Broadcom is exploring an $80 billion debt facility to finance expanded custom silicon production and networking solutions for Anthropic’s AI chip roadmap.
Broadcom is in discussions with lenders to raise between $70 billion and $80 billion in debt to finance a chip and computing infrastructure arrangement designed to benefit artificial intelligence companies, including Anthropic. The financing comes as the AI sector races to secure the hardware necessary to support rapid expansion. According to Bloomberg reporting cited by Reuters, the proposed structure could ultimately push the total raise to as much as $100 billion.
The transaction would further cement Broadcom’s expanding role in AI infrastructure and deepen its ties with Anthropic, which is simultaneously laying the groundwork for its own custom semiconductor operations. Lenders are currently targeting a senior tranche of approximately $45 billion and a junior tranche of roughly $35 billion, though CNBC reported that these figures remain subject to change as negotiations progress. Blackstone and Apollo Global Management are among the investors evaluating participation. Blackstone declined to comment, while Broadcom and Apollo did not respond to requests for comment.
This proposed financing builds upon a partnership announced in June between Broadcom, Apollo, and Blackstone. Under that agreement, the companies committed $35 billion to expand Anthropic’s computing infrastructure utilizing Broadcom’s custom chips and networking technology. The initial investment was structured to add approximately one gigawatt of computing capacity, with the broader partnership aiming to deliver more than 20 gigawatts of compute power to leading AI laboratories by 2028.
The arrangement underscores the massive infrastructure demands of the AI industry, where firms are fiercely competing for processors, data centers, and networking capacity. As Broadcom scales its AI chip division, Anthropic is concurrently bolstering its semiconductor capabilities. The developer of Claude has hired Amir Salek, a founder of Google’s custom chip program, to join its compute team. Salek previously led Google’s Tensor Processing Unit business until 2022 and contributed to the development of the company’s first seven generations of chips. At Anthropic, he will report directly to James Bradbury.
Currently, Anthropic sources processors from multiple vendors, including Nvidia, Google, and Amazon. Developing proprietary silicon would grant the company greater oversight over the hardware required to train and deploy its models. Custom chips could also help mitigate supply chain bottlenecks and allow Anthropic to tailor architectures to its specific computational needs. OpenAI is pursuing a comparable strategy; the ChatGPT creator recently unveiled Jalapeno, a custom chip developed alongside Broadcom, with plans to begin deployment later this year.
Broadcom itself designs custom processors for major technology firms, including Alphabet, Meta, Anthropic, and OpenAI, helping them reduce their dependence on Nvidia, the market’s dominant AI processor supplier. This semiconductor push coincides with Anthropic’s rapid expansion of its chip and data center partner network. The company has secured an initial order valued at approximately $250 million with UK-based chip startup Fractile, with provisions to scale the agreement. Additionally, Anthropic has signed computing capacity deals with Riot Platforms and Volta Infra Holdings.
Anthropic’s strategy increasingly emphasizes securing compute capacity across multiple providers while cultivating internal expertise to manufacture its own silicon. Salek brings extensive experience from both Google and Nvidia, along with his most recent position as senior managing director at Cerberus Capital Management.