With AI pushing electricity use to breaking point, Meta joins the club of big tech companies betting on orbital solar energy to power its data centers.
Scaling AI infrastructure requires rethinking how electricity is generated and delivered. While power demands surge, raising questions about whether Earth can keep up, Meta has joined big tech companies that are increasingly looking skyward to solve AI’s power problems.
“Today’s clean energy technologies have real limits: solar depends on sunlight, wind depends on weather, and the grid still needs more storage to make the most of both,” Meta says in a blog post.
As companies accelerate their artificial intelligence ambitions, the next push may be to harvest solar energy from space to power their rapidly expanding data centers.
The math is simple – in space, there is no nighttime, so the flow of energy is 24/7 without expanding land use or building entirely new facilities.
Big tech is desperate for electricity
Tapping into a constant flow of electricity is on the minds of most big tech companies. AI’s growing appetite is now driving historic spikes in power demand. By the end of 2025, AI systems were projected to devour nearly 50% of all data center power, reaching 23 gigawatts.
Data centers account for 3% of EU electricity demand, and their consumption is expected to increase rapidly this decade as artificial intelligence expands. US government research suggests that data centers consume about 12% of total US power.
Microsoft CEO Satya Nadella warned that energy costs and infrastructure will be the key factors determining how AI scales worldwide. Meanwhile, the White House convened major players – Microsoft, Amazon, Meta, and Anthropic – to discuss pledges around AI data center power costs under the Trump administration.
Microsoft is testing superconducting power lines to address AI's ballooning electricity demands, while in Europe, the company and Amazon are repurposing aging coal plants into AI data centers.
The International Energy Agency projects global data center electricity demand will more than double by 2030 to around 945 terawatt-hours, slightly exceeding Japan's total energy consumption.
What does AI cost the environment?
A comprehensive study by scientists at Cornell University calls net-zero emissions targets by tech giants unrealistic.
They suggested that AI server deployment across the US alone could generate between 24 and 44 million metric tons of carbon dioxide annually by 2030.
The numbers are the same as adding 5 to 10 million cars to US roads. Water consumption should also rise massively, reaching levels equal to the annual household water usage of 6 to 10 million Americans.
Goldman Sachs Research forecasts that approximately 60% of this increased demand will be met through fossil fuels, potentially adding 220 million tons to global carbon emissions. A sort of new fracking boom, it seems.
So, it's no surprise that finding renewable solutions is on the table.
How can you harvest solar energy from space?
On April 27th, Meta announced two new partnerships aimed at securing an energy supply from space. One of the partners is a Virginia-based aerospace startup, Overview Energy, which develops technology to capture solar energy in orbit and transmit it back to Earth.
Overview’s satellites are operating around 22,000 miles above the surface, in the geosynchronous orbit, where sunlight is constant. Satellites collect solar energy and beam it down to existing solar farms as low-intensity infrared light. Those facilities then convert the signal into electricity and feed it into the grid just like conventional solar power.
Generating more energy is only half the equation. Storing it is the other. For that, Meta is partnering with California-based company Noon Energy, which is developing ultra-long-duration storage technology designed to hold power longer.
According to Meta, unlike conventional lithium-ion batteries, Noon Energy’s system uses solid oxide fuel cells and carbon-based storage to deliver more than 100 hours of energy capacity.
The goal is to smooth out the natural volatility of renewable energy, ensuring that power generated during peak conditions can be used long after the sun sets or the wind drops.
Have thoughts about this topic? Others do, too. Join them in the discussion.
If successful, the first orbital demonstration is expected in 2028, with potential commercial deployment beginning around 2030. These partnerships add to Meta's growing portfolio of energy investments, which have already contracted more than 30 gigawatts of renewable energy globally.
The company has previously backed geothermal projects with Sage Geosystems and XGS Energy. The company has also committed to nuclear energy through agreements with Vistra, TerraPower, Oklo, and Constellation Energy.
Big tech is experimenting with space solar energy
Meta is not the only tech company exploring space possibilities. Google’s Project Suncatcher has been investigating solar-powered satellite constellations to run AI workloads in orbit.
In space, solar panels receive uninterrupted sunlight with no clouds, weather, or nighttime, dramatically increasing energy yield compared to ground-based solar.
In January 2026, SpaceX filed with the FCC to launch a constellation of up to 1 million solar-powered satellites designed to function as orbital data centers for AI workloads. The satellites would harness continuous solar energy in orbit, where panels are roughly 6x more effective than on Earth.
Elon Musk has stated he believes space will become the lowest-cost location for AI data centers within 2-3 years.
Jeff Bezos has publicly envisioned gigawatt-scale data centers in space within 10-20 years. Amazon is moving towards this goal by partnering with the aerospace technology company Starcloud, which will launch AWS Outposts hardware to space in October this year.
The company plans to deploy a total of 88,000 satellites to train and operate AI models and other cloud computing services in outer space.
Unlock exclusive Cybernews content on YouTube.
Your email address will not be published. Required fields are markedmarked