Box CEO Aaron Levie asserts that tech executives are succumbing to AI psychosis, a delusion driven by their detachment from ground-level operations. Levie argues that CEOs, often insulated from the labor-intensive tasks required to implement AI, overestimate its capabilities after superficial experimentation. This theory gains gravity amid a backdrop of 115,430 layoffs in tech since January 2026, with companies citing AI as justification—a claim Levie dubs "AI washing." While Levie advocates for hands-on AI use to gauge its limits, many leaders remain entrenched in optimism, ignoring data showing no robust link between AI adoption and productivity gains. UC Berkeley’s meta-analysis and MIT’s projections of AI achieving only 80%–95% task success by 2029 underscore the gap between hype and reality.

The phenomenon reflects a broader disconnect: CEOs like Zeb Evans of ClickUp claim to have replaced 22% of staff with AI agents, envisioning a "100x org." Yet MIT researchers warn that AI agents still fall short of human quality, and Harvard Business Review studies highlight that AI’s bottleneck lies in executive oversight. When everyone automates, approval processes choke, risking chaos. Evans’ strategy, while innovative, assumes a future where AI flawlessly executes tasks—a timeline Levie and others deem overly optimistic. The tension between visionary claims and empirical evidence exposes a crisis of credibility in tech leadership.

Investors and businesses face a critical choice: Will they heed Levie’s caution or follow CEOs blindly into organizational chaos? The data suggests AI’s near-term impact is overstated, yet the psychological grip it holds on executives remains potent. Until AI delivers on its promises—or until leaders confront its limits—the tech industry risks repeating past cycles of overpromise and underdelivery. The stakes are not just financial but existential for companies betting on unproven scalability.