Musk Warns of AI Power Shortfall as Zuckerberg Cites Data-Center Labor Gap

Elon Musk warned G20 technology ministers that AI could face a significant power shortfall next year. Mark Zuckerberg, speaking at the same Chapel Hill meeting, said Meta cannot find enough skilled-trades workers to construct the data centers it wants to build. Together, the remarks put two physical constraints—electricity supply and construction labor—at the center of an event devoted to AI’s economic promise.

The G20 Innovation Ministerial runs September 1–2 at the Carolina Inn in Chapel Hill. The U.S. Department of Commerce and White House Office of Science and Technology Policy are co-hosting the gathering, which was set up to bring G20 partners together around technology-driven growth and policy principles for AI and other emerging technologies.

The buildout problem arrives before the systems do

Musk’s forecast was a near-term warning, not a projection for a distant phase of AI adoption. He said a power crunch could give China an advantage, and said Tesla was building generation capacity for its own data centers. The remarks did not establish how large a shortfall he expects or where it would occur.

What the consensus is there will be a significant power shortfall next year, so not in the distant future.

Elon Musk, speaking at the G20 Innovation Ministerial

Zuckerberg described a different bottleneck. He said the volume of demand has left Meta unable to hire the electricians, carpenters and other skilled workers it needs for its planned facilities. He estimated that the wider data-center buildout could require hundreds of thousands, and perhaps millions, of skilled-trades jobs.

A proposed answer puts more work on builders

Zuckerberg said Meta created America’s Workforce Academy to train people for trade roles. On power, White House science adviser Michael Kratsios argued that data-center developers can benefit host communities if they build, bring or buy their own electricity rather than simply compete with residents for existing supply.

The local bargain under discussion

• Kratsios said communities ultimately decide whether to accept data-center projects.
• Critics have raised concerns about data centers’ environmental effects and potential effects on utility costs.
• One account reported that a majority of Americans oppose building AI data centers, citing electricity bills and facility noise.

Growth claims meet practical dependencies

The infrastructure warnings sat beside expansive claims for AI’s economic effects. Musk estimated that AI could increase the global economy by 20% to 30%, or roughly $20 trillion to $30 trillion a year. Zuckerberg said lower startup costs could yield more startups and thriving small businesses, though those outcomes are forecasts rather than commitments from the ministerial.

The meeting also provided a policy setting for those arguments. Its organizers are seeking principles that promote innovation, investment in manufacturing and supply chains, and public support for emerging technologies. Musk separately called for innovation to be relatively free of regulation and criticized European technology rules, while DeepMind chief Demis Hassabis urged responsible development.

The next day tests whether the discussion turns concrete

Nvidia chief executive Jensen Huang and OpenAI chief executive Sam Altman were scheduled to participate on the ministerial’s second day. The immediate question is whether the gathering produces a shared approach to the infrastructure requirements its speakers highlighted, or remains a forum for competing views on how rapidly AI development should proceed and how much government should shape it.