SpaceX's forthcoming IPO serves as a compelling indicator of its profound impact on the venture capital landscape. The company has fundamentally reshaped the industry's perception of long-term, capital-intensive space ventures, creating an environment where even a talented founder without prior space experience can secure funding for an ambitious space data center enterprise.
Orbital, a nascent firm that emerged in May from a16z’s Speedrun startup accelerator program, recently closed a $5 million seed round. This company is the latest to propose performing AI inference in space, a plan contingent on the regular operation of SpaceX's Starship rocket. Its diverse investor base includes Basis Set, Human Element, Wayfinder, Antler, Anti Fund, Ascent, Rubik, Zero Knowledge Ventures, LYVC, Feld Ventures, New Legacy, FNDR, UpHonest, and Asterisk.
Founder and CEO Euwyn Poon brings a strong entrepreneurial background, having previously founded the e-scooter company Spin in 2017 and successfully sold it to Ford a year later, subsequently joining the automotive giant. According to a16z partner Andrew Chen, when Poon was ready for his next venture, Speedrun was eager to support him. Chen told TechCrunch that Poon explored several concepts before ultimately focusing on space data centers.
The core proposition for space-based computing is widely understood: the relentless demand for AI compute faces slow deployment on Earth. The allure of space offers limitless solar energy and streamlined environmental reviews. However, the primary impediment remains the formidable economics of launching payloads into orbit, which currently renders the business case financially unviable.
Like many of its peers, Orbital is placing its strategic bet on SpaceX successfully developing and commercializing its Starship rocket. Poon articulated this reliance, stating, “We will get to full scale when Starship comes online.” He further explained that the cost associated with the current cutting-edge Falcon 9 makes the venture “not economically feasible” at present.
Currently, Poon and his team of approximately a dozen professionals in Los Angeles, boasting experience from Amazon LEO, SpaceX, and Northrop Grumman, are preparing for a demo flight. This test will involve deploying an Nvidia Blackwell chip on a partner’s satellite to rigorously evaluate Orbital’s radiation shielding and thermal management technologies. The company aims to launch its inaugural data-processing spacecraft, equipped with Nvidia’s Space-1 Vera Rubin-class GPUs, in 2028.
At that juncture, Orbital plans to commence piece-wise inference work, a strategy designed to generate revenue with each satellite launched. This approach mirrors that of rival data center startup Starcloud, which already has a GPU in orbit and intends to deploy additional units to create income until Starship enables the deployment of its full constellation.
Orbital’s ambitious long-term objective is to deploy 10,000 satellites, collectively delivering a distributed gigawatt of computing power, with each satellite contributing 100 kW. For context, Elon Musk has indicated that SpaceX anticipates its AI satellites will produce up to 150 kW, while Starcloud projects fielding larger 200 kW-rated spacecraft for chip operation.
Not all companies are willing to wait for Starship. Cowboy Space Company, another space data center startup backed by a16z, recently opted to develop its own rockets. Similarly, Jeff Bezos’ Blue Origin has announced its intention to launch data centers into space utilizing its New Glenn launch vehicle.
Poon expresses confidence that the expansive demand for AI will create opportunities for numerous companies to thrive. He conveyed to TechCrunch, “There’s so many lanes for companies in our space to pursue,” before elaborating on the diverse options, including companies focusing on different AI workloads, design philosophies, and conceptualizations of what a space data center entails.
Andrew Chen highlighted Poon’s proven ability to scale operations, citing his success in deploying 250,000 scooters across 100 cities, as evidence of his capability to manage the intricate demands of building an aerospace company. Chen acknowledged that such a project might span a decade and require $5 billion or more, but noted that venture firms are now more amenable to these extended timelines.
Chen reflected on the industry's evolution, stating, “This kind of thing would have sounded crazy 10 years ago when we were all building mobile apps.” He added, “Starting it in 2026 just lets you tap into all the energy and excitement that’s that’s happening in the capital markets.”
Poon’s entry into the space data center business followed a unique trajectory. After departing Ford, he acquired an Nvidia A100 on an impulse, co-locating it in a Santa Clara data center to serve open-weight models. This direct, hands-on experience solidified his conviction in the immense value of delivering compute in the current AI era.
Now, his formidable task is to deploy several thousand GPUs into space.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.