The NANDA report claims that 95% of AI enterprise projects fail, which has sparked concern among industry watchers. Despite the alarming statistic, the author argues that this failure rate is not unique to AI. Previous studies show high failure rates in general IT projects—84% according to a Forbes interview and up to 98% for large, complex projects per the CHAOS report. The NANDA report defines success as projects that lead to noticeable productivity or profit gains, a stricter standard than some other metrics in IT. AI projects face unique challenges. The technology is still relatively new, with useful models like GPT-4 only emerging recently. Most enterprise AI projects are not based on well-understood problems, making them inherently complex. The author questions the reliability of the NANDA report's 95% failure statistic, pointing out that it may be based on a limited set of interviews and unclear definitions of success. For instance, only 8.3% of surveyed companies that pursued AI projects achieved a notable impact, suggesting that the failure rate may not be as grim as presented. The discussion highlights that the landscape of enterprise AI is still developing. Companies are grappling with various approaches and best practices, which adds to the complexity of these projects. The high failure rate might reflect the growing pains of a new technology rather than a fundamental flaw in AI's potential to deliver value in business.