When a chipmaker starts raising what could be the largest corporate debt deal in history just to keep AI companies supplied with compute, something has fundamentally changed about how capital is flowing through the AI industry.

On August 20, Bloomberg reported that Broadcom is in talks with a group of lenders including Apollo Global Management and Blackstone to raise more than $60 billion in debt for an AI chip financing deal. The total package could reach as much as $100 billion, structured with roughly $60-70 billion in senior-secured debt and a junior tranche of around $30 billion.

The primary beneficiary is Anthropic, the AI lab behind Claude, though the platform is designed to serve multiple AI companies needing access to custom silicon at scale.

Why This Is Different

The scale here is worth pausing on. For context, the largest corporate bond deals in history have typically topped out in the $40-50 billion range for major acquisitions. A $60-100 billion AI chip financing package would rank among the biggest ever seen.

But what’s also interesting is the structure. This isn’t Anthropic raising money directly. It’s Broadcom — the chip and infrastructure company — raising debt that would then be used to fund AI customers’ access to custom compute.

This builds on a partnership that Broadcom, Apollo, and Blackstone established in June, where an initial AI XPU financing platform was set up at roughly $35 billion, primarily to support Anthropic’s AI data center build. The August talks represent a significant scale-up of that arrangement.

A special-purpose vehicle would issue the debt, which helps insulate Broadcom’s core balance sheet while creating a dedicated pool of capital for AI chip supply.

The AI Compute Arms Race Has Its Own Financial Layer

This deal reflects something important that business leaders are starting to grapple with: AI compute is becoming a financial product as much as a technology one.

Traditional enterprise software had relatively modest capital requirements. You built the software, you sold it, the margins were high. AI at scale is different. The compute costs are so substantial that a new financial architecture is emerging around them.

Broadcom’s approach essentially lets AI companies like Anthropic lock in chip supply without carrying the full infrastructure cost directly on their balance sheets. For Anthropic, this means more predictable access to the custom silicon needed to train and run frontier models. For Apollo and Blackstone, it means deploying capital into a sector with strong structural demand.

The long-term goal flagged in earlier reporting is to support more than 20 gigawatts of AI computing infrastructure by 2028 — an extraordinary number that speaks to how much physical infrastructure the AI industry expects to consume.

What This Means for Business

If you’re running a business that relies on AI services from companies like Anthropic, this kind of deal has practical implications:

Supply security improves. Deals like this are about ensuring AI labs have consistent access to chips at scale, which reduces the risk of compute constraints throttling the AI products your business depends on.

Prices may stabilize. One of the unpredictable variables in enterprise AI has been compute cost. More structured financing for chip supply creates more predictability in the economics of AI services.

The big labs are getting bigger. Capital at this scale tends to concentrate around established players. Anthropic, already valued at roughly $2 trillion by some estimates, is cementing its position at the frontier tier where compute access is guaranteed rather than competitive.

The infrastructure gap widens. Smaller AI providers and open-source alternatives will find it harder to compete on raw compute. For enterprise buyers, this is worth factoring into vendor selection.

The Broader Pattern

This isn’t an isolated move. The past 12 months have seen a wave of large-scale AI infrastructure financing: Microsoft’s partnership with OpenAI, Saudi Aramco’s AI investments, Masayoshi Son’s Stargate commitments. What’s shifting is that traditional financial institutions — private equity firms like Apollo and Blackstone — are becoming core infrastructure players in the AI build-out.

That’s a significant development. When PE firms at that scale start deploying capital into AI compute, it suggests the underlying demand is being treated as durable, not speculative. It also means the financial system is increasingly enmeshed with the AI industry’s infrastructure.

For Enterprise DNA’s clients — whether you’re thinking about AI adoption strategy, data infrastructure, or workforce planning — the takeaway is straightforward: the compute layer underpinning enterprise AI is getting a massive structural investment. The AI tools you’re evaluating today are backed by capital at a scale that would have seemed implausible three years ago.

---

Enterprise DNA helps businesses navigate AI adoption through data education, AI agent deployments, and strategic advisory. Learn more at enterprisedna.co.