Sam Altman calls AI a “bubble” while seeking $500B valuation for OpenAI

During a recent private dinner, Sam Altman, the CEO of OpenAI, expressed concerns about the current excitement surrounding AI investments. He suggested that many investors may be overly enthusiastic, stating that someone could potentially face massive financial losses. This commentary comes as OpenAI is in discussions for a secondary share sale, aiming for a staggering $500 billion valuation—an increase from $300 billion just a few months prior. Altman likened the current AI investment landscape to the dot-com bubble of the late 1990s, asserting that the market might be experiencing an overzealous phase. He also made bold predictions about the future of OpenAI, indicating that the company may soon invest trillions in data center construction. Furthermore, he claimed that ChatGPT could eventually cater to billions of users daily, which would mean reaching nearly half of the global population—a target that many view as overly ambitious compared to Facebook's approximately 3 billion monthly active users. These remarks coincided with the release of new research from MIT, which highlighted a significant number of failures in enterprise AI initiatives. According to the study titled 'GenAI Divide: State of AI in Business 2025,' a staggering 95 percent of enterprise AI pilots do not yield the expected revenue growth. The research included an analysis of 300 public AI deployments and surveys of 350 employees, alongside 150 interviews with business leaders. Critics, however, have pointed out that the findings seem to reiterate well-known challenges: while simple tools like ChatGPT are popular, complex enterprise systems often fail to meet user needs. The study attributes these failures to issues in implementation rather than deficiencies in AI model quality. Fortune reported that proprietary AI tools succeed 67 percent of the time, whereas internally developed systems succeed only about 33 percent of the time. This trend suggests that the difficulties lie more in corporate IT departments' attempts to create their own solutions rather than in the capabilities of existing AI technologies.