In a move that underscores the sheer capital intensity of the artificial intelligence arms race, semiconductor giant Broadcom is reportedly in advanced negotiations with lenders to secure a debt package exceeding $60 billion. According to reports from Bloomberg, this massive capital infusion is designed to finance the acquisition of custom AI chips and networking equipment primarily for Anthropic, one of the leading developers of frontier AI models.
The proposed deal, which could potentially scale as high as $100 billion depending on market appetite and final negotiations, represents a significant escalation in the "compute-as-a-service" model. By leveraging a complex structure involving special-purpose vehicles (SPVs) and institutional giants like Blackstone and Apollo Global Management, Broadcom is effectively outsourcing the hardware financing burden, allowing AI labs to deploy state-of-the-art infrastructure without immediate, crushing capital expenditures.
The Structural Mechanics: How the Deal Works
At the heart of this multi-billion dollar arrangement is a sophisticated, bifurcated debt structure. The package is expected to be divided into two distinct tiers: a junior tranche, estimated at approximately $30 billion, and a senior-secured tranche, ranging between $60 billion and $70 billion.
The financial architecture relies heavily on the use of a Special Purpose Vehicle (SPV)—a legal entity created specifically to hold the assets (in this case, the silicon and networking gear) and manage the corresponding debt. By housing the debt within the SPV, the financial liability is kept off the balance sheet of the primary partners, while providing investors with a direct claim on the underlying hardware assets.
Crucially, this arrangement creates a "leasing" model rather than a direct sale. Anthropic does not technically purchase the chips; instead, investors finance the hardware acquisition, and Anthropic leases the computing power. To ensure the viability of these tranches, Broadcom has reportedly agreed to guarantee a portion of the senior debt. This "backstop" is essential, as it elevates the credit quality of the senior notes to investment-grade status, significantly reducing borrowing costs and making the deal attractive to risk-averse institutional lenders.
Chronology: From AI XPV Partnership to Multi-Billion Dollar Scaling
The trajectory of this partnership has been rapid, reflecting the hyper-growth phase of the AI industry.
- May 2026: Initial reports surfaced indicating that Apollo and Blackstone were shopping a massive financing deal to investors, signaling a shift toward asset-backed financing for AI infrastructure.
- June 2026: Broadcom, Apollo, and Blackstone officially announced the formation of the "AI XPV" partnership. The debut transaction raised $35 billion, aimed at significantly expanding Anthropic’s compute capacity.
- August 2026: Building on the success of the initial June raise, reports emerged that Broadcom is seeking an additional $60 billion to $100 billion in debt to continue the expansion.
- Ongoing: The partnership has set a long-term goal of financing 20 gigawatts of computing power for leading AI labs by 2028. With the first $35 billion tranche accounting for roughly one gigawatt of capacity, the partners are barely five percent of the way to their stated goal—a testament to the astronomical costs of scaling modern AI.
Supporting Data: The Financial Scale of the AI Arms Race
The sheer scale of the financing required for AI is unprecedented in the history of the technology sector. To put the 20-gigawatt goal in perspective, it is equivalent to the power output of roughly 20 nuclear power plants.
Anthropic’s Financial Profile
The need for such massive capital is reflected in Anthropic’s own financial trajectory. By the end of July 2026, the company’s annualized revenue run rate had reached an eye-watering $65 billion, a stark contrast to its $11.5 billion revenue for the second quarter of the year. However, this growth comes at a steep price: the company recorded a net loss of nearly $42 billion in 2025, a five-fold increase from the $8.3 billion loss in 2024.
Broadcom’s Market Position
Broadcom is positioning itself as the primary alternative to Nvidia for firms seeking to reduce their dependency on a single supplier. The company is currently designing custom silicon for industry titans like Alphabet and Meta. With supply agreements already locked in with Anthropic and OpenAI, Broadcom’s CEO recently projected that AI chip sales alone could exceed $100 billion in the coming year. This, combined with massive legacy agreements—such as a $30 billion deal with Apple and a $200 billion deal with Samsung—cements Broadcom as a linchpin of global digital infrastructure.
Official Responses and Stakeholder Silence
As is typical with highly sensitive, multi-billion dollar financial negotiations, official confirmation has been sparse. When approached for comment regarding the $60 billion to $100 billion debt package, spokespeople for Broadcom, Anthropic, Apollo, and Blackstone all declined to issue statements.
The sources providing the details—individuals with direct knowledge of the private discussions—have requested anonymity. These stakeholders acknowledge that while the current outline is clear, the terms remain fluid. The specific split between the junior and senior tranches, the exact nature of the Broadcom guarantee, and the full roster of participating lenders are subject to change as the deal moves toward finalization.
Market Implications: The Era of "Shadow" Credit
The implications of this deal extend far beyond the immediate participants. Broadcom’s strategy is emblematic of a broader shift in how the tech industry is funding the "AI build-out."
A New Class of Asset-Backed Finance
We are witnessing the emergence of a massive "shadow credit" market. Just this month, Nvidia announced a coalition involving BlackRock and Goldman Sachs to line up over $500 billion for AI infrastructure. The SEC has even weighed in, signaling that data center securitization—the process of turning these infrastructure projects into tradable debt—falls outside of traditional Dodd-Frank risk retention rules. This regulatory green light is expected to accelerate the flow of capital from institutional investors into the AI sector.
Broadcom’s Competitive Edge
For Broadcom, the ability to facilitate this financing is a competitive masterstroke. By essentially becoming a "one-stop shop" that provides both the hardware and the financing mechanism to pay for it, Broadcom is making it frictionless for companies like Anthropic to choose their silicon over competitors. This is particularly relevant in a market where Alphabet, Amazon, and Microsoft are all scrambling for compute, and where every major player is looking to manage the immense capital expenditures associated with training frontier models.
The IPO Horizon
For Anthropic, this financing is a bridge to the public markets. The company is currently finalizing a $10 billion revolving credit facility and is preparing for an Initial Public Offering (IPO) that is expected to rival or exceed the $86.2 billion record set by SpaceX. With the backing of Morgan Stanley, Goldman Sachs, and JPMorgan, Anthropic is clearly aiming for a valuation that reflects its central role in the AI economy.
Conclusion: A High-Stakes Bet on the Future
The $60 billion to $100 billion debt package currently under negotiation is more than just a corporate loan; it is a declaration of confidence in the long-term viability of generative AI. By structuring these deals through SPVs and institutional partnerships, Broadcom is effectively socializing the risk of the AI build-out, inviting the world’s largest asset managers to bet alongside them.
Whether this massive investment in silicon and power capacity will yield the returns necessary to justify such debt levels remains the central question for the tech industry. For now, however, the momentum is undeniable. With the backing of the world’s largest lenders and the participation of the most significant AI labs, the physical infrastructure of the AI revolution is being built at a speed and scale that few could have imagined just a few years ago. As the race toward the 20-gigawatt target continues, the financial machinery behind it will remain the most critical, yet often overlooked, component of the AI era.
