SpaceX (SPCX) Stock: Morgan Stanley Says Power Scaling Is Its Biggest AI Edge

TLDR

• Morgan Stanley analyst Stephen Byrd says SpaceX has a key competitive advantage in scaling power solutions faster and cheaper than rivals.
• SpaceX ended Q2 2026 with 1.4 gigawatts of AI computing capacity, targeting 10 GW by end of 2027.
• Elon Musk says he is “highly confident” SpaceX will begin launching NVIDIA’s AI computers into orbit as early as 2027.
• NVIDIA has been named the exclusive chip supplier for the orbital computing plan, and already holds a stake in SpaceX.
• Analysts do not expect meaningful revenue from orbital AI before around 2029, with 2027-2028 milestones seen as feasibility tests.

---

SpaceX (SPCX) stock slipped 0.6% on Monday, closing at $151.85, missing a broader tech rally that saw the Nasdaq Composite rise 2.3%. The stock was up 0.5% in premarket trading Tuesday.

Space Exploration Technologies Corp., SPCX

Despite the short-term dip, the bigger story around SpaceX right now is its AI ambitions, both on the ground and in orbit.

Elon Musk has said he is “highly confident” SpaceX will begin launching NVIDIA’s most advanced AI computers into orbit as early as 2027. The plan involves adapting Starlink satellites with NVIDIA chips and solar panels, with larger deployments targeted for 2028 and an eventual constellation of up to one million compute satellites floated as a long-term goal.

NVIDIA has been named the exclusive chip supplier for the project. The company already holds a stake in SpaceX, meaning its interest goes well beyond a typical supplier relationship.

For NVIDIA, any large-scale orbital buildout would represent a new source of chip demand stacked on top of existing earthbound data center appetite.

SpaceX’s Power Edge on the Ground

Before any of that happens, SpaceX has a growing terrestrial AI business to run, and power is the central challenge.

Morgan Stanley analyst Stephen Byrd projects AI power demand will grow to nearly 100 gigawatts by 2028. That is roughly 8% of total U.S. electricity generating capacity, and utilities simply cannot scale that fast.

SpaceX, Byrd argues, has a leg up. “SpaceX has proven capabilities in scaling time-to-power solutions faster and cheaper than its competitors,” he wrote. “We view this as a key competitive advantage given the magnitude of our projected US power shortfall.”

SpaceX ended Q2 2026 with 1.4 GW of AI computing capacity. It is targeting 10 GW by end of 2027. Byrd sees terrestrial compute peaking at around 15 GW in 2031.

The Orbital Plan: Ambitious, But Not Imminent

The space-based data center concept has real logic behind it. Orbital locations offer solar power, near-unlimited space, and no planning objections from local communities. The vacuum of space also helps with cooling, in theory.

But cooling remains one of the biggest technical hurdles. Without air to carry heat away, a single orbital data center could require radiators covering millions of square feet. Add radiation shielding, data transfer logistics, the inability to send engineers for repairs, and regulatory unknowns, and the challenge is formidable.

Most analysts do not expect meaningful orbital AI revenue before 2029. The 2027 launch target is viewed as a feasibility test, not a commercial milestone.

SpaceX’s Starship rocket is central to making orbital data centers cost-competitive. Starship’s 14th test flight is currently scheduled for September 28.

SpaceX stock has risen roughly 45% since the start of August. It IPO’d in mid-June 2026.