Orbital secures $5M for AI data centres in space

AI’s appetite for power and physical space is outstripping what Earth can provide. One Los Angeles startup is proposing a radical fix: take the whole operation off the planet.

Orbital, a space-infrastructure company focused on building AI data centres in low Earth orbit, has secured a $5 million oversubscribed pre-seed round led by a16z speedrun, with participation from a broad group of venture investors. The capital will fund its first in-orbit technology demonstration, Pathfinder, a hosted GPU payload scheduled to launch on a SpaceX Falcon 9 rideshare in 2027. It also supports early development of Orbital-1, the company’s first purpose-built compute satellite. Founder and CEO Euwyn Poon previously co-founded the e-scooter company Spin.

The business case is anchored in a genuine and escalating crisis. The International Energy Agency projects that global data-centre electricity consumption will more than double by 2030, reaching roughly 945 terawatt-hours , about the same as Japan’s entire annual usage. On Earth, strained power grids, cooling demands, land scarcity, and permitting delays have all become choke points for new construction.

Orbital’s solution is to relocate the hardware where those constraints disappear. In the right orbit, solar power is continuous, and waste heat radiates directly into the vacuum of space, eliminating the need for water and fans. “The sun is the most abundant and accessible source of energy in the universe, yet we’ve barely begun to tap into it,” Poon said. “We’re building AI data centres in orbit, where solar power is continuous and heat dissipates into the void of space. Advances in launch infrastructure are making this an imminent reality, not science fiction.”

The company’s architectural approach is distinctive. Instead of assembling one massive structure in space, Orbital plans to distribute compute across many small, independently deployable satellites , a constellation it can scale one satellite at a time. The hardware is being designed around Nvidia’s announced Space-1 Vera Rubin-class GPUs, with a focus on AI inference, the fastest-growing segment of compute demand. Each production satellite is engineered to deliver 100 kilowatts of compute, with a long-term ambition of more than 100,000 satellites providing over 10 gigawatts of capacity, supported by a Los Angeles assembly plant the company calls Factory-1.

That vision enters a crowded and skeptical market. Starcloud has already raised $170 million at a $1.1 billion valuation and run a language model in orbit. SpaceX has filed plans to launch up to a million data-centre satellites, and Google is paying SpaceX for orbital compute. Against those balance sheets, $5 million is a modest sum, and Orbital’s first hardware won’t fly until 2027.

The deeper doubts are physical, not just financial. Even SpaceX, in its pre-IPO filing, warned that orbital AI data centres depend on “unproven technologies” and may never achieve commercial viability. Scientists continue to highlight the thermal challenge: dissipating heat in a vacuum is extremely difficult, requiring roughly 1,200 square metres of radiator , about four tennis courts , to shed a single megawatt. Orbital’s distributed design is partly an attempt to engineer around those thermal and manufacturing limits.

For now, Orbital represents a small cheque, a rideshare slot two years out, and a very large idea. Whether orbital data centres become real infrastructure or remain a pitch deck depends on physics and economics that nobody has solved yet. But with AI’s energy demand outpacing the grid, investors are increasingly willing to fund the moonshots , in this case, almost literally.