US AI stocks are set to plunge not because the economics do not add up but because the underlying product is next to worthless

There is a big difference between a genunely potentially super powerful new tech in its infancy which needs trillions to be developed and a useless, flawed, overhyped tech which will never bring the results it promises.

Unfortunately, US AI belongs to that category.

Too much attention was spent on chips, clouds, infrastructure and not enough on understanding what kind of tasks AI was supposed to do to raise productivity and how it could be designed to do them profitably.

Bill Gates is a big part of the problem. The arrogant and "really evil" billionaire was a key figure in developing frontier AI and LLM. But he never thought to ask the most basic questions about the goals, objectives of AI and how it could improve productivity.

A solipsistic LLM model which rerferences only its own data base to solve every problem as a string of mere words put together by chaotic algorithms while using vast amounts of energy, chips,money, is a priori a fundamentally flawed and uneconomical product.

Despite lacking education, cluture, curisoty, he really believes  whatever random thought goes through his mind must be an act of genius and worth hyperscaling and spending trillions on and there has been no one to tell Gates the truth about his abilities.

 From media

Growing skepticism now challenges the economics of frontier AI, with critics including operators, auditors, and regulatory filings. Leaked audited financials reveal OpenAI's 2025 operating loss near $21 billion on $13.07 billion revenue, with total costs around $34 billion, as it prepares for an IPO. Palantir CEO Alex Karp slammed the token business model as "insane," arguing enterprises gain little value while surrendering competitive advantages. Intellectual property concerns are also rising, exemplified by Anthropic's product launch impacting Figma. Oracle, a key compute provider for OpenAI, warned of nonpayment risks, and its massive Stargate data center project faces financing hurdles. Even Alphabet is raising $80 billion for AI, highlighting capital intensity. These credible voices question whether the AI model layer is a sustainable business or a subsidized commodity heading for a price war, a thesis to be tested by OpenAI's public offering.

https://www.forbes.com/sites/josipamajic/2026/07/04/credible-ai-lab-critics-pile-up-as-the-bubble-math-worsens/