If you thought the AI revolution was all about chips, code, and compute—think again. The unsung hero (or ticking time bomb) of this story might just be… private credit.
Welcome to the trillion-dollar question: who exactly is financing the infrastructure needed to power the AI future?
The answer, increasingly, is a shadowy constellation of non-bank lenders, structured products, and funds sitting outside the traditional financial system. And while that may sound efficient and exciting, recent cockroach sightings suggest we might want to turn the lights on.
AI’s $6 Trillion Appetite
Let’s start with the scale of AI’s infrastructure demand.
Whether it’s ChatGPT-5 or DeepSeek, one thing is clear: the “bitter lesson” of AI is that performance scales with compute. And compute needs data centers. McKinsey estimates $6.7 trillion will be needed for data center infrastructure by 2030. Morgan Stanley forecasts $2.9 trillion of AI infrastructure spending by 2028 alone.
Breakdown of capital expenditure for data center infrastructure
Source: McKinsey
These numbers dwarf the annual capital expenditures for AI investments of the entire S&P 500 this year at $1.2 trillion.
Where is all this money coming from?
- $1.4 trillion from hyperscaler cash flows (Amazon, Google, Meta, etc.)
- $800 billion from private credit
- $200 billion from corporate bonds
- $150 billion from securitized assets (ABS, CMBS)
- $350 billion from PE, VC, sovereigns, and banks
Source: Morgan Stanley
In short, roughly 40% of this spending will be debt-financed, and the largest slice of that debt is expected to come from private credit.
The Cockroach Problem
In October, JPMorgan CEO Jamie Dimon dropped an eerie warning: “When you see one cockroach, there are probably more.”
He was referring to the collapse of Tricolor Holdings, a subprime auto lender that filed for bankruptcy after allegedly double-pledging collateral. The kicker? Big banks had lent hundreds of millions to Tricolor where some of it through opaque, off-balance-sheet channels.
It didn’t stop there. Soon after:
- First Brands Group, an auto-parts supplier, filed for bankruptcy due to short-term loan issues.
- Zions Bancorp and Western Alliance disclosed exposure to credit-related fraud.
- 777 Partners, a private investment firm, saw one of its co-founders charged with conspiracy for—you guessed it—double-pledging loans.
The sector may be labeled “private,” but the consequences are public.
This “cockroach chain” highlights a growing risk: as traditional banks get regulated out of risky lending, they’re increasingly funding the funds that do it anyway. The loop looks like this:
And we’re currently in the “Banks Lend to Private Credit Funds” section
Shadow Finance Lights Up AI
So how did private credit become such a major player in the AI buildout?
The rise of non-depository financial institutions (NDFIs) plays a key role. These are lenders that don’t take deposits like traditional banks — think private credit funds, specialty finance companies, and other “shadow banks” operating outside typical regulatory oversight. A decade ago, they accounted for just 3.6% of all U.S. bank lending. Today, that number is 10.4%, representing $1.2 trillion in loans, nearly half of which goes to private credit and private equity firms.
Share of U.S. bank lending to NDFIs has tripled since 2015
Source: Moody’s
Where do loans to NDFIs go? Nearly half goes to private credit and private equity
Source: Moody’s
Why?
Because AI infrastructure is uniquely complex, global, and capital-hungry. Building a GPU-loaded data center isn’t like building a mall. It requires:
- Customized, asset-backed financing structures
- Speed and scale
- Global reach
- A tolerance for regulatory grey zones
Private credit fits that bill. It’s fast, flexible, and invisible—until it isn’t.
From Bubble Watch to Regime Shift?
It’s tempting to warn of a bubble. High valuations. Accelerating capex. Shadow finance. We’ve seen this before.
But perhaps the better frame is this: private credit is not a flaw — it’s the vessel.
If AI turns out to be a tool rather than a revolution, the loans may look reckless in hindsight. But if AI continues to reshape compute, productivity, and power, then we may look back at this moment as the early days of a capital formation regime shift.
What used to be considered risky (opaque, unrated, bespoke debt) is being normalized, because the magnitude of the opportunity demands it.
Closing Thought: Trillions Don’t Tiptoe
We’re now financing the future in trillions, not billions. That changes the scale — and the stakes.
Whether private credit is a ticking time bomb or a necessary lubricant depends on what you believe about AI itself. If the promise materializes, the financing structure will be seen as bold and adaptive. If not, we’ll be picking through the ashes of another misallocated boom.
But either way, the financial architecture is evolving. Private credit isn’t just enabling the AI race. It’s being shaped by it.
Exponential ambitions require exponential mechanisms. And sometimes, those mechanisms are invisible—until they’re not.
Tara Mulia
For more blogs like these, subscribe to our newsletter here!
