‘The cost of compute is far beyond the costs of the employees’: Nvidia executive says right now AI is more expensive than paying human workers

Major technology firms are executing significant workforce reductions and accelerating artificial intelligence spending despite mounting evidence that AI remains economically uncompetitive with human labor on a cost basis. The apparent contradiction between labor cuts and AI investment reveals a market dynamic where capital deployment decisions may be driven by strategic positioning rather than demonstrated productivity gains.

Bryan Catanzaro, vice president of applied deep learning at Nvidia, recently articulated this cost disparity in direct terms, telling Axios that compute expenses for his team substantially exceed employee compensation. This statement from a senior executive at a cornerstone AI infrastructure provider carries particular weight given Nvidia's central role in the AI capex cycle. An MIT study from 2024 corroborates this experience at enterprise scale: researchers analyzing the technical requirements for AI models to perform human-level work found that automation would be economically justified in only 23% of roles where vision comprises the primary workload. The remaining 77% of positions showed continued human labor as the more cost-effective option.

Despite this evidence gap, Big Tech continues deploying capital at unprecedented scale. According to Morgan Stanley, technology firms have announced $740 billion in capital expenditures this year—a 69% increase from 2025—with much of this directed toward AI infrastructure and development. This spending magnitude has prompted some companies to fundamentally restructure budget allocations. Concurrent workforce reductions at Meta (10% headcount reduction plus 6,000 eliminated open positions) and Microsoft (its largest voluntary buyout offering on record) suggest firms are pursuing parallel strategies: reducing near-term labor costs while making long-dated bets on AI infrastructure.

The economic logic underlying this divergence remains unclear. The Yale Budget Lab has noted an absence of widespread data supporting job displacement from AI, contradicting the labor-replacement narrative often cited in public commentary. Furthermore, documented reliability failures—including cases where AI systems have caused significant infrastructure damage through autonomous decision-making—add operational risk to the cost calculus. Until AI systems demonstrate both superior economics and consistent reliability in production environments, the current trajectory of compute spending relative to its economic justification presents a notable investment thesis: whether current AI capex represents rational hedging against future disruption or a speculative bubble in technology infrastructure.