According to foreign media reports, a number of Wall Street financial institutions and asset management giants, including Bank of America, Blackstone, Citigroup and Morgan Stanley, will jointly provide Anthropic with large-scale debt financing of up to $63 billion (approximately RMB 420.679 billion), earmarked for the purchase of Broadcom's artificial intelligence chips and networking equipment. This is also one of the largest debt transactions in the field of AI computing power construction.
According to the disclosed financing structure, the transaction will be implemented through the establishment of a brand-new special-purpose entity, which will be responsible for purchasing Broadcom's AI accelerator chips and networking equipment, and then leasing them to Anthropic, the developer of Claude. In terms of the specific debt arrangement, Bank of America, Citigroup and Morgan Stanley will arrange the first-lien financing, while Blackstone will participate as the committed provider of second-lien financing. In addition, Broadcom will provide "residual value support" to the senior debt lenders, meaning that in the financing structure it offers lenders a guarantee: if future declines in asset values cause losses, Broadcom is willing to bear the corresponding losses, with a cap consistent with the amount of the first-lien debt; however, the announcement did not disclose the specific size of that debt tranche.
Under the current plan, this batch of hardware equipment is expected to begin formal delivery in 2027, which also forms an important part of the previously announced strategic computing power supply agreement involving Broadcom. Notably, before this, Apollo and Blackstone had already reached a related agreement with Broadcom, contributing $35 billion (approximately RMB 233.711 billion) to expand Anthropic's AI computing power foundation.
The completion of this latest massive financing fully highlights that traditional banks and alternative asset managers are playing an increasingly critical role in providing infrastructure financing for the explosively growing AI computing power demand. At the same time, Anthropic officially submitted its IPO application on Monday local time, and this listing is expected to set a historical record; however, its heavy R&D investment and continuously disclosed security risks will likewise test the capital market's long-term confidence in this top AI giant.