OpenAI chief executive Sam Altman has cautioned that the rapid run-up in artificial-intelligence valuations is starting to look like a bubble, telling audiences that investors are “overexcited” and warning that some could lose “phenomenal” amounts of money if expectations do not match reality. Altman drew parallels with the dot-com era, saying the technology’s long-term importance is unquestioned but short-term enthusiasm may have outrun fundamentals. Altman’s remarks landed the same week MIT published “The GenAI Divide: State of AI in Business 2025,” a study finding that 95 percent of corporate generative-AI pilot projects fail to improve profit and loss statements. Based on 150 executive interviews and reviews of 300 deployments, the report estimates U.S. companies have poured $30-40 billion into the technology, yet only 5 percent of pilots achieve rapid revenue acceleration. Researchers blamed poor workflow design and a focus on sales and marketing tasks with lower returns. The twin signals are fueling a broader debate over whether AI equity prices can be sustained. Shares of high-profile beneficiaries such as Nvidia and Palantir have edged lower in recent sessions as investors reassess growth assumptions, although bulls including Wedbush analyst Dan Ives contend the industry remains in the early innings of a multiyear build-out. The divergent views underscore the growing divide between AI’s strategic potential and its still-uncertain near-term pay-off for most businesses.