Are We Headed Towards A Repeat Of The 2008 Financial Crisis? - November 14, 2025

There Aren’t Any NINJA Loans, But There’s A Lot of Incestuous Dealmaking Happening

Every day we read about the latest industry explosion, that is Artificial Intelligence.  We hear about its possibilities, and the endless ways in which it can improve our lives. We also read about the risks to our society that may result from AI and what we should do to prevent real harm from occurring.  There is confusion on how AI will be used and how AI should be used. Amid this confusion, there is even more confusion as to who’s going to regulate this industry.  

And now, in the midst of all this excitement, fear and confusion, Wall Street and other related entities are introducing new creative techniques to finance the massive capital expenditure that is ongoing and will continue for the foreseeable future.  Gartner is estimating that $1.5 trillion will be spent on AI (infrastructure, software, services) in 2025, and that the 2026 spend on AI will exceed two trillion dollars.  McKinsey & Company estimates that annual AI spending will reach $6.7 trillion by 2030.  Up until recently hyperscaler companies like Microsoft, Meta, Google and Amazon were funding the AI buildout themselves.  Not anymore.  As mentioned above, Wall Street and other financing companies are getting into this game.   When I read about these newly emerging capital financing tools being developed, my first thoughts were,  if we are in the midst of an AI bubble, are we on our way to another financial disaster like the real estate crash of 2007 and the resulting financial crisis of 2008? 

A Refresher On The Crazy Schemes That Lead To The 2007 and 2008 Crises

17-18 years ago Wall Street dreamt up all kinds of crazy investment vehicles that were supported by questionable assets (loans and mortgages given to people who had no basis  qualifying for such debt). These investment vehicles were falsely certified as legitimate and secure, and then traded and sold as securities on the basis of these false certifications. The book by Michael Lewis entitled “The Big Short”, and movie of the same name, do a great job describing the insanity that was taking place at the time.  Some of these securities included:

1. Mortgage-Backed Securities (MBS): this was when banks pooled thousands of home mortgages together, then sold pieces of those pools as “bonds” (MBS) to investors.  These were risky because many of the underlying mortgages were subprime — issued to borrowers with poor credit or low income. Rising home prices masked the risk at first. When homeowners began defaulting, the MBS values collapsed. They transformed risky home loans into “safe” AAA-rated securities — until the defaults hit.

2. Collateralized Debt Obligations (CDOs): these were bundles of other bundles — Wall Street took pieces of MBS (including risky tranches) and repackaged them into new securities. These were dangerous because they layered risk on top of risk — essentially “leverage built on leverage.” Rating agencies gave high ratings (AAA) to CDO tranches that didn’t deserve it. When mortgage defaults rose, the whole structure collapsed. “It’s like taking spoiled meat, putting it in a new grinder, and calling it steak.” 

3. Synthetic CDOs: these were even more abstract — these didn’t contain actual loans. They were bets on the performance of other CDOs or MBS, using derivatives called credit default swaps (CDS).  They multiplied exposure — investors could bet on the same mortgage bonds multiple times. So when defaults rose, the losses were magnified many times over. Example: If $1 billion of real mortgages went bad, synthetic CDOs could spread tens of billions in losses.

4. Credit Default Swaps (CDS): these were a kind of insurance policy on bonds — you paid a premium, and if the bond defaulted, you got paid out.  CDS were sold without adequate reserves to cover potential losses. AIG, the giant insurer, sold hundreds of billions worth of CDS on CDOs — and couldn’t pay when defaults surged. This is what led to AIG’s massive $182 billion government bailout; and, finally;

5. Adjustable-Rate Mortgages (ARMs) and “NINJA” Loans:  These were mortgages with very low initial rates that over time reset sharply higher — often issued to borrowers with No Income, No Job, No Assets (“NINJA” loans). 

The heart of the 2008 financial crisis - The meltdown wasn’t caused just by bad loans — it was driven by financial engineering gone wild. Banks, rating agencies, and investors created complex products that seemed safe, but were actually built on shaky foundations. 

What Crazy Schemes Is Wall Street Dreaming Up This Time?

According to an investor offering sheet obtained by the New York Times DealBook, private equity giant, Blackstone, is on the cusp of closing a $3.5 billion commercial-mortgage-backed securities (“CMBS”) offering to refinance debt held by QTS, the biggest player in the artificial intelligence infrastructure market. It would be the largest deal of its type this year. The bonds would be backed by 10 data centers in six markets (including Atlanta, Dallas and Norfolk, Va.) that together consume enough energy to power the city of Burlington, Vt., for half a decade. Blackstone’s offering is part of the latest push in the AI infrastructure financing blitz. 

Now, the tech giants are turning to financing maneuvers that may add to the risk. To obtain the capital they need, hyperscalers (companies such as Google, Amazon, Microsoft and Meta)  have leveraged a growing list of complex debt-financing options, including corporate debt, securitization markets, private financing and off-balance-sheet vehicles. That shift is fueling speculation that A.I. investments are turning into a game of musical chairs whose financial instruments are reminiscent of the 2008 financial crisis. 

