Jeff Gundlach suggests that classifying AI chips as an asset class appears to signal a market peak.

      Jeff Gundlach has expressed skepticism regarding Nvidia's initiative with six asset managers to raise over $500 billion for AI infrastructure, questioning the rationale behind using assets of uncertain lifespan as collateral for long-term debt. Mark Cuban likened chips as an investment category to cryptocurrency.

      Gundlach believes Wall Street has revealed its inconsistencies. “Assets of unknown life as collateral for long-term debt?” the CEO of DoubleLine remarked on X, suggesting that Nvidia's financing consortium “is unlikely to endure well.”

      He is referring to a significant agreement. Nvidia recently forged partnerships with Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs, and KKR to develop financing platforms aimed at generating more than $500 billion to enable customers to access its chips.

      His analogy is intentionally irreverent. Gundlach likens it to backing 30-year asset-backed securities with warehouses of bananas, even “newly engineered bananas of unknown life.”

      Beneath the humor lies a concern about duration. The debt is structured for the long term, while the collateral consists of hardware that the industry frequently updates.

      The participants have acknowledged this. KKR’s co-CEOs described computing as essential infrastructure needing “long-duration capital,” while Jensen Huang maintains that ongoing CUDA software development is “extending its useful life and improving its economics over time.”

      These are not equivalent assertions. One indicates that capital must be patient, while the other asserts the asset will remain viable, and the latter requires a significantly greater degree of substantiation.

      Gundlach’s broader argument pertains to recognizing patterns. He noted that no one alerts investors at the peak of risk markets; rather, they should be mindful of claims regarding new asset classes founded on financial engineering, accompanied by what he termed “questionable” credit ratings.

      Others are echoing this sentiment. Mark Cuban stated that “chips as an asset class will be the new crypto,” and Michael Burry has been betting against AI companies, arguing that the processors used today will become outdated before the investments are recouped.

      The credit market has already shown its response. Nvidia’s credit default swaps reached an all-time high after the company announced $750 billion in AI deals, indicating the market's concerns rather than merely discussing them.

      The broader systemic implications are already being considered, with the Bank for International Settlements cautioning that an AI downturn could impact credit markets similarly to the 2008 crisis. Gundlach's objection, however, is more focused and manageable. It is not a question of whether AI will succeed, but whether a graphics processor will still be a viable collateral option in year seven.