Musk reveals timeline for SpaceX’s orbital AI data center, with first Nvidia-powered satellites launching in Q4 2027 and “significant scale” expected by 2028.

Musk reveals timeline for SpaceX’s orbital AI data center

Musk reveals timeline for SpaceX’s orbital AI data center, confirming that the first batch of Nvidia-powered AI satellites is now planned for launch in the fourth quarter of 2027, with the network expected to reach “significant scale” in 2028. The announcement, made on August 24, 2026, marks an acceleration of SpaceX’s earlier plans, which had initially targeted 2028 for the first orbital data center deployments.

The project, part of SpaceX’s broader push into space-based computing, is being developed in close partnership with Nvidia. Together, the companies have designed a space-optimised AI supercomputer system based on Nvidia’s Vera Rubin NVL72 platform, intended to run AI workloads in orbit rather than in traditional ground-based data centers.

First launch moved forward to late 2027

According to Musk’s comments reported by Bloomberg and confirmed on X:

• The first Starmind AI satellites equipped with Nvidia chips are scheduled to launch in Q4 2027.
• The orbital data center network is expected to reach “significant scale” in 2028, with a much larger constellation following the initial deployment.
• This timeline is at least one year earlier than SpaceX’s previously disclosed target of 2028 for the first orbital data centers.

SpaceX had earlier suggested in its prospectus that orbital data centers might begin launching in 2028. Musk’s latest update moves that target forward, signalling growing confidence in both the technology and the underlying launch cadence of Starship and other vehicles.

Nvidia partnership and Vera Rubin NVL72

The orbital AI system is being co-developed with Nvidia and is centred around the Vera Rubin NVL72 platform:

• Vera is described by Musk as the first CPU “built for agents,” designed to accelerate orchestration, code execution, and data processing for AI agents.
• Vera Rubin NVL72 is a space-optimised variant of Nvidia’s data center architecture, adapted for the radiation, thermal, and power constraints of orbit.
• The system will power SpaceX’s planned Starmind AI satellites, which form the backbone of the orbital data center network.

This partnership deepens an existing relationship: last month, Musk announced that SpaceX would exclusively use Nvidia chips for its future AI infrastructure, and Nvidia confirmed that SpaceX plans to leverage the Vera Rubin platform for its space-based compute.

Scale and compute targets

Musk also outlined ambitious compute goals for SpaceX’s overall AI infrastructure, both on the ground and in orbit:

• By the end of 2026, SpaceX aims to amass more than 2 gigawatts (GW) of AI compute capacity.
• By the end of 2027, that figure is expected to rise to a level “closer to 10 GW than to 5 GW.”
• The orbital data centers are a key part of this expansion, intended to complement ground-based AI clusters with additional capacity in space.

While Musk did not specify exactly how much of this capacity will be orbital versus terrestrial, the implication is that space-based compute will become a non-trivial component of SpaceX’s AI footprint by 2028.

Why put AI data centers in orbit?

SpaceX’s rationale for orbital AI data centers includes several potential advantages:

• Access to abundant solar power – Satellites can harvest continuous solar energy without night-time interruptions, potentially enabling very high-power AI workloads.
• Reduced land and cooling constraints – Space-based systems avoid many of the land-use, water, and cooling challenges facing large terrestrial data centers.
• Global low-latency coverage – A constellation of AI satellites could provide compute closer to end users in different regions, reducing latency for certain applications.
• Strategic diversification – Distributing AI infrastructure across ground and space reduces dependence on any single geography or regulatory environment.

Critics, however, point to challenges such as launch costs, radiation hardening, maintenance, and the complexity of operating large-scale AI systems in orbit. Musk’s accelerated timeline suggests SpaceX believes these hurdles can be overcome sooner than previously expected.

Implications for AI and the space industry

If successful, SpaceX’s orbital AI data centers could:

• Create a new category of space-based cloud and AI services, competing with or complementing terrestrial providers.
• Enable novel applications in real-time Earth observation, communications, and autonomous systems that benefit from on-orbit processing.
• Intensify competition among tech and aerospace companies to control both AI compute and launch capacity.

For Nvidia, the partnership reinforces its position as a key supplier not just for ground-based AI clusters, but also for next-generation space infrastructure.

Summary: Musk reveals timeline for SpaceX’s orbital AI data center, with first Nvidia-powered Starmind AI satellites planned for Q4 2027 and “significant scale” expected by 2028. The project, built around Nvidia’s Vera Rubin NVL72 platform, is part of a broader push to reach up to 10 GW of AI compute by the end of 2027 across ground and space.