Michael Burry Nvidia Warning: AI Boom Built on Customer Concentration and Hidden Debt Risks

TLDR:

• Nvidia now depends on three customers for 64% of receivables, increasing concentration risk sharply.
• Burry argues much AI spending reflects benchmarking activity rather than sustainable long-term demand.
• Tech giants reportedly hold $662 billion in off-balance-sheet AI commitments hidden from investors.
• Private equity, insurers, and offshore reinsurers may amplify risks if AI spending slows abruptly.

Michael Burry has raised fresh concerns about Nvidia’s revenue structure, warning that the chipmaker’s financials rest on a dangerously narrow customer base.

Burry argues that current AI spending patterns resemble a temporary buildout phase rather than permanent demand.

He introduces the concept of the “bezzle” to describe inflated spending that may sharply reverse. His analysis also connects Nvidia’s risk to a broader web of hidden financial commitments across the tech sector.

Nvidia’s Customer Concentration Raises Red Flags

Michael Burry Nvidia analysis centers on a striking shift in accounts receivable data. Three customers now account for 64% of Nvidia’s total accounts receivable.

That figure stood at 33% in 2020, meaning concentration has nearly doubled in just a few years. The jump of eight percentage points in a single quarter alone draws attention.

This level of concentration means Nvidia’s revenue depends heavily on the spending decisions of very few buyers. Any slowdown from those buyers would create a sizable gap in Nvidia’s reported numbers.

Burry describes the current AI spending environment as companies “flying empty airplanes around.” The reference points to benchmarking activity, model testing, and leaderboard competition rather than real, recurring demand.

Burry’s core argument is that this benchmarking phase will eventually end. When it does, those concentrated customers will have far less reason to maintain current chip order volumes.

The financial model holding Nvidia’s growth narrative together may then face a serious stress test. Markets have largely priced in sustained demand, which makes any deceleration more painful.

The danger is not that artificial intelligence as a technology is fraudulent. Rather, the concern is that a large portion of today’s AI infrastructure spending serves a temporary competitive signaling function. Once that function is served, the underlying procurement rationale shifts considerably.

Hidden Commitments and Offshore Risk Structures

Beyond Nvidia’s customer base, Burry points to a broader financing architecture that amplifies the risk. Microsoft, Amazon, Alphabet, Meta, and Oracle carry a combined $662 billion in off-balance-sheet AI commitments, according to Moody’s.

Standard accounting rules allow companies to exclude these figures from reported financials entirely. That means the true scale of AI infrastructure obligations remains largely invisible to public investors.

Private equity firms have moved to finance this buildout by acquiring life insurance companies. Those insurers collect premiums from ordinary policyholders and redirect that capital into the PE firms’ own illiquid assets.

The risk is then pushed offshore through captive reinsurers set up in Bermuda, where capital requirements are lighter. Each layer of this structure adds distance between the underlying risk and public disclosure.

The interlocking nature of these arrangements is what makes a potential unwind so disruptive. The same PE firms own the insurance vehicles funding AI debt.

The same Bermuda structures carry the reinsurance exposure. If a major hyperscaler exits a data center commitment, every connected counterparty faces pressure simultaneously.

Burry’s warning, therefore, covers two overlapping risks. One is Nvidia-specific and tied to customer concentration.

The other is systemic, involving hundreds of billions in commitments that have been structured to remain off the books until they cannot be.

The post Michael Burry Nvidia Warning: AI Boom Built on Customer Concentration and Hidden Debt Risks appeared first on Blockonomi.