Wall Street's latest move to package AI chips as a tradeable asset class has drawn a sharp warning from Jeffrey Gundlach, the prominent bond investor known as the "New Bond King," who suggests this trend may actually signal that risk markets are approaching their ceiling.
Gundlach's criticism centers on a major collaboration announced last week by Nvidia Corp, which signed a memorandum of understanding with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to mobilize over $500 billion for AI infrastructure funding. He took to social media to caution that this consortium plan will "likely not stand the test of time."
His core concern revolves around a fundamental mismatch between debt maturity and the useful life of chip technology. In a pointed analogy, he compared using GPUs as collateral for long-term debt to issuing 30-year asset-backed securities against bananas in a warehouse—suggesting that even "brand-new engineered bananas with an unknown shelf life" cannot support a durable credit structure.
Financial Innovation Narratives: A Familiar Prelude to Market Peaks
According to the press release from Nvidia Corp last week, the company and its six institutional partners signed the memorandum of understanding to create a financing platform for customers, helping businesses secure funding for Nvidia chips and broader "AI factory" assets. The participating parties are positioning AI computing power as a long-term infrastructure asset suitable for institutional investors, aiming to attract substantial long-term capital into the market.
The framework's underlying logic is straightforward: securitize the cash flow generated by AI chips, transform them into a financeable asset class through financial innovation, and thereby unlock hundreds of billions of dollars for AI infrastructure development.
Gundlach fundamentally challenges this reasoning. He points out that while AI processors can generate robust cash flows during periods of strong demand, the rapid pace of AI technology evolution means current-generation chips are highly likely to depreciate significantly before the loans financing them reach maturity. The disconnect between debt duration and the actual value lifespan of the collateral forms the core risk of this financing structure.
He also questions the credit ratings that such transactions might rely on, suggesting those ratings may not be dependable. Gundlach wrote that market tops never come with "bell-ringing warnings," but investors should remain alert to narratives of new asset classes built on "dubious" credit ratings under the guise of financial innovation—this is a typical phenomenon accompanying risk markets reaching their extremes.
Rapidly Depreciating Technology Assets Struggle to Support Long-Term Debt Structures
Gundlach's "banana ABS" analogy precisely highlights the structural weakness of this financing model. Bananas have an obvious short shelf life, making the absurdity of issuing 30-year ABS against them readily apparent. The GPU situation is even more complex—while its technological lifecycle is difficult to predict with accuracy, the fast-paced evolution of AI chip architecture means current mainstream products always face the risk of being replaced by more advanced generations.
When the uncertainty of collateral value far exceeds the duration of the debt, the credit foundation of long-term financing products built on such assets cannot be adequately secured. For institutional investors seeking stable long-term returns, this is a risk that cannot be overlooked.
Comments