As Wall Street attempts to reshape AI chips into a new asset class suitable for long-term allocation, some astute market observers have already detected potential crises. Jeffrey Gundlach, CEO of the renowned investment firm DoubleLine Capital, pointed out that the over $500 billion AI infrastructure financing plan promoted by Nvidia in collaboration with several top financial institutions might be a clear signal that the risk market is approaching a cyclical peak. He struck at the core of the issue, questioning whether rapidly iterating compute hardware possesses the qualifications to support long-term debt.

The focus of this controversy lies in a massive financing blueprint recently disclosed by Nvidia. The company has reached preliminary intentions with Apollo, BlackRock, Blackstone, Brookfield Asset Management, Goldman Sachs, and KKR to build a dedicated financing platform aimed at raising over $500 billion for AI infrastructure construction. This huge sum of money will primarily be used to help enterprises purchase Nvidia chips and expand computing centers. The financial giants involved expect to define computing power as a long-term asset similar to traditional infrastructure, thereby attracting long-term capital such as pension funds and insurance funds. This means that AI chips are gradually shedding their pure technological equipment attributes and being deeply packaged as financial targets that are financeable, collateralizable, and possess long-term holding value.

However, Gundlach sees a massive logical flaw in using rapidly iterating hardware as collateral for long-term debt. He frankly stated that although current demand for AI chips is strong and generates considerable returns, the pace of their technological evolution is staggering. Today's top-tier processors could be completely outpaced by new-generation products in just a few years. If the financing debt has a long maturity while the collateral value shrinks rapidly with technological iterations, a severe maturity mismatch will occur between the actual lifespan of the assets and the debt cycle. Gundlach even joked that this is tantamount to issuing 30-year asset-backed securities with the underlying collateral being a batch of bananas with an uncertain shelf life. He bluntly stated that this massive financing alliance plan would hardly withstand the test of time.

In Gundlach's analytical framework, such phenomena are often highly correlated with market cycle peaks. He mentioned that historical experience shows that when the market is at a high level, there will always be someone claiming to have invented a brand-new asset class, accompanied by complex financial engineering and questionable credit ratings. The securitization of AI computing power possesses exactly these typical characteristics. This is not a wholesale denial of the long-term prospects of the AI industry, but rather a reminder to investors that when financial markets begin to wildly leverage and design complex securitization structures around a single hot asset, the risk may have quietly spread from equity valuations to the credit market.

Billionaire investor Mark Cuban echoed this sentiment, bluntly stating that chips as an asset class could evolve into a new cryptocurrency. This also points to the market endowing hardware with overly strong financial attributes and attempting to build a new speculative trading system around its price and financing. This view coincides with previous warnings from renowned investor Michael Burry. Burry has always been skeptical of the massive capital expenditure of large tech companies on AI chips, believing that against the backdrop of surging technology, some heavily purchased equipment could lose its economic value before the end of its depreciation period.

Stripping away the surface, the essence of this debate is not questioning the sustainability of AI demand, but returning to a fundamental financial principle: the matching degree of asset and liability maturities. If the demand for computing power continues to surge and data centers maintain high utilization rates, the cash flow generated by the chips is sufficient to cover financing costs, and the logic of related asset securitization can be self-consistent. Conversely, once technological iterations exceed expectations or the cost per unit of computing power drops sharply, the shrinkage rate of the residual value of old equipment will far exceed the model predictions of financial institutions. At that time, investors may face the dilemma of loans not yet cleared while the collateral has depreciated significantly.

Currently, the capital expenditure model in the AI field is undergoing a profound shift. As the volume of infrastructure investment expands exponentially, the involvement of private credit, asset-backed financing, and large asset management institutions is deepening day by day. The prosperity of AI is crossing the boundaries of tech stocks and penetrating deeply into the bond and credit markets. The $500 billion financing blueprint has been laid out, and the gears of computing power assetization are accelerating. Meanwhile, the credit risks hidden behind the maturity mismatch, with the deep involvement of private credit and asset management institutions, are quietly spreading to the broader bond market.