The tech industry is experiencing a peculiar kind of madness. In 2026, with artificial intelligence (AI) becoming more accessible, many CEOs are making sweeping decisions based on what appears to be a delusional belief in AI's immediate capabilities. This phenomenon, termed 'AI psychosis' by Box founder Aaron Levie, is leading to mass layoffs, organizational chaos, and a disconnect between executive vision and operational reality.
The Origin of the Term
Aaron Levie, the long-time CEO of cloud storage company Box, recently took to X (formerly Twitter) to describe the mindset he sees pervading C-suites. 'CEOs are uniquely prone to AI psychosis because they’re sufficiently distant from the last mile of work that still has to happen to generate most value with AI,' he wrote. Levie explained that executives often demo AI tools, see a prototype generate a contract or a piece of code, and then leap to the conclusion that entire workflows can be automated. However, they are not the ones who must review the code for bugs, handle edge cases, or integrate AI outputs into complex business processes. This disconnect, he argues, is dangerous.
Levie is not an AI skeptic. His X feed is filled with positive takes on AI, and he has invested in multiple AI startups. Yet even he cautions against the euphoria. His warning echoes previous periods of technological hype, such as the cloud computing boom of the early 2010s, when companies spent lavishly on cloud services before realizing the hidden costs and complexity. The difference now is the scale: AI promises to replace not just infrastructure but knowledge workers themselves.
The Evidence of Psychosis
The most visible symptom is the wave of layoffs sweeping the tech sector. According to Layoffs.fyi, in just the first five months of 2026, 115,430 people have been laid off from 152 tech companies. That is nearly as many as the entire 2025 total of 124,636 layoffs across 275 companies. The majority of these companies cite AI as a reason for the cuts. In some cases, the link is direct: Zeb Evans, CEO of ClickUp, proudly laid off 22% of his workforce after deploying 3,000 AI agents for internal tasks. Evans claimed the move was not about cost reduction but about creating a '100x organization' where humans oversee agents. However, critics argue this is a form of AI washing—using AI as a convenient excuse for cost-cutting measures driven by other factors like investor pressure or overhiring.
The data on AI productivity gains is far from convincing. A meta-analysis published in the California Management Review in October 2025 found 'no robust relationship between AI adoption and aggregate productivity gain.' Another study from the National Bureau of Economic Research (March 2026) acknowledged that AI improves productivity but noted a 'productivity paradox' where perceived gains exceed measured gains. This gap between hype and reality is the hallmark of AI psychosis: CEOs believe AI is transforming their companies, but the metrics do not support it.
Why CEOs Are Especially Vulnerable
Levie's theory highlights a structural issue in corporate hierarchies. CEOs are responsible for strategic vision, not operational execution. They see AI's potential in demos and prototypes, but they rarely encounter the gritty details of implementation. For example, a CEO might see an AI agent draft a legal contract in seconds. What they do not see are the hours spent training the model on company-specific clauses, the meticulous review required to catch outdated references, or the hallucinations where the AI invents non-existent legal precedents. Similarly, in software development, AI can generate code quickly, but that code often needs extensive debugging and security analysis before deployment. The 'last mile' of work—testing, validation, integration—remains stubbornly human-intensive.
This phenomenon is not entirely new. During the early days of cloud computing, executives often signed hefty contracts for cloud services without understanding the operational challenges of migration, leading to runaway costs and project failures. Now, with AI, the stakes are higher because the technology claims to replace cognitive labor, not just infrastructure. The difference is that AI agents today are still unreliable. MIT researchers recently concluded that agents are not yet producing human-quality work, and they predicted that by 2029, AI models will achieve 80–95% success rates on text tasks at a minimally sufficient level. Outperforming humans will take a few more years. Yet CEOs are acting as if that future is already here.
The Organizational Impact
When CEOs act on AI psychosis, the consequences ripple through their organizations. Layoffs create anxiety and erode trust. Survivors are often burdened with overseeing agents, a role that requires new skills and can be more stressful than doing the work themselves. A study in the Harvard Business Review pointed out that when AI enables everyone to produce more output, the bottleneck shifts to executives who must authorize all the new work. If decision-making authority is not reorganized, chaos ensues. OpenAI experienced this firsthand in 2025 when rapid AI tool adoption led to governance issues and internal friction.
Moreover, mass layoffs may backfire. If AI cannot yet handle the 'last mile' tasks, companies that cut too deep may find their quality declining, customer satisfaction dropping, and competitive advantage eroding. The productivity gains that CEOs dream of may never materialize. Instead, they may get a workforce that is demoralized, overworked, and forced to fix errors made by overconfident AI agents.
What Should CEOs Do Instead?
Levie recommends a pragmatic approach: use AI extensively but with full awareness of its limits. 'CEOs should use AI a ton, test its capabilities and limitations, and come out the other side with an appreciation for both the upside and the real work,' he said. This means spending time with the technology at the operational level, not just in executive briefings. It also means being honest about productivity metrics, avoiding the temptation to exaggerate AI's impact to investors or the public. The sobering data from academic research should serve as a warning: AI is a powerful tool, but it is not a miracle cure for efficiency. The tech industry may need to relearn the lesson of every previous technology wave: innovation requires patience, investment, and human expertise. Until CEOs overcome their AI psychosis, the layoffs and organizational turmoil are likely to continue, leaving a trail of burned-out workers and unfulfilled promises.
Source: TechCrunch News