AI Industry Faces Squeeze as Hardware Costs Rise

The artificial intelligence industry is facing a structural economic squeeze characterized by collapsing token prices and rising costs for essential hardware. Semiconductor suppliers, including Nvidia Corp. and Samsung Electronics Co., have increased prices for servers and advanced chipmaking by up to 15% due to structural undersupply. Memory chip manufacturer SK Hynix is exploring joint-venture models to fund new plants to address a memory crunch.

Financial pressure is also mounting in credit markets. Broadcom Inc. is in talks to raise over 60 billion dollars in debt to fund AI chips, but the cost of insuring such debt is rising as investors express nervousness over financing.

Despite high adoption rates among U.S. businesses and strong cloud revenue for hyperscalers, Milos Maricic, founder of the AI advisory firm Maximand, reports a lack of realized productivity gains. A Maximand study of the 60 largest U.S.-listed financial firms found that only one could report a realized dollar return from AI across three years of earnings calls. Investors are consequently shifting their focus from the viability of the technology to the uncertainty of which entities will ultimately capture the financial payoff.