Artificial intelligence may be transforming workplaces around the world, but a senior Nvidia executive has argued that, for many businesses, the technology still costs more to operate than employing people.
Speaking about the economics of AI deployment, Bryan Catanzaro, Nvidia's vice president of applied deep learning, said the expense of the computing power required to run advanced AI systems currently outweighs the cost of the employees using them. His comments challenge the widespread assumption that AI is already a cheaper alternative to human labour.
Catanzaro said the high cost of AI stems largely from the enormous computing resources needed to train and operate large language models, as well as the energy and infrastructure required to support them. While AI has become increasingly capable, the financial burden of delivering those services remains substantial.
The remarks come as businesses across multiple industries continue investing heavily in AI despite uncertainty over when those investments will generate meaningful returns. Major technology companies are expected to spend hundreds of billions of dollars this year on AI infrastructure, including data centres, networking equipment and specialised chips needed to power next-generation applications.
According to estimates cited by McKinsey, global spending on AI could reach between $5.2 trillion and $7.9 trillion by the end of the decade, depending on the pace of adoption. Around $1.6 trillion of that total is expected to be invested in data centres, while a further $3.3 trillion could be directed towards IT hardware and related equipment.
Despite the significant expenditure, many organisations remain optimistic that AI will eventually lower operating costs as models become more efficient and computing prices fall. However, Catanzaro suggested the industry has not yet reached that point, with current-generation systems requiring vast amounts of processing power that make them expensive to use at scale.
The comments also highlight a broader debate over AI's impact on employment. Although concerns persist that automation could replace large numbers of workers, some experts argue that the economics do not yet support wholesale substitution in many sectors.
Research from the Massachusetts Institute of Technology has suggested AI automation is currently cost-effective in only a minority of jobs, with human workers remaining the cheaper option in most roles.
At the same time, demand for AI infrastructure continues to grow rapidly, driving up the cost of key components such as memory chips, cooling systems and electricity. Those rising costs have prompted warnings that the AI boom could place additional pressure on supply chains and contribute to higher technology prices across the wider economy.
Nvidia has emerged as one of the biggest beneficiaries of the global AI race, supplying the graphics processing units (GPUs) that underpin many of today's most advanced AI models. Demand for its hardware has surged as technology companies race to expand their AI capabilities, although the cost of building and operating the infrastructure needed to support those systems has also increased significantly.
While AI continues to improve at a rapid pace, Catanzaro's comments suggest businesses weighing up automation against human labour may need to focus on more than capability alone. For now, the financial equation remains complex, with the cost of computing power still representing one of the biggest barriers to achieving widespread economic gains from artificial intelligence.
