Data Centers Are Going to Space. Here’s What You Need to Know.

By Zack Huhn, Co-Founder & National Director, Enterprise Technology Association
March 30, 2026

Starcloud just raised $170 million in a Series A round that values the company at $1.1 billion— making it the fastest startup in Y Combinator’s 21-year history to reach unicorn status. Their bet: building AI-ready data centers not on Earth, but in low Earth orbit.

The Big Picture

AI’s insatiable appetite for compute is slamming into the physical limits of terrestrial infrastructure. New data centers can take five years to permit and build. Communities are pushing back on power consumption and water usage. Grid capacity is strained. Starcloud’s thesis is that orbital infrastructure sidesteps all of this: in space, solar energy is essentially unlimited, cooling happens passively by radiating heat into deep space, and there are no permitting battles or land acquisition wars to fight.

The round was led by Benchmark and EQT Ventures—the latter managing $100 billion in assets and operating more than 70 terrestrial data centers. Additional investors include Macquarie Capital, NFX, and Y Combinator, alongside angel backers like former Boeing CEO Dennis Muilenburg and retired U.S. Air Force General Stephen Wilson. It’s a signal that serious infrastructure capital is treating orbital compute as a legitimate asset class.

Where They Are Today

Starcloud launched its first satellite carrying an Nvidia H100 GPU in November 2025—making it the most powerful commercial GPU ever deployed in orbit. The company used it to train an AI model in space (a first) and run inference workloads, including a version of Gemini. That was built in 21 months with just $3 million in pre-seed funding.

Later this year, Starcloud-2 launches with 100x the power generation of its predecessor. It will feature Nvidia’s Blackwell B200 chip, an AWS server blade, and notably, a bitcoin mining computer. It will be their first satellite running commercial customer workloads, with early customer Crusoe and partnerships with AWS, Google Cloud, and Nvidia.

What’s Next—and What’s at Stake

The real play is Starcloud-3: a three-ton, 200-kilowatt spacecraft designed to fit SpaceX’s Starship deployment system. The company projects it could deliver power costs around $0.05 per kilowatt-hour—competitive with terrestrial data centers—but only if Starship achieves commercial launch costs near $500 per kilogram. CEO Philip Johnston expects that access to open up in 2028–2029.

They’re not alone. SpaceX has filed to deploy up to one million data center satellites. Blue Origin has expressed similar ambitions. Google’s Project Suncatcher and startups like Aetherflux (raising at a $2 billion valuation) and Aethero are all building in this space. The competitive landscape is heating up fast.

WHAT YOU NEED TO KNOW

→ The AI infrastructure bottleneck is real—and it’s driving serious capital toward solutions that would have seemed like science fiction two years ago.

→ Orbital compute doesn’t replace terrestrial data centers. Even Starcloud forecasts space-based
compute will represent less than 1% of new capacity in the near term. But a decade out, it could scale dramatically.

→ This is an energy play as much as it is a space play. The convergence of AI demand, grid limitations, and community opposition to new facilities is creating structural openings for alternative infrastructure models.

→ Workforce and ecosystem implications are significant. Space-based data centers require talent across aerospace engineering, AI/ML operations, power systems, satellite communications, and manufacturing—a new category of jobs at the intersection of industries.

→ For economic development leaders, this represents an emerging sector to watch. The regions that build training pipelines and attract orbital infrastructure companies early will have a meaningful advantage.

The ETA Perspective

At ETA, we track inflection points—moments where technology shifts create new economic opportunities and workforce demands. Starcloud’s raise is one of those moments. It’s a reminder that the infrastructure layer of AI is evolving in ways that will reshape job markets, regional economies, and the technology supply chain in ways most leaders aren’t planning for yet.

This is exactly the kind of development we dig into at our events—where technology leaders, policymakers, and economic development professionals come together to make sense of what’s next and build strategies around it.

Join the Conversation

We’re breaking down developments like this at upcoming ETA events across the country,
from AI Week regional conferences to the US AI Congress in Washington, D.C.
Come be part of the conversation shaping America’s technology future.