Equities: AI profitability doubts grow – Nordea

Nordea's analysts Kirsti Sunde Midttun and Ole Håkon Eek-Nielsen express growing concerns over the profitability of AI companies. They highlight several structural pressures, including high inference costs, rapid model depreciation, and increasing competition from free, open alternatives. The analysts question the long-term viability of current AI business models and note a shift in investor sentiment, with a growing skepticism towards AI-related equities.

This has led to a rotation out of tech stocks and into cyclical, defensive, and value-oriented sectors. Despite AI's rapid growth, the analysts see several challenges to profitability and adopt a more skeptical perspective on the industry's prospects. The central economic issue for AI developers remains high inference costs, and the leading model companies are currently not profitable.

Frontier models are seen as infrastructure with a short useful life, requiring extraction of value before the technology becomes obsolete. Publishing these models for free also suppresses willingness to pay across the market and undermines the business models of developers who charge for access. Overall, the picture is bleak: building frontier models is expensive, they depreciate quickly, and they face intense competition from both other models and free alternatives.