The AI race is no longer just about building better models.
It is now about who can finance the infrastructure required to run them.
This week, Apollo Global Management and Blackstone finalized a massive $35 billion debt package to help Anthropic expand its AI infrastructure. The money will be used to fund Google’s custom AI chips, which Anthropic will lease to power future versions of Claude and its other AI products.
The size of the deal is remarkable. At $35 billion, it ranks among the largest private credit transactions ever completed.
But the bigger story is what it tells us about the future of AI.
AI Is Becoming an Infrastructure Business
For the last few years, most investors focused on AI models.
OpenAI built ChatGPT.
Anthropic built Claude.
Google developed Gemini.
The conversation revolved around which model was smartest, fastest, or most useful.
Today, the bottleneck is increasingly becoming infrastructure.
Training and running advanced AI systems requires enormous amounts of computing power. Companies need data centers, electricity, networking equipment, and most importantly, specialized AI chips.
Those chips cost billions of dollars.
As AI models become larger and more capable, the capital required to support them continues to rise.
This is creating an entirely new financing market.
What Exactly Is Being Financed?
Interestingly, Apollo and Blackstone are not directly lending money to Anthropic.
Instead, the deal uses a structure commonly seen in aircraft leasing and infrastructure finance.
The process works like this:
- A special purpose vehicle (SPV) raises debt and equity capital
- The SPV purchases Google’s custom AI chips
- Anthropic leases those chips from the SPV
- Lease payments from Anthropic are used to repay investors
This structure allows Anthropic to access massive computing capacity without having to purchase all of the hardware outright.
In simple terms, Anthropic is renting the infrastructure instead of owning it.
That preserves cash while still allowing the company to scale rapidly.
Why Broadcom’s Role Is So Important
One of the most interesting aspects of the deal is the role played by Broadcom.
Broadcom helps Google develop its Tensor Processing Units (TPUs), which are custom AI chips designed specifically for machine learning workloads.
More importantly, Broadcom is providing what is known as residual value support.
That sounds complicated, but the idea is straightforward.
If Anthropic were unable to make lease payments, the chips would be sold.
If the resale value of those chips was not enough to repay investors, Broadcom would cover the shortfall for the senior debt holders.
This protection significantly reduces risk for lenders.
As a result:
- Borrowing costs become lower
- More investors are willing to participate
- Larger financing packages become possible
Essentially, Broadcom’s balance sheet is helping unlock billions of dollars of AI infrastructure financing.
A New Asset Class Is Emerging
Traditionally, investors financed things like:
- Office buildings
- Airports
- Power plants
- Aircraft fleets
- Telecommunications networks
Now AI chips are joining that list.
That may sound surprising today, but it makes sense.
AI hardware generates predictable cash flows through long-term leases and usage agreements.
Institutional investors such as insurance companies, pension funds, and private credit funds are always looking for assets that can produce steady returns over many years.
AI infrastructure is increasingly becoming one of those assets.
This transaction may be remembered as one of the first major examples of AI hardware being financed at infrastructure scale.
The Numbers Are Getting Bigger
A few years ago, a billion-dollar AI investment seemed enormous.
Today, the numbers are reaching entirely different levels.
Consider the recent developments:
- Anthropic recently raised capital at a valuation approaching $1 trillion
- OpenAI continues to invest heavily in infrastructure partnerships
- Microsoft, Amazon, Google, and Meta are spending tens of billions annually on AI capacity
- New data center projects are being announced across multiple continents
The common theme is clear.
The winners in AI may not simply be the companies with the best models.
They may also be the companies that secure the most compute.
What This Means for Investors
Most discussions around AI focus on software companies.
But deals like this highlight another side of the opportunity.
The AI ecosystem includes:
- AI model developers
- Semiconductor companies
- Data center operators
- Power generation companies
- Networking providers
- Infrastructure financiers
As the industry matures, value creation may spread across the entire ecosystem rather than concentrating solely within AI labs.
The financing layer itself could become a major investment theme.
Just as real estate created REITs and aircraft created leasing companies, AI infrastructure may create entirely new categories of financial products.
The Bigger Picture
The most important takeaway from this deal is not the $35 billion headline.
It is the realization that AI is evolving from a software story into a capital-intensive infrastructure story.
Building frontier AI models now requires enormous amounts of hardware, energy, and financing.
That means Wall Street is becoming just as important to the AI race as Silicon Valley.
Apollo, Blackstone, Broadcom, Google, Anthropic, and other large institutions are effectively building the financial plumbing needed to support the next generation of AI systems.
This deal may be one of the clearest signs yet that AI’s next phase will be funded not only by venture capital, but by the global credit markets themselves.