AI's $61 Billion Power Play Is Over. The Next Trillion-Dollar Race Is Efficiency.

The artificial intelligence boom is entering a new, critical phase where brute force gives way to strategic efficiency, according to former privacy chief . Speaking to CNBC, he argued that the next big winners in AI won't just build bigger models, but will be those who can drastically cut the immense costs and power consumption of training them, a timely warning as the industry's infrastructure spending spree tops this year.

Hyperscalers have poured money into a global data center buildout, with infrastructure dealmaking accumulating over in alone, according to S&P Global. has committed over to AI initiatives in the coming years, including massive partnerships with GPU leader and infrastructure giants and .

But this frenzy is straining an already fragile electric grid. A joint project by and , announced in September, includes at least of data centers. That's roughly equivalent to the annual power consumption of U.S. households, or the entire peak summer demand of New York City in .

believes this challenge is the industry's next great opportunity. "We run our brains on . We don't need gigawatt power centers to reason," he said. "I think that finding efficiency is going to be one of the key things that the big AI players look to."

The pressure to reduce costs is already mounting. In a sign of what's to come, Chinese firm claimed in December to have launched a free, open-source large language model for under — a fraction of the cost for its U.S. competitors.