Admin heyokha
Share
If you thought the AI revolution was all about chips, code, and compute—think again. The unsung hero (or ticking time bomb) of this story might just be… private credit.
Welcome to the trillion-dollar question: who exactly is financing the infrastructure needed to power the AI future?
The answer, increasingly, is a shadowy constellation of non-bank lenders, structured products, and funds sitting outside the traditional financial system. And while that may sound efficient and exciting, recent cockroach sightings suggest we might want to turn the lights on.
AI’s $6 Trillion Appetite
Let’s start with the scale of AI’s infrastructure demand.
Whether it’s ChatGPT-5 or DeepSeek, one thing is clear: the “bitter lesson” of AI is that performance scales with compute. And compute needs data centers. McKinsey estimates $6.7 trillion will be needed for data center infrastructure by 2030. Morgan Stanley forecasts $2.9 trillion of AI infrastructure spending by 2028 alone.
Breakdown of capital expenditure for data center infrastructure
Source: McKinsey
These numbers dwarf the annual capital expenditures for AI investments of the entire S&P 500 this year at $1.2 trillion.
Where is all this money coming from?
- $1.4 trillion from hyperscaler cash flows (Amazon, Google, Meta, etc.)
- $800 billion from private credit
- $200 billion from corporate bonds
- $150 billion from securitized assets (ABS, CMBS)
- $350 billion from PE, VC, sovereigns, and banks
Source: Morgan Stanley
In short, roughly 40% of this spending will be debt-financed, and the largest slice of that debt is expected to come from private credit.
The Cockroach Problem
In October, JPMorgan CEO Jamie Dimon dropped an eerie warning: “When you see one cockroach, there are probably more.”
He was referring to the collapse of Tricolor Holdings, a subprime auto lender that filed for bankruptcy after allegedly double-pledging collateral. The kicker? Big banks had lent hundreds of millions to Tricolor where some of it through opaque, off-balance-sheet channels.
It didn’t stop there. Soon after:
- First Brands Group, an auto-parts supplier, filed for bankruptcy due to short-term loan issues.
- Zions Bancorp and Western Alliance disclosed exposure to credit-related fraud.
- 777 Partners, a private investment firm, saw one of its co-founders charged with conspiracy for—you guessed it—double-pledging loans.
The sector may be labeled “private,” but the consequences are public.
This “cockroach chain” highlights a growing risk: as traditional banks get regulated out of risky lending, they’re increasingly funding the funds that do it anyway. The loop looks like this:
And we’re currently in the “Banks Lend to Private Credit Funds” section
Shadow Finance Lights Up AI
So how did private credit become such a major player in the AI buildout?
The rise of non-depository financial institutions (NDFIs) plays a key role. These are lenders that don’t take deposits like traditional banks — think private credit funds, specialty finance companies, and other “shadow banks” operating outside typical regulatory oversight. A decade ago, they accounted for just 3.6% of all U.S. bank lending. Today, that number is 10.4%, representing $1.2 trillion in loans, nearly half of which goes to private credit and private equity firms.
Share of U.S. bank lending to NDFIs has tripled since 2015
Source: Moody’s
Where do loans to NDFIs go? Nearly half goes to private credit and private equity
Source: Moody’s
Why?
Because AI infrastructure is uniquely complex, global, and capital-hungry. Building a GPU-loaded data center isn’t like building a mall. It requires:
- Customized, asset-backed financing structures
- Speed and scale
- Global reach
- A tolerance for regulatory grey zones
Private credit fits that bill. It’s fast, flexible, and invisible—until it isn’t.
From Bubble Watch to Regime Shift?
It’s tempting to warn of a bubble. High valuations. Accelerating capex. Shadow finance. We’ve seen this before.
But perhaps the better frame is this: private credit is not a flaw — it’s the vessel.
If AI turns out to be a tool rather than a revolution, the loans may look reckless in hindsight. But if AI continues to reshape compute, productivity, and power, then we may look back at this moment as the early days of a capital formation regime shift.
What used to be considered risky (opaque, unrated, bespoke debt) is being normalized, because the magnitude of the opportunity demands it.
Closing Thought: Trillions Don’t Tiptoe
We’re now financing the future in trillions, not billions. That changes the scale — and the stakes.
Whether private credit is a ticking time bomb or a necessary lubricant depends on what you believe about AI itself. If the promise materializes, the financing structure will be seen as bold and adaptive. If not, we’ll be picking through the ashes of another misallocated boom.
But either way, the financial architecture is evolving. Private credit isn’t just enabling the AI race. It’s being shaped by it.
Exponential ambitions require exponential mechanisms. And sometimes, those mechanisms are invisible—until they’re not.
Tara Mulia
For more blogs like these, subscribe to our newsletter here!
Admin heyokha
Share
