21
In December, a discussion that had long existed mainly in academic and aerospace circles touched a broader audience when Lado Okhotnikovfounder of Holiversepublicly suggested that the long-term energy needs of artificial intelligence might eventually require solutions that go beyond terrestrial power systems. Among the possibilities he raised was the idea of harvesting solar energy in space and transmitting it down to Earth to support AI infrastructure. Okhotnikov also stated that his team is exploring this direction.
The suggestion was not presented as a near-term solution or a roadmap for implementation, but as a means of highlighting the scale of the challenge ahead: AI’s accelerating demand for reliable, continuous electricity.
That framing raises a key editorial question: if the idea of powering AI from space is taken seriously by some technologists, does that reflect visionary foresight — or is it an indicator of how pressing the underlying energy constraint has become?
Why Earth-Based Energy May Be Stretched
Data centres already account for a significant share of global electricity consumption. In 2024, data centre electricity use reached around 415 terawatt-hours (TWh), equivalent to approximately 1.5 per cent of the world’s total electricity consumption. Projections by the International Energy Agency (IEA) indicate this figure could more than double to around 945 TWh by 2030 as demand for AI and other digital services grows, making data centres a prominent driver of power demand in advanced economies.
In the United States, data centres consumed roughly 4 per cent of national electricity in 2023 and could account for approximately 9 per cent by 2030, with individual hyperscale facilities consuming as much electricity as tens of thousands of homes.
AI-optimised servers are significantly more energy-intensive than traditional computing hardware. According to industry research, AI-focused accelerated servers may account for roughly 44 per cent of total data centre power consumption by 2030, up from less than a quarter today, and their electricity usage is projected to increase nearly fivefold between 2025 and 2030.
Terrestrial renewable energy sources are expanding, and wind, solar PV and hydro together generated approximately 27 per cent of the electricity consumed by data centres in recent years. Yet intermittency and transmission constraints remain challenges: solar and wind do not produce energy continuously, and many grids require balancing power and storage capacity to match supply to load.
These figures illustrate the scale of demand that will need to be met in the years ahead — not only to support current workloads but to sustain the next generation of large-scale AI systems.
Energy from Space: Concept and Practical Questions
Space-based solar power is conceptually straightforward. In orbit, solar panels could capture sunlight without atmospheric interference or night cycles, theoretically producing a continuous supply of energy. Energy would then be transmitted wirelessly — likely via microwave or laser systems — to receiving stations on Earth where it would be converted into electricity.
The underlying physics is well understood, and small demonstrations of wireless power transmission have been conducted. Longstanding research by space agencies and institutions has shown that the concept is possible in principle. But feasibility in theory does not imply feasibility in practice, particularly at the scale required to feed large portions of terrestrial energy demand.
The Cost and Complexity of Orbital Infrastructure
Generating significant power from orbit would demand infrastructure that extends far beyond current capabilities. Solar arrays in space would need to be substantially larger than any operational satellites. Autonomous or semi-autonomous assembly and maintenance in orbit would be necessary, as would highly sophisticated energy-transmission systems with robust safety controls. Ground-based receiving stations would need to be integrated into existing power networks, requiring significant grid upgrades.
The economic and logistical challenges of building and operating such systems are substantial. Launch costs, material durability in space environments, repair and replacement cycles, and international regulatory frameworks for orbital infrastructure all contribute to uncertainty about whether space-based solar energy can ever be deployed at meaningful scale.
Reliability and Systemic Risk
AI infrastructure is sensitive to disruptions. Uptime requirements for large models and real-time inference mean that energy systems feeding them must be exceptionally reliable. Introducing an energy source dependent on orbital mechanisms would create new dependencies far beyond Earth’s surface.
Current terrestrial grids already face reliability challenges: extreme weather events and ageing infrastructure have stressed power networks in regions experiencing rapid data centre growth, revealing grid fragilities that policymakers and utilities are still addressing.
Whether an energy system that includes space-based components could match the reliability, serviceability and resilience of terrestrial power sources without substantial redundancy and contingency mechanisms is an open question.
Governance and Control
Orbital energy infrastructure would not only be an engineering challenge but a governance one. Questions about ownership, access priorities, regulatory authority and dispute resolution loom large. Unlike terrestrial grids, which operate under established regulatory regimes within national jurisdictions, space-based systems would likely require new international agreements and oversight frameworks to manage competition, safety, and equitable access.
These issues do not have clear answers today, and their resolution will be necessary before space-based energy can be seriously pursued as part of global power infrastructure.
Exploring Possibilities, Not Predicting Certainties
Space-based solar energy remains, at best, a long-term research direction. There is no articulated plan for large-scale deployment, nor a clear timeline for when — or if — such systems could contribute meaningfully to global energy supply. Yet the fact that the idea has entered public conversation is indicative of how seriously some experts take the energy question.
If the growth of AI continues to push data centre electricity demand upward — doubling or more by the end of the decade — energy planners and technology leaders will need to consider new paradigms alongside incremental improvements to existing systems. These could include more efficient hardware, advanced cooling technologies, better grid integration with renewables, and innovative demand management.
In this context, the idea of gathering solar energy in space does not sit on the fringe for its own sake. It highlights the tension between anticipated demand and the pace of energy infrastructure development on Earth.
Whether AI data centres will one day be supported in part by energy collected in orbit is a question without a clear answer. The technical, economic, regulatory and reliability challenges remain significant, and timelines are uncertain. What is clear is that space-based solar power for AI infrastructure is theoretical and long-term, and that the conversation was brought into wider view through figures such as Lado Okhotnikov, who have said their teams are exploring this direction.
The result is not certainty, but suspense: an invitation to examine the limits of current energy systems and to explore all plausible paths forward — including those that lie well beyond the near term.
Comments are closed.