Nvidia reportedly trims a proposed $250 billion OpenAI data-center commitment, according to the Wall Street Journal

Nvidia has reportedly scaled back a proposed $250 billion guarantee tied to an OpenAI data-center project, the Wall Street Journal reported, citing people familiar with the matter. The update, published via a Yahoo Finance report on Aug. 14, indicates Nvidia’s plan would be smaller than originally discussed, though the revised size and timing were not spelled out in the available coverage.

A “guarantee” in this context refers to an arrangement meant to underwrite demand or financing risk associated with building large-scale data-center capacity. For Nvidia, these kinds of deals can help secure long-term sales of data-center systems, including GPUs (graphics processing units) and related networking and infrastructure used to run AI workloads.

The report characterizes the development as a downsizing of plans rather than a cancellation, but it does not provide granular details about what has changed. That includes what the new commitment level would be, which contracts or counterparties would be affected, or how the revised terms might alter procurement schedules for hardware.

While Nvidia is best known for its semiconductor products, the economic logic of large AI data-center commitments depends on a broader ecosystem: chips must be paired with networking gear, power delivery, and data-center design that can support high-throughput training and inference at scale. If the scope of a guarantee changes, it can ripple through pricing, delivery plans, and the pace at which customers expand capacity.

The company did not comment in the Yahoo-linked report beyond relaying the existence of a Wall Street Journal account. With no official filing or company announcement included in the coverage, investors and customers are left to interpret the shift largely through secondhand descriptions and timing cues.

For the AI infrastructure market, the reported adjustment highlights how fast-changing demand assumptions can affect long-horizon commitments. Over the past year, the pace of AI buildouts has been influenced by customer capex plans, component availability, power constraints, and performance-per-dollar improvements across hardware generations, all of which can lead parties to recalibrate deal structures.

What remains unclear is whether the revised guarantee would still cover the same categories of spending and delivery milestones, and whether any alternate terms were added, such as different pricing, demand tiers, or revised take-or-pay language. The coverage also does not address whether OpenAI, Nvidia, and data-center partners have renegotiated earlier schedules.

Going forward, attention will likely shift to whether Nvidia provides any clarification in earnings materials or official communications, and whether other suppliers or customers disclose corresponding changes to capacity plans tied to major AI infrastructure builds.