SpaceX expects its orbital data centers to reach what it calls "significant scale" in 2028, with Nvidia now on board as a partner in the effort, Light Reading reports.

The 2028 target gives the space-infrastructure and data-center industries a concrete timeline for what has so far remained a speculative concept: racks of compute hardware operating in orbit rather than on terrestrial power grids and fiber networks.

Why does Nvidia's involvement matter?

Nvidia's participation signals that mainstream accelerated-computing hardware — the GPUs underpinning current AI buildouts — now sits at the center of the orbital data center concept rather than specialized or experimental processors.

For SpaceX, the partnership aligns with its existing launch capacity and its Starlink broadband constellation. The company controls both the ride to orbit and a communications backbone capable of linking space-based compute to ground users.

For Nvidia, the arrangement extends its silicon into a new deployment environment as demand for AI compute continues to strain terrestrial power and cooling constraints.

What does 'significant scale' mean for the timeline?

SpaceX has not yet attached specific capacity figures, rack counts, or launch cadence to the 2028 milestone. The company's use of the phrase "significant scale" indicates deployment beyond demonstrator units — an operational fleet rather than a proof of concept.

The three-to-four-year runway implies:

• Hardware qualification for radiation, thermal cycling, and vibration during launch
• Power systems sized for GPU-class compute loads in orbit
• Orbital servicing, replacement, and decommissioning procedures
• Integration with Starlink or other links for data transfer to customers

What are the open questions?

The announcement leaves key engineering and commercial details unanswered. These include thermal rejection in vacuum — orbital hardware cannot rely on air cooling — the cost per deployed compute unit versus terrestrial equivalents, and the regulatory treatment of large commercial compute platforms in orbit.

Market observers will also watch how the 2028 date holds against launch-schedule realities, and whether capacity in orbit competes with, or merely supplements, the wave of terrestrial AI data centers now under construction.

The 2028 milestone now serves as the industry's reference point for judging whether orbital compute moves from concept to infrastructure.