Thursday, August 6, 2026
DarkSubscribe
AI Infrastructure · News & Analysis
HomeData CentersReport
Data Centers · Report

Elon Musk commits SpaceX exclusively to Nvidia GPUs, planning 10GW of AI compute capacity by 2027, in major supply allocation and capacity expansion announcement.

Exclusive supplier commitment from major infrastructure player signals Nvidia's continued dominance in chip allocation and validates aggressive data center expansion timelines.
Trade pressSlicast · August 5, 2026 · US · Source: Google News
importance 93

SpaceX has committed to using NVIDIA GPUs exclusively to power its AI operations. In its latest earnings call, CEO Elon Musk confirmed that the company selected NVIDIA's Vera Rubin platform for AI workloads across its global data centers, including deployments in space. According to Musk, the choice was driven by the fact that "they're the best."

SpaceX will deploy the NVIDIA Vera Rubin NVL72 rack, codenamed Kyber, to power its AI infrastructure. The company plans to install 2GW of compute capacity by the end of this year, with a total deployment target of 10GW by the end of 2027. This represents a significant win for NVIDIA as it continues to secure major customers among cloud providers and large-scale AI infrastructure operators.

A notable aspect of SpaceX's strategy involves in-orbit deployments. NVIDIA has already introduced its space-certified Space-1 Vera Rubin module, designed for orbital operations. This module integrates four Rubin GPUs, two Vera CPUs, and substantial memory capacity to support AI workloads beyond Earth's atmosphere.

SpaceX's orbital ambitions build on its previous partnership with Anthropic to develop a multi-Gigawatt "Orbital" AI data center. The company sees orbital compute installations as a solution to critical terrestrial constraints—specifically power, land, and cooling limitations that restrict ground-based data centers. By locating infrastructure in space, these bottlenecks can theoretically be eliminated through unlimited power and cooling resources.

However, the orbital AI datacenter model presents its own challenges. With abundant power and cooling in space, GPU capacity itself becomes the constraint. Additionally, the logistical burden of transporting hundreds of thousands of GPUs to orbit represents a formidable hurdle. Despite these obstacles, the pace of AI advancement has made what seemed like a distant vision increasingly feasible in the near term.

Read the original
Elon Musk commits SpaceX exclusively to Nvidia… · Slicast