'The honeymoon is over': Deutsche Bank warns of 'most challenging year ever' for AI

The "honeymoon" for investors in AI stocks is over, Deutsche Bank said. Adrian Cox and Stefan Abrudan, analysts at Germany's largest bank, warned clients that 2026 will be "the most difficult year" for AI . It will be shaped by three converging trends: disillusionment, imbalance and distrust.

The first factor - frustration - is related to the transition of companies from pilot projects to full-fledged AI implementation. Businesses are faced with the problems of model accuracy, unpredictability of results, and the unobvious cost-effectiveness of technology compared to human labor. For many businesses, this process resembles "not replacing a horse with a tractor, but rather switching to a more comfortable saddle," Seeking Alpha quotes Deutsche Bank analysts as saying.

The second theme, market imbalance, refers to the gap between sharply rising demand and limited capacity supply, which provokes supply chain disruptions, power shortages and skills shortages.

The third trend, mistrust, reflects growing public and regulatory concerns. "Anxiety about AI will shift from a quiet hum to a loud roar this year. It will be reflected in lawsuits over everything from copyright to privacy, data center siting and protecting young people from chatbots that encourage self-harm or worse," CNBC quoted a note from bank analysts. In the list of key concerns, the experts also included job losses in AI-affected sectors: "Disguising layoffs as the effect of AI implementation will be an important feature of 2026," they wrote.

Deutsche Bank predicts that 2026 will be a pivotal year for independent AI model developers, including OpenAI. "OpenAI is in a particularly vulnerable position and perhaps most at risk as it does not yet appear to have found a workable business model to cover losses of $9 billion last year and likely $17 billion this year," Barron's quoted the bank's analysts as saying in a note. The pressure on OpenAI will only intensify as it gets closer to an IPO, which could take place in early 2027, they said.

Despite volatility and fears of a bubble at the end of 2025, AI stocks ended the year with strong results, outperforming the broad U.S. stock market and the technology sector, investment research agency Morningstar recalls. Midway through the fourth quarter, these securities traded at an average 18 percent premium to their fair value, but after a correction at the end of the year, the premium fell to 4 percent, the agency calculated. As a result, shares of Marvell, Adobe and Salesforce look undervalued, while last year's growth leaders Micron, Lam Research and Palantir are still overvalued, Morningstar said.

This article was AI-translated and verified by a human editor