Doubts mount over Musk's space data center pitch as rivals circle

Elon Musk's vision for data centers orbiting Earth is facing skepticism from some of the industry's biggest players. At a shareholder meeting, SoftBank founder and CEO Masayoshi Son cast doubt on the economic viability of space-based infrastructure, noting that the immediate race for AI dominance will be won or lost in the next few years—not a decade down the line.

The skepticism spreads#

Son's remarks have sparked broader conversation about whether orbital data centers represent a genuine solution to the world's pressing compute shortage, or simply another far-fetched moonshot. The reality is that compute constraints are acute right now: data center buildout on Earth faces numerous hurdles ranging from environmental concerns to local opposition, making space seem like an attractive alternative. But skeptics argue the engineering and financial barriers to launching, maintaining, and operating satellites as computational infrastructure are formidable—and their payoff lies too far in the future.

OpenAI's Sam Altman has expressed similar reservations about the timeline. Meanwhile, SoftBank itself has committed massive resources to traditional data center projects on terra firma, putting its money where its mouth is on near-term solutions.

Following the incentives#

What complicates the discourse is that nearly every executive weighing in on AI's future—including those raising questions—stands to benefit from their own predictions. Musk's vision would directly boost SpaceX's launch business if orbital infrastructure requires a constant stream of replacement satellites. Son is betting billions on Earth-based data centers. Each player has skin in the game, making purely objective analysis difficult to find.

The broader trend#

Compute constraints have sparked a feeding frenzy of sorts. Groq secured $650 million in fresh funding; SpaceX has signed multiple rental agreements to monetize its spare computing capacity; even Allbirds pivoted away from footwear into AI infrastructure after emerging from bankruptcy. Those with idle compute are racing to become cloud providers. The open question isn't whether space-based data centers might work someday—it's whether they'll ever matter more than solutions that can actually ship now.