NVIDIA and Broadcom are offering billions of dollars in loans to accelerate sales of their AI chips, South Korea's Chosun Ilbo reports.

The lending programs, as described in the report, put the two chip designers in the unusual position of financing their own customers' purchases. For AI infrastructure buyers, the arrangement matters because capital constraints — not demand — have become a gating factor on how many GPUs and custom accelerators a cloud provider or enterprise can deploy in a given quarter.

Details on loan sizes, interest terms, recipient names, and deal counts were not included in the report, so any estimate of how much incremental chip revenue the financing unlocks remains an assertion rather than a measured outcome. What the report does establish is the direction of travel: the two largest AI chip suppliers are now willing to carry balance-sheet risk to keep order books moving.

For vendors selling alongside these chips — server makers, integrators, and GPU-cloud resellers — supplier-backed lending effectively lengthens the payment chain. A chip loan that defers or eases a customer's upfront cost does not disappear from the deal; it shifts when and where cash constraints bind, and sellers further down the stack should expect procurement teams to factor that timing into their own payment negotiations.

The move also signals how NVIDIA and Broadcom read current AI hardware demand: strong enough to justify billions in credit exposure, but contingent enough on customer balance sheets that financing is now part of the sales motion rather than an afterthought. That is a workflow change for their own account teams, and a competitive signal for rivals such as AMD and custom-silicon brokers deciding whether to match vendor credit terms.

Chosun's report did not specify a start date for the programs or name the banks involved, if any. Watch for confirmation in the companies' upcoming earnings disclosures, where any material financing commitments would need to appear on the balance sheet.