Back
SpaceX to use only NVIDIA technology, targeting over 2 gigawatts by year-end

SpaceX to use only NVIDIA technology, targeting over 2 gigawatts by year-end

Elon Musk said SpaceX will use Nvidia GPUs exclusively, while JPMorgan expects the company to receive a very significant share of Nvidia shipments in 2027 as AI capacity expands.

Fact Check
Elon Musk's own X post directly confirms SpaceX committed to using Nvidia GPUs exclusively. The ~220,000 NVIDIA GPU figure is corroborated by Tom's Hardware (222,000+ GPUs in Colossus), and the ~2 gigawatt compute target for late 2026 is corroborated by CNBC reporting on SpaceX/Anthropic compute deals. The aggregator's reference to an '$SPCX earnings call' is questionable framing, but the substantive claims about NVIDIA exclusivity, GPU count, and gigawatt-scale compute are well supported.
Summary

SpaceX is committing to Nvidia GPUs exclusively for its AI infrastructure, with Elon Musk saying the company chose Nvidia because its chips are "the best" and JPMorgan describing a deepening partnership centered on Nvidia’s Vera Rubin platform. The bank said SpaceX has decided to deploy Nvidia’s Vera Rubin GPUs across its data centers and expects to receive a "very significant percentage" of Nvidia’s GPU shipments in 2027. JPMorgan said the move supports SpaceX’s plan to scale AI compute capacity from more than 2 gigawatts by the end of 2026 to between 5 and 10 gigawatts by the end of 2027, and added that SpaceX also plans to use Vera Rubin architecture in its Starmind AI satellites. The exclusive strategy marks a shift from Musk’s earlier indication that SpaceX and Tesla would use both AMD and Nvidia chips, and it reinforces Nvidia’s standing in AI accelerators even as rivals push to win share. CNBC’s Jim Cramer, a long-time Nvidia bull, highlighted Musk’s post as further support for his thesis on the chipmaker.

Terms & Concepts
  • GPU: A graphics processor widely used to train and run artificial intelligence models.
  • Vera Rubin platform: Nvidia’s AI computing platform built around its Vera Rubin architecture and related GPU systems.
  • AI accelerators: Specialized chips designed to speed up artificial intelligence workloads such as training and inference.