Why a Three-Month Data Center Delay Costs $10 Billion, and What It Means for Power Stocks

A single 1-gigawatt data center requires roughly $50 billion to construct and can yield about $40 billion in annual revenue, according to remarks made by former Nvidia CEO Jensen Huang before his resignation on November 12. Huang pointed out that a mere three-month construction setback results in $10 billion of lost revenue potential, a financial risk that causes widespread anxiety across the artificial intelligence sector.

While Nvidia has sought to alleviate bottlenecks by moving more assembly processes into the supply chain rather than relying solely on on-site workers, industry experts warn that construction hurdles are secondary to a much larger obstacle. Financial institutions and consultancies, including McKinsey & Co., Morgan Stanley, J.P. Morgan, and Goldman Sachs, have identified energy availability as the ultimate binding constraint on the ongoing AI infrastructure build-out.

To navigate these power limitations, everyday investors are looking beyond traditional tech holdings toward energy providers supplying the necessary electricity. Higher-risk opportunities include next-generation nuclear innovators such as NuScale Power and Oklo Inc., which specialize in small modular reactors designed for faster construction. Meanwhile, more risk-conscious investors are considering established players like Constellation Energy, a major nuclear producer expanding its output through power purchase agreements, and Bloom Energy, which offers fuel cell systems allowing data centers to generate independent electricity.

What to watch next: Monitor upcoming corporate power purchase agreements between major AI operators and energy providers, alongside regulatory approvals for small modular nuclear reactors.