NVIDIA to spend $3 billion on power infrastructure tied to AI data centers, report says

NVIDIA is reportedly preparing a major expansion into the electrical infrastructure that underpins modern AI data centers, with a planned $3 billion bet tied to the physical power needed to operate AI chip platforms at scale. The move, described in a report from The Information and relayed by Yahoo Finance, points to a growing focus beyond semiconductors as companies race to build the computing capacity for artificial intelligence.

According to the report, NVIDIA’s planned spending would go toward the “power behind AI,” an apparent reference to the grid and on-site electrical systems that support the high energy draw of large-scale AI deployments. The Information report was published Aug. 7, with Yahoo Finance carrying the summary in its Aug. 16 technology coverage.

For NVIDIA, the strategic logic is straightforward but increasingly complex. AI data centers require not only advanced processors and the supporting networking gear, but also power delivery and management systems that can handle rapidly scaling loads. The company’s customers, including large cloud and enterprise operators, face long lead times for power-related upgrades, which can become a bottleneck even when chips and servers are available.

NVIDIA has long positioned itself as more than a chip supplier by building broader data center platforms, including software and systems designed to integrate across training and inference. In that context, the reported emphasis on power infrastructure suggests NVIDIA wants to reduce deployment friction for AI customers and keep demand flowing as power constraints become more prominent across the industry.

The reported $3 billion figure also highlights how the economics of AI are shifting from purely computing hardware to the physical infrastructure around it. Power systems involve engineering, permitting, and construction across both utility interfaces and internal data center electrical architecture. Even when companies can procure servers quickly, electrical upgrades can still determine when equipment can be turned on and operated reliably.

NVIDIA did not provide details in the Yahoo Finance summary about the specific structure of the spending, which utilities or sites it targets, whether it involves direct capital expenditures, partnerships, or service contracts, or how the company expects to measure returns. Without additional disclosure, it remains unclear how much of this spending is new incremental investment versus commitments tied to existing customer relationships.

Still, the direction of travel is consistent with a broader industry reality: AI’s growth is increasingly constrained by power availability. If NVIDIA can help customers navigate those constraints, it may strengthen the stickiness of its platform ecosystem, because power readiness and uptime are prerequisites for running AI workloads at meaningful scale.