AI is being touted as a lot of things. A productivity tool to get the most out of your work, a bit of software to help you draft an email, and, perhaps most worrying, a way to do away with those pesky wage costs. However, it seems like, to those near the top of the chain, AI might already be starting to become more costly than its human counterpart.
As reported by Axios, Bryan Catanzaro, the vice president of applied deep learning at Nvidia said: “For my team, the cost of compute is far beyond the costs of the employees.”
A point worth mentioning here is that this quote does not specify whether that is ‘more in total’ or ‘more per an average worker’s work’. My guess is on the former, given how much Nvidia and similar companies are pumping into AI development.
Still, it’s a sign of how much commitment is currently being invested in the tech. Nvidia owes a lot to AI, with it being responsible for making it the world’s first $5 trillion company. OpenAI, the creator of ChatGPT, picked up $110 billion in funding just a few months ago, and the likes of Meta and Amazon are also further committing to AI.
One of the central concerns around AI right now, other than plagiarism, AI psychosis, and the environment, is the effect it will have on the general working public. Microsoft’s AI CEO reckons lawyers, accountants, project managers, and marketing people will all be replaceable by AI in the next 12 to 18 months (though curiously, not AI CEOs).
AI can cost more than human workers now https://t.co/qJbEBTa4OlApril 26, 2026
Nvidia’s own Jensen Huang said last year that, even if AI replaces some workers, there will be roles out there for them. He said, “If you’re an electrician, if you’re a plumber, if you’re a carpenter, we’re going to need hundreds of thousands of them. To build all of these factories.”
Not everyone seems to agree with this perspective, though. A former Google executive believes AI will lead to a “short-term dystopia” as the idea that it will create new jobs for the ones it’s replacing is “100% crap”.
Still, what remains a lingering question over AI development is how much those workers will eventually cost. If a handful of companies control the majority of AI models, they can afford to set the rates of tokens. We saw this recently, with the Uber CTO getting through the entire 2026 AI budget due to token costs.
One has to wonder when all that AI development will be worth it, and how much it will cost after all of that effort.