At the G20科技部长会议上，科技界的两位重量级人物罕见地同台发声，共同向全球政策制定者传递了一个明确的信号：制约AI发展的核心瓶颈正在从算力转向电力和人力。

Elon Musk offered a stark timeline, suggesting electricity shortages could become apparent as early as next year, with a projected shortfall of at least 15 gigawatts for AI chips by 2027. Meanwhile, Mark Zuckerberg acknowledged that Meta is already grappling with a shortage of skilled technical workers needed for its data center projects.

Analysts suggest this warning carries direct and profound market implications, signaling that competition in AI infrastructure has expanded beyond chips and data centers to encompass power generation and energy solutions. Musk revealed that companies like Google and Anthropic are now leasing computing capacity from SpaceX, largely because SpaceX has built its own power plants to secure electricity supply, making energy self-sufficiency a new frontier in AI competition.

The meeting, held on September 1st in Chapel Hill, North Carolina, focused on the opportunities and disruptions presented by AI. Both Musk and Zuckerberg called for accelerated data center construction, while Demis Hassabis, CEO of Google's DeepMind, joined via video to describe AI's potential impact as "10 times the industrial revolution."

Power Crisis Intensifies as AI Chip Growth Outpaces Grid Expansion

Musk delivered his most specific warning yet on AI's energy constraints during the conference. Citing analyst consensus estimates, he projected that AI chips would face a power deficit of at least 15 gigawatts by 2027. The fundamental imbalance lies in the growth rates: AI chip capacity expands roughly 40% to 50% annually, while available power supply outside China grows only 10% to 20% per year.

"Clearly, something growing faster will eventually overwhelm something growing slower," Musk said, articulating this structural conflict. He also noted that Google, Anthropic, and other companies are already leasing computational power from SpaceX, which has overcome power supply bottlenecks by constructing its own electricity generation facilities. He emphasized that his companies build power infrastructure alongside data centers as a matter of standard practice.

Labor Shortage Emerges as Critical Hurdle in Data Center Development

Beyond electricity, the scarcity of skilled labor is proving equally challenging for tech giants. Zuckerberg, speaking via video link, highlighted the massive demand for skilled technical positions in AI data center construction, potentially requiring "hundreds of thousands or even millions" of workers. He stated plainly that Meta is already struggling to find enough qualified personnel to build its facilities, a constraint that is happening now rather than in some distant future.

Together, Musk's power warning and Zuckerberg's labor concerns paint a clear picture of AI infrastructure expansion facing dual bottlenecks: the hardware-level electricity supply and the workforce-level talent pool. Both point to the same conclusion, that AI's pace of development is outstripping the capacity of existing infrastructure systems to support it.

Regulatory Divide Draws Criticism as Musk Targets European Model

Infrastructure issues were not the only point of contention, as regulatory differences also became a flashpoint at the gathering. Musk criticized Europe's heavy-handed approach to technology regulation, arguing that innovation thrives in less restrictive environments. He advocated for new technologies to be "legal by default" rather than "illegal by default," suggesting that while European regulations won't halt technological progress entirely, they will significantly slow its momentum.

This position aligns with the broader policy tone emerging from Washington, following former President Trump's previous criticism of AI data center opponents as "backward and poor," indicating a political inclination toward accelerating AI infrastructure development.However, the tech industry's aggressive push faces headwinds at home. Polls show a majority of Americans oppose building more data centers in their communities, with concerns centering on rising electricity costs and noise pollution. This creates a policy paradox relative to the power crisis Musk warns about: solving electricity shortages requires large-scale construction, yet that very construction is facing public resistance.

Economic Vision Envisions AI Adding Over $20 Trillion Annually to Global Economy

Despite these formidable challenges, tech leaders remain highly bullish on AI's long-term economic value. Musk estimated that AI could expand the global economy by 20% to 30%, equivalent to an additional $20 trillion to $30 trillion per year. He characterized this as a "rough estimate" but one he would be "willing to bet heavily on."

Demis Hassabis offered an even broader perspective, suggesting that achieving human-level AI could produce impacts ten times greater than the industrial revolution, while emphasizing that nations must pursue this development responsibly.