The next AI pivot will be toward efficiency and lowering costs, ex-Facebook privacy chief says

Chris Kelly, the former Chief Privacy Officer at Facebook, highlighted a pivotal shift in the artificial intelligence sector during a recent interview on CNBC's "Squawk Box." He emphasized that the upcoming evolution within AI will center around enhancing efficiency and reducing operational costs. As the demand for AI infrastructure escalates, Kelly pointed out the necessity for the industry to optimize its energy-intensive developments. "Our brains operate on just 20 watts. There's no reason we should require gigawatt power facilities to make decisions," he stated. He believes that the quest for efficiency will be crucial for major players in the AI landscape. Kelly, who also served as Facebook's general counsel, indicated that companies that succeed in minimizing data center expenses will likely emerge as leaders in the AI arena. The data center market has seen an impressive $61 billion in infrastructure transactions as hyperscalers engage in a global building spree, according to S&P Global. OpenAI has notably committed over $1.4 trillion to AI initiatives in the coming years, forming substantial partnerships with GPU powerhouse Nvidia and infrastructure leaders like Oracle and Coreweave. However, this surge in data center construction has raised alarms about the sourcing of the necessary power, especially given the current strain on the electric grid. In September, Nvidia and OpenAI unveiled plans for a project requiring at least 10 gigawatts of data center capacity, which is equivalent to the annual energy consumption of about 8 million U.S. households. This figure mirrors New York City's peak summer energy demand projected for 2024, as reported by the New York Independent System Operator. Cost concerns gained further traction when DeepSeek introduced a free, open-source large language model in December, claiming development costs of under $6 million—significantly lower than its U.S. counterparts.