OpenAI CEO Sam Altman dismissed SpaceX’s proposal for orbital data centers as “ridiculous” during a public interview in New Delhi, drawing a sharp line between his company’s ground-level AI infrastructure strategy and Elon Musk’s interest in moving computing power into space. The remark came as SpaceX seeks federal approval to place up to 1 million solar-powered data centers in orbit, a proposal that has attracted regulatory attention. The exchange captures a real and growing disagreement over where the next wave of AI computing capacity should be built.
Altman’s Blunt Critique at Express Adda
Speaking at the Express Adda event in New Delhi, Altman was interviewed by Anant Goenka against the backdrop of an AI-focused gathering that brought together policymakers, technologists, and business leaders. When asked about the prospect of space-based data centers, Altman did not hedge. He called the concept “ridiculous” in the current environment and tied his objection to two specific problems: the cost of launching hardware into orbit and the near-impossibility of maintaining it once it is there, especially at the scale modern AI workloads demand.
His sharpest line targeted the practical absurdity of remote repairs. Altman asked, “How do you fix a broken GPU in space?” according to coverage in India, distilling the core engineering challenge into a single question. Terrestrial data centers already require constant hardware swaps, cooling adjustments, and power management by on-site technicians. Replicating that maintenance cycle hundreds of miles above Earth, with no physical access and extreme launch costs, remains far beyond current logistics. Altman went further, suggesting that space-based data centers would not matter this decade, according to The Indian Express, a timeline that challenges SpaceX’s ambitions and reinforces OpenAI’s preference for scaling conventional infrastructure.
What SpaceX Actually Filed With the FCC
The target of Altman’s criticism is a formal application SpaceX submitted to the FCC’s International Communications Filing System outlining a massive orbital compute network. In that document, the company proposed deploying up to 1 million solar-powered data centers in orbit, a scale that would eclipse any existing satellite constellation and signal a new phase for commercial space infrastructure. Reporting from financial media described the filing as SpaceX’s most explicit move yet to expand beyond broadband internet service and into the AI compute market, positioning the company as a potential rival to terrestrial cloud providers.
The technical pitch leans on two advantages that Earth-based facilities struggle to match: abundant solar energy collection and passive radiative cooling in the vacuum of space. In its filing, SpaceX framed orbital platforms as a way to capture continuous sunlight without weather or nighttime interruptions, and to shed heat directly into space without the water-intensive cooling towers used on the ground. The company even invoked the Kardashev scale, a theoretical framework for measuring a civilization’s energy use, casting the project as a step toward harnessing a larger share of the Sun’s output. Yet as technology reporters have pointed out, the proposal also raises serious congestion concerns in already busy orbital shells, since adding a million data-center satellites would demand unprecedented coordination to avoid collisions and manage space debris.
Regulatory Momentum Behind SpaceX
Despite Altman’s skepticism, SpaceX has already secured regulatory decisions on its Starlink expansion that show U.S. agencies are actively weighing the implications of a denser orbital environment. The FCC recently approved the company’s Starlink Gen2 Version 3 authorization, clearing another tranche of satellites after a review that required close interagency consultation on spectrum and interference issues. Assistant Secretary Evelynne Roth of the National Telecommunications and Information Administration publicly praised the process in a formal statement, signaling that federal agencies are not simply resisting SpaceX’s growth but are actively working through the technical and policy implications of a denser orbital environment.
That regulatory track record matters because it shows SpaceX is not merely circulating ideas outside the formal regulatory process. The company has repeatedly demonstrated an ability to move ambitious proposals through the FCC pipeline, and each approval builds procedural precedent for the next round of applications. If the data center constellation filing follows a similar path, SpaceX could secure spectrum rights and operating authority long before the engineering challenges Altman highlighted are fully resolved. That would create a gap between regulatory permission and technical readiness, in which SpaceX could experiment, iterate on early prototypes, and shape global norms for orbital compute infrastructure while competitors are still debating the feasibility of the idea.
The Economics That Divide Them
Altman’s argument rests on a straightforward cost comparison between rockets and real estate. Even if SpaceX’s Starship program drives launch prices lower than any previous generation of rockets, putting a server rack into orbit still requires expensive radiation shielding, hardened electronics, and specialized deployment hardware, on top of the launch itself. By contrast, terrestrial data centers benefit from cheap land in remote regions, existing high-voltage transmission lines, relatively low-cost fiber optic connectivity, and a workforce that can physically replace a failed drive or GPU within hours. Insurance, regulatory compliance, and construction costs on Earth are significant, but they are well understood and amortized over decades of cloud expansion.
SpaceX’s counter-argument, embedded in its FCC filing and echoed by supporters, is that Earth-based facilities face their own tightening constraints that will eventually reshape the cost curve. AI training runs are consuming electricity at a pace that is straining regional grids and sparking local political resistance to new substations and transmission corridors. Cooling systems for dense GPU clusters are becoming a major engineering and environmental burden, particularly in water-stressed regions where data centers compete with agriculture and households for limited supplies. In orbit, solar panels can collect energy without weather interruptions, and radiative cooling requires no water or mechanical chillers, potentially reducing both operating expenses and environmental impact. The unresolved question is whether those orbital advantages can close the enormous upfront cost gap quickly enough to matter commercially. Altman is effectively betting that they will not, at least not before 2030, while SpaceX appears to be wagering on a longer horizon in which terrestrial energy bottlenecks make orbital compute comparatively more attractive.
A Rivalry With Broader Stakes
This dispute is not just a technical argument between two high-profile executives; it reflects a deeper split in the AI ecosystem over how to manage explosive growth in compute demand. OpenAI and its partners are channeling billions of dollars into new ground-based data centers, racing to secure long-term power purchase agreements, grid interconnections, and construction permits in multiple countries. Their strategy assumes that, with enough investment and regulatory negotiation, terrestrial grids can be upgraded, new renewable projects can be built, and data centers can be integrated into local economies through jobs and tax revenue. That approach also keeps compute infrastructure closer to users and to the physical locations where training data is generated and stored, reducing latency and simplifying compliance with data residency rules.
SpaceX, by contrast, is sketching an entirely different supply chain for compute, one that bypasses land-use conflicts, local permitting fights, and some forms of geopolitical risk by lifting infrastructure into orbit. If even a fraction of its orbital data center vision becomes reality, it could create a parallel layer of cloud services that operate above national borders yet still depend on ground stations and regulatory approvals for spectrum and downlink rights. That possibility raises new questions about who controls access to orbital compute, how liability for collisions and debris is assigned, and whether international frameworks will be updated fast enough to keep pace. Altman’s dismissal of the project as “ridiculous” underscores how far apart leading figures in AI are on these issues, but his critique may also serve as an early marker in a longer debate over whether the future of AI infrastructure is rooted firmly in the ground or destined to reach into space.
More From The Daily Overview
*This article was researched with the help of AI, with human editors creating the final content.
Grant Mercer covers market dynamics, business trends, and the economic forces driving growth across industries. His analysis connects macro movements with real-world implications for investors, entrepreneurs, and professionals. Through his work at The Daily Overview, Grant helps readers understand how markets function and where opportunities may emerge.
