Artificial intelligence is the most popular active of Wall Street in 2025. Startups in the sector have raised more than $ 44 billion in the first half, more than last year, and the total investments should approach the $ 200 billion by December, according to Goldman Sachs. But despite the avalanche of capital and rhetoric that AI will forever change the global economy, a MIT study reveals a much less glamorous scenario: 95% of business attempts are failing.
The report, entitled The GenAI Divide: State of AI in Business 2025analyzed as companies from different sectors are integrating tools such as ChatgPT and Microsoft Copilot. The result is discouraging: only 5% of projects can generate “rapid revenue acceleration”, while the overwhelming majority show measurable financial impact.
Read more:
Take your business to the next level with the help of the country’s leading entrepreneurs!
According to the researchers, the reason is in the very architecture of the general AI systems. They do not learn from feedback, do not adapt to context and do not improve over time. Instead of functioning as “autonomous assistants” capable of transforming integer processes, they remain trapped in fragile and cool workflows from the real routine of companies. Thus, at most, they raise the individual productivity of some employees, but are far from changing the profit structure.
This creates a dangerous mismatch. The most optimistic projections already assumed that AI would add more than $ 6 trillion to the global economy by 2030. To get there, productivity gains would have to accelerate at almost exponential pace, something far from the current reality, where even the best systems can only complete 30% of office tasks for which they are tested.
The study also questions the most repeated thesis about the “mass replacement” of jobs. At least in the coming years, MIT says, impacts tend to limit themselves to external cost optimizations, not deep internal restructuring.
Continues after advertising
Read more:
Meanwhile, the financial market continues to precet the opposite. The case of the palantir caught the eye. The company saw its shares shoot more than 100% in the year, supported by the expectation that its AI contracts with governments and companies would create an unstoppable growth trajectory. But in a matter of days, the euphoria became a free fall, the action lost 17% in a week, with $ 73 billion market value evaporated after criticism that its appreciation was disconnected from reality.
The threat of bubble burst softened late of the week after, taking Wall Street a. PALANTIR’s action, for example, rose 2.2% on Friday afternoon. But the study of MIT has a warning that the sector is not out of danger.
Continues after advertising
Even Sam Altman, CEO of OpenAi, the greatest symbol of the revolution, recently admitted that the sector lives a bubble. If the promise does not get out of paper, the risk is not just a damage to startups or in some big techs: it is a broader collapse, as each year without concrete deliveries makes the account heavier, and the disappointment potentially more devastating to the global economy.