Blackstone’s $3.46 billion CMBS offering may seem like small potatoes compared with some other debt-fueled deals, such as Meta’s $30 billion corporate offering to finance its data center in Louisiana. But it’s unprecedented for the CMBS market, where issuance for data-center-backed deals was just $3 billion for all of 2024. To complicate matters further, the share of single-asset-single-borrower securities (SASB) — for example, the assets inside the bond being sold are all from the same company or a single data center — is rising, with 13% of all SASB deals coming from data centers, according to Goldman Sachs. “It’s one company, and these assets are quite similar. If there’s a problem with A.I. data centers, such as if their current chips are obsolete in five years, you could have big losses in these deals. That’s the knock on SASB: When things go bad, they go really bad.” 

Also at play: a financial tool that came into vogue before the financial crisis. Called a special purpose vehicle (S.P.V.), it’s a legal entity that allows a company to take on a lot of debt without having to hold it on its own balance sheet. When Meta structured its $30 billion debt offering for its new data center in Louisiana — the largest private capital transaction on record — Morgan Stanley arranged the debt to sit in one of these custom, off-balance-sheet vehicles. Although the S.P.V. was created to service Meta, the debt technically belongs to the S.P.V., not Meta, which makes Meta look healthier on paper. The maneuver made it easier for Meta to raise another $30 billion in the more traditional corporate bond market. Overall, according to Morgan Stanley, $800 billion in private credit will be needed over the next two years to fund data centers. And S.P.V.s are becoming a more popular way to structure it. Following Meta’s lead, Elon Musk’s xAI is also tapping an S.P.V. to potentially hold $20 billion in debt to buy Nvidia chips and then rent them to xAI.

Are crazy financial instruments spreading the risk of the AI spending frenzy? According to Menlo Ventures, only 3 percent of consumers pay for AI-related services, amounting to about $12 billion per year. If hyperscalers are unable to generate enough profit to offset the costs related to capital expenditures, systemic risk could enter credit markets.

Unlike 2008’s opaque derivatives, today’s creativity is mostly in capital structuring — how investors and firms are funding massive AI buildouts.

Here are the most prominent examples of today’s creative capital infrastructure financing:

🏗️ 1.  AI Infrastructure Securitization

• Some firms (e.g., data-center REITs, hyperscalers’ partners) are bundling long-term AI compute contracts or data-center leases into securitized vehicles. These are similar in spirit to MBS from the 2008 timeframe — they transform future revenue streams (from renting GPUs or compute) into upfront cash. Example: data-center operators or chip lessors selling “AI capacity-backed securities.”

👉 Similarity to 2008: Transforming long-term, uncertain cash flows into tradable assets. 👉 Difference: Underlying assets (servers, power contracts) are tangible and transparent, not subprime mortgages.

2. “Compute Leasing” and AI Hardware Financing

Cloud providers and startups are using leaseback or “as-a-service” models for GPUs. Specialized funds (often backed by private equity or sovereign wealth) finance large GPU clusters, then lease them to AI startups or cloud firms.

👉 Similarity to 2008: High leverage; future earnings must justify upfront capital. 👉 Difference: Hardware retains residual value (unlike bad mortgages).

3. Joint Ventures and Off–Balance Sheet Financing

Tech giants (Microsoft, Google, Amazon) are forming joint ventures with data-center developers, utilities, and chipmakers to share the huge capex burden. Example: Microsoft’s multi-billion-dollar partnerships with energy firms to secure power for data centers. Off-balance-sheet treatment makes corporate balance sheets look lighter — reminiscent of SIVs pre-2008, though more regulated now.

👉 Similarity to 2008: Shifting risk off primary balance sheets. 👉 Difference: Regulatory scrutiny is higher, and assets are income-producing, not speculative.

Where the Hidden Risks May Lurk

While we may not in a systemic bubble (yet), some red flags echo 2007-08 patterns:

1. Overconfidence in a single narrative: “AI demand will grow exponentially forever” = “housing prices never fall.”

2. Leverage in private markets: Many AI infrastructure deals are debt-financed with long-duration paybacks. If rates stay high or demand cools, refinancing risk appears.

3. Concentration risk: Nvidia, Microsoft, Amazon, Google dominate — if one stumbles, valuations and financing terms across the ecosystem could reprice sharply.

4. Opaque private deals: Private credit and venture-backed AI infrastructure deals aren’t marked to market daily — risks may be hidden until liquidity tightens.

   
   

So while AI financing creativity is real, and occasionally worrying, it’s not yet systemically dangerous. The risk is more in overvaluation and overcapacity, not financial collapse.

The Big Picture

• The 2008 structures were toxic complexity hiding weak assets.

• The 2025 AI structures are complex ownership and funding models around strong assets — but with potentially unrealistic growth assumptions.

 Bottom Line

The plumbing of today’s AI finance is sophisticated but more transparent and asset-backed than 2007’s. However, there is a familiar danger: too much capital flowing into a single “hot” sector based on linear growth assumptions. If AI demand plateaus, those financed revenue streams and leveraged data center loans could start to stress the system — especially in private credit. In essence: 2008 was about hidden risk in fake safety.  2025 could become about visible risk in overconfidence.