AI is more expensive than human employee, admits Nvidia VP: 'Computational cost exceeds salary'

Bryan Catanzaro, Vice President of Deep Learning at Nvidia, went straight to the point: the cost of running artificial intelligence is higher than paying a human employee. The statement, made in an interview with Fortune, contrasts with the big tech race to implement AI in all areas, even without consistent evidence of real productivity gains.

Why does the cost of AI exceed human salary?

Catanzaro explained that the computational cost involves not only hardware acquisition, but also energy consumption, refrigeration, maintenance and, above all, the need to train and adjust models repeatedly. While a human employee has a fixed monthly cost (wage + charges), the AI requires continuous and unpredictable investments in processing capacity. A recent study showed that employees spend 6.4 hours a week 'taking care' of AI — and 75% claim that productivity increases (read more), which can even reduce gains.

Impact of Nvidia’s claim on the AI market

The statement by a senior executive of Nvidia, a company that dominates the GPU market for AI, draws attention because part of the ecosystem itself that benefits most from the massive adoption of AI. If even Nvidia admits that replacing humans with AI may not be economically advantageous, companies that invest billions in automation may need to reassess their strategies. However, Catanzaro points out that AI is still valuable for specific tasks of high complexity, where the human cost would be even higher.

Reaction of big techs at the cost of AI

Despite the alert, giants like Google, Microsoft and Meta continue to expand their datacenters and launch new models. The current logic is that AI can generate indirect value, such as data insights or improvements in customer experience, which are not captured by simple productivity metrics. But for small and medium-sized enterprises, the account may not close, increasing the risk of an investment bubble.

Frequently asked questions about the cost of AI

Will the AI replace all jobs?

Nope. According to Catanzaro, for most routine tasks, the cost of AI still outweighs the benefit, making mass replacement economically impossible in the short term.

What is the main hidden cost of AI?

The cost of continuous training and adjustment of models, which consumes expensive energy and hardware, and the need for experts to keep the system running.

Should companies invest in AI now?

Depends on the use case. For tasks of high added value and low error tolerance, AI may be worth it. For simple automation, the human cost is even more competitive.