Credit Bubble and Neural Networks: UBS Predicts 13% Default Rate

We're used to hearing that AI is the new oil, endless stock growth, and universal prosperity. However, analysts at UBS Group AG decided to add a spoonful of bitter truth to this digital honey. They calculated that if AI begins to reshape the economy too aggressively, the U.S. private credit market could face defaults at the 13% level. This is not just an abstract figure, but a potential financial storm that will affect thousands of companies that failed to jump on the automation train's departing steps. While Silicon Valley celebrates, Wall Street is beginning to nervously eye the debt obligations of mid-market businesses.

Why are they discussing this now? The private credit market has ballooned over recent years to an incredible $1.7 trillion, becoming a quiet haven for investors avoiding the volatility of public markets. Investors have been pouring billions there for years, counting on steady returns from loans to companies that are too small to issue bonds but too large for conventional banks. But UBS reminds us: stability ends where technological singularity begins. If a company's business model was built on processes that are now being replaced by a single neural network or cloud service, that company's chances of repaying the loan are rapidly approaching zero.

In the worst-case scenario laid out by UBS, so-called aggressive disruption would render entire industries unprofitable practically overnight. Bankers fear that the speed of generative AI deployment will significantly outpace the adaptation speed of corporate borrowers. Imagine a company engaged in customer support outsourcing or simple data analysis. It has loans for equipment, offices, and payroll. Then along comes a startup with five employees and an annual subscription to an advanced LLM, doing the same work ten times cheaper. The old company loses contracts, but debts don't go anywhere. 13% defaults—this is a level comparable to deep recession, but this time the cause will be not a mortgage crisis, but software code.

The problem is compounded by the fact that private credit is far less transparent than public debt. We don't know exactly how many skeletons are hidden in the closets of funds lending to mid-market businesses. AI acts here as a catalyst for natural selection. Those who can't integrate algorithms into their processes will simply become extinct. But the irony is that even successful integration requires massive capital investments, which cash-strapped companies may simply not have. It's a vicious circle: to survive in an AI world, you need investments in technology, but to get them, you first need to settle old debts that have become unmanageable because of the same AI.

UBS analysts point out that current optimism about labor productivity growth may be premature for the debt market. Yes, AI will make the economy more efficient in the long term, but the transition period promises to be bloody for balance sheets. This warning should be taken as a signal for investors to start reshuffling their portfolios.

The era when you could simply lend to any stable business is fading into the past. Now creditors will have to understand transformer architecture and automation capabilities as well as they understand balance sheets. If the forecast comes true, we're in for a massive transfer of capital from the old economy to the new one, and this process is unlikely to pass painlessly for those accustomed to a quiet life on interest payments.

The key takeaway: AI is not only a tool for profit, but also a powerful factor of systemic risk to financial stability. Are credit markets ready for the fact that their borrowers may become obsolete much faster than they can repay even the principal?