Anthropic (a non-listed company) has reached multiple agreements with technology giants such as Amazon and Microsoft. On one hand, these companies provide funding, and on the other hand, they supply computing power for AI models of Anthropic.
The aforementioned three companies have raised concerns in the market regarding circular financing: these companies provide funds to Anthropic, and after receiving the funds, Anthropic then goes back to purchase products and services from those same companies.
Today, Broadcom ( AVGO ) has also joined this ranks. It is reported that Broadcom will provide up to $42 billion in loans to Anthropic for the construction of its computing power infrastructure.
In return, it is expected that Anthropic will purchase a large amount of technical products from Broadcom; by 2027, this company, which develops the Claude large model, will become Broadcom's largest customer in its chip business.
In April this year, Anthropic officially announced a partnership with Broadcom and Google, with Google providing the computing power of tensor processing units (TPU) to Anthropic. Broadcom is also the designer of Google's TPU chips.
The report indicates that the cooperation between the two parties is not merely about purchasing chips; Anthropic will also rely on Broadcom for equipment leasing and accompanying financing services.
On September 5, 2025, in San Jose, California, USA, Broadcom's laboratories were opened to the public. The image shows a Broadcom circuit board used for chip testing. At that time, Broadcom was preparing to launch a new optical chip technology in order to compete with NVIDIA.
Amid the boom in computing power construction represented by AI, Wall Street analysts and investors have repeatedly expressed concerns about the circular investment model. Broadcom's competitor, NVIDIA, as well as AMD, also provide funding to their own customers, which includes OpenAI and Anthropic; after receiving this funding, these AI laboratories then pay to purchase high-performance chips from both companies.
The concerns of the market regarding such financial arrangements are that if there is a problem with any link in the chain, it could trigger a chain reaction that would severely impact the AI sector, and in turn drag down global stock markets that have benefited from the prosperity of AI and are highly dependent on the business conditions of AI companies.
