimately, the viability of this system depends on "external cash flow." If a large number of enterprises, consumers, and governments are ultimately willing to continuously pay for Claude and Claude-based applications, then the entire cycle will be supported by genuine end-user demand, and today's massive infrastructure investment may prove justified, potentially becoming one of the largest infrastructure investment cycles in history. However, if the profits generated by end-user demand are far lower than infrastructure capital expenditure, the net cash flow generated by the entire ecosystem to the external economy will be insufficient. The massive growth will merely be capital circulating rapidly among the same group of companies, without corresponding profit generation.