From Scooter Startups to Space Data Centers: How Orbital Is Revolutionizing AI Compute
Orbital, backed by a16z and other prominent investors, is set to revolutionize AI compute with its ambitious plans for space data centers. Learn how this startup is tackling the challenges of launching into orbit.
Admin User
SpaceX's impending IPO has shifted the venture industry's perspective on long-term, capital-intensive ventures in space. A16z’s Speedrun accelerator program recently funded Orbital, a new firm promising to do inference in space – just as soon as Starship is flying regularly.
About Orbital
Orbital emerged from the a16z startup accelerator and secured $5 million in seed funding. The company’s founder, Euwyn Poon, brings valuable experience from selling his previous venture, e-scooter company Spin, to Ford.
The Business Model
Orbital aims to deploy 10,000 satellites that provide a distributed gigawatt of computing power. Each satellite will offer 100 kW of power. The main challenge is the high cost of launching payloads into orbit – currently, the Falcon 9 rocket makes this economically unfeasible.
Technology and Partnerships
The company is working on a demo flight that will test its radiation shielding and thermal management technology using an Nvidia Blackwell chip. By 2028, Orbital plans to launch its first data-processing spacecraft with Nvidia’s Space-1 Vera Rubin-class GPUs.
Competitors in the Market
Other space data center startups like Starcloud and Cowboy Space Company are also eyeing this lucrative market. Blue Origin has announced plans to launch data centers using its New Glenn launch vehicle, showing that multiple players are entering this space.
Challenges and Future Prospects
The path to profitability is uncertain, but Poon believes the insatiable demand for AI compute will drive success. Each satellite launched could generate revenue through piece-wise inference work. Over the long term, a project like this might take a decade and $5 billion or more, but venture firms are now more comfortable with such extended timelines.
The Journey to Space
While Poon’s journey into space data centers is unique, his experience scaling up a company that deployed 250,000 scooters across 100 cities provides valuable insights. He bought an Nvidia A100 on a whim and co-located it in a Santa Clara data center to serve open-weight models, gaining firsthand experience with the value of delivering compute in the era of AI.
