Oracle and NVIDIA expand Vera Rubin AI server deployment to multi-gigawatt scale

Reuters explicitly reports that Oracle and NVIDIA will expand Oracle’s cloud deployment of NVIDIA Vera Rubin AI servers to a multi-gigawatt scale, making this a new, forward-looking commercial and infrastructure commitment. The arrangement could affect future Oracle cloud capacity and NVIDIA AI-system demand; however, the deal value, deployment schedule, financing arrangements, and supplier allocation are undisclosed. Oracle’s planned use of its own Acceleron networking chips in some NVIDIA-based systems is included as an implementation detail of the same arrangement, rather than as a separate development. The commercial_agreement type is more specific than operations_infrastructure because the source describes an expanded chip-supply deal between the two companies. Importance is high but below the maximum because the scale is material while financial terms remain undisclosed. Surprise is moderate-to-high: Oracle’s Vera Rubin deployment was already publicly established, but the multi-gigawatt expansion is a significant new scale commitment. Confidence is high based on the direct Reuters report. The timestamp uses October 27, 2026, the announcement date indicated by the source context; the exact announcement time is not stated, so midnight UTC is used as a date-level estimate. ORCL and NVDA are the only approved stakeholders because they are the directly named parties to the expansion; other companies mentioned in the article relate to separate chip offerings or indirect supply-chain exposure.