Broadcom has begun raising $60 billion in debt to finance the production of AI chips for Anthropic, according to a report by Quartz. The figure ranks among the largest debt financings ever undertaken by a semiconductor company and signals how capital-intensive custom AI silicon has become.

The debt program covers chip manufacturing commitments tied to Anthropic's compute buildout. Broadcom has built its AI business on designing and supplying application-specific integrated circuits (ASICs) for large cloud and AI customers, a model that requires the company to underwrite substantial capacity ahead of revenue recognition.

Why does Broadcom need $60 billion?

Custom AI chip programs do not generate income until silicon ships in volume. Before that point, the supplier must pay for:

• Design and engineering work on custom processors
• Wafer capacity secured at foundries such as TSMC
• Packaging and advanced memory supply chains, including high-bandwidth memory
• Multi-year capacity reservations that carry fixed payment obligations

A $60 billion raise indicates the scale of the commitments Broadcom has taken on for the Anthropic program. Debt, rather than cash on hand, allows the company to spread these outlays across the program's life while preserving balance-sheet flexibility for other customer programs.

What does the deal say about the AI chip market?

The Anthropic engagement places Broadcom at the center of a shift among frontier AI labs away from buying general-purpose GPUs and toward custom accelerators. Labs that can commit to sufficient volume can contract directly with a designer like Broadcom for silicon tailored to their workloads, then pay foundries to fabricate it.

For that model to work, someone must finance the enormous upfront cost. The $60 billion figure quantifies it. It also demonstrates that AI infrastructure spending has reached a scale where individual customer programs justify corporate debt issuances comparable to large sovereign or industrial financings.

The raise follows a broader pattern in the sector. Hyperscalers including Microsoft, Meta, Alphabet and Amazon have committed hundreds of billions of dollars to AI data center capacity, and OpenAI has announced multi-trillion-dollar infrastructure initiatives with manufacturing partners. Anthropic, as one of the leading AI labs, now requires compute capacity at a scale that its own capital cannot cover — pushing the financing burden onto its chip supplier.

Who carries the risk?

Debt financing shifts execution risk onto Broadcom's balance sheet. If demand for Anthropic's products grows as projected, chip revenue services the debt. If it falls short, Broadcom still owes the creditors.

That asymmetry makes the credit terms of the raise a metric to watch. Analysts will scrutinize the pricing, maturity structure and covenants of the debt as indicators of how lenders assess long-term AI compute demand.

The transaction also consolidates Broadcom's position as the dominant independent supplier of custom AI silicon, deepening its dependence on a small number of very large customers whose spending plans are themselves unproven over the full life of the debt.

What happens next?

The raise is now underway, per the Quartz report. Market participants will track the final size, pricing and investor demand as the bonds or loans reach the market — each data point will serve as a real-time read on lender confidence in AI infrastructure economics.