- Goldman Sachs Institute's Jared Cohen says the U.S. will need to outsource its data center capacity, and choose its global partners wisely, as the AI race heats up.
- Big Tech companies are set to spend roughly $600 billion on AI data center infrastructure and development.
- "Data may be the new oil, and it's ultimately nations, not nature, that's going to determine the future of AI infrastructure built," says Cohen.
Data centers are a key piece of artificial intelligence infrastructure. Where they pop up around the globe could have lasting geopolitical impacts for the U.S., says Goldman Sachs' president of global affairs.
Jared Cohen, the former CEO of Google's Jigsaw and now co-head of the Goldman Sachs Global Institute, compares the momentum in AI to the next Industrial Revolution in a new op-ed this week in Foreign Policy. In this technology wave, the opportunity exists for "data center diplomacy," he wrote.
"This is a technology that is on par with the creation of the internet, except it's happened much more abruptly," Cohen told CNBC in an interview. "Data may be the new oil, and it's ultimately nations, not nature, that's going to determine the future of AI infrastructure built."
Get top local stories in Connecticut delivered to you every morning. Sign up for NBC Connecticut's News Headlines newsletter.
AI relies on massive amounts of data for training and gigantic data centers. Mega-cap tech companies like Google, Amazon, Microsoft and Meta are spending heavily to build the infrastructure, and plan to shell out roughly $600 billion on those efforts in the next three years, according to Goldman.
The key geopolitical factor is China. Despite a slowing economy, Beijing has been investing in AI data centers, and launched a $6.1 billion national initiative called "Eastern Data, Western Computing." The U.S. has its own slate of initiatives, including a task force on AI infrastructure.
Countries that have money to spend face a "binary" choice for investing in AI: The U.S., or China. Cohen says that while the U.S. is still ahead in AI, data center buildout become a "bottleneck," and the "U.S. is going to have to have some sort of overflow option to meet demand."
Money Report
The U.S. has already partnered with countries including Canada, Australia and France. But the other option is what Cohen calls "geopolitical swing states," or countries with a "disproportionate amount of capital, a willingness to deploy it around the world," and a higher propensity to swing towards China. He recognizes the Middle East as a key partner.
An influx in capital to oil-rich nations in the Gulf and a need to diversify their economies, has made them a match for capital intensive AI businesses. Roughly $11.3 trillion is managed by sovereign funds, with five of the ten most active based in the Middle Eastern Gulf states. They've become major backers of some of Silicon Valley's artificial intelligence darlings, from OpenAI to Anthropic.
Cohen says countries like United Arab Emirates, Qatar, and Saudi Arabia may be in the strongest position to build data center capacity, and "do this fast."
"The Arab Gulf countries of the Middle East present many promising opportunities for AI data centers," Cohen wrote. "With young, ambitious leaders, these countries aim not just to export oil, but also AI. As one prominent official from the United Arab Emirates recently emphasized, 'We missed the first industrial revolution, but we are not missing the AI revolution.'"
