Broadcom, Apollo, and Blackstone Expand AI XPV Platform to a Landmark $100 Billion Custom Chip Debt Package
The alternative financing of artificial intelligence infrastructure has reached a staggering new scale. On August 20, 2026, reports surfaced that Broadcom (AVGO) is negotiating a massive debt financing package of up to $100 billion to fund custom AI chips and infrastructure for Anthropic, OpenAI, and other premier clients. This represents a massive expansion of the initial $35 billion Broadcom, Apollo, and Blackstone Expand AI XPV Platform to a Landmark $100 Billion Custom Chip Debt Package launched in June 2026 with Apollo Global Management and Blackstone.
The Debt-Funded AI Arms Race
The financing model demonstrates that the buildout of global AI infrastructure is increasingly moving off corporate balance sheets and into the private credit markets.1 Goldman Sachs Research estimates that total AI-related debt issuance could approach $500 billion by the end of 2026, as the massive capital requirements of gigawatt-scale data center builds outstrip traditional cash reserves.
The proposed $100 billion Broadcom package is structured to bypass balance-sheet constraints:
- Tranches: It pairs $60 billion to $70 billion of senior secured debt with approximately $30 billion of junior debt.
- Off-Balance-Sheet SPV: The debt is expected to be issued through an independent Special Purpose Vehicle (SPV), keeping it off Broadcom's own balance sheet, though Broadcom may guarantee a portion of the senior notes.
- Strategic Targets: The proceeds will secure custom AI accelerators (such as Google/Broadcom's custom TPUs and OpenAI/Broadcom's custom "Jalapeño" inference chip) and associated infrastructure.2
This financialization enables frontier AI labs to challenge Nvidia's GPU dominance by securing massive custom silicon capacity through alternative, highly structured credit facilities.
Verbatim Quotes
From SiliconANGLE's coverage:
"Broadcom Inc. is reportedly seeking to borrow up to $100 billion as part of a new artificial intelligence chip financing deal... The proposed deal could include between $60 billion and $70 billion worth of senior notes... Additionally, the company could reportedly add about $30 billion in junior notes..." — Maria Deutscher, SiliconANGLE
From TipRanks' report:
"AI chipmaker Broadcom is discussing a financing package that could reach $100 billion to help customers such as Anthropic secure more AI computing capacity, according to Bloomberg... If Broadcom guarantees part of the debt, it also takes on financial risk while trying to expand hardware sales.3 At the same time, this raises questions about how much risk chipmakers should take on in order to help customers fund large AI investments." — TipRanks
From Cryptopolitan's analysis:
"This transaction highlights the reality of how the development of AI across the world is being financed by loans rather than company funds... Providing funding for such custom chips at this magnitude makes that option more achievable... Competition is increasingly about who can secure the cheapest and deepest pools of capital, not simply who can build the fastest processor." — Ashish Kumar, Cryptopolitan
Interpretation
The shift to debt-funded custom silicon represents a structural challenge to Nvidia's accelerated-computing moat. By establishing a $100 billion financing conduit, Broadcom and its Wall Street partners (Apollo and Blackstone) are offering hyperscalers and frontier labs a viable second-source alternative that does not require massive up-front capital expenditures. However, this model introduces significant systemic credit risk, as chipmakers begin guaranteeing senior debt tranches to fund their own hardware pipelines, tying their financial health directly to the long-term cash-generation capability of AI workloads.
-
An instance of Multi-billion-dollar AI hardware builds have outgrown dilutive equity. — This illustrates how the extreme capital required to fund custom silicon and hardware builds is driving companies to use massive structured private credit packages. ↩︎
-
An instance of Custom, model-tailored silicon is breaking the general-purpose GPU monopoly. — The debt-funded custom silicon projects allow tech giants to bypass generic GPU pools in favor of specialized, model-tailored custom accelerators. ↩︎
-
An instance of Hardware vendors must assume client credit risk to sustain historic infrastructure expansion. — It shows a leading chipmaker taking on direct credit and financial underwriting risks to help its customers secure off-balance-sheet debt for hardware purchases. ↩︎