Two solar executives say making better use of existing power plants, transmission lines and batteries could help meet the energy demands of AI. Their proposals range from using underused grid capacity to beaming solar power from space, where the sun, inconveniently for night-time demand, keeps shining.
Watch Desk analysis
What happened
At Fortune’s 500 Innovation Forum in Detroit, Overview Energy co-CEO Abdullah Al-Shakarchi argued that gas plants and transmission lines are often underused. He said gas plants run at 60 to 70 per cent utilisation in some areas, while transmission can sit idle. Using those assets more effectively, he said, could reduce how much new infrastructure needs to be built.
Sunrun CEO Mary Powell made the case for power generated nearer to where it is used, including rooftop solar and home batteries. She said Sunrun had about 1.06 million subscribers and five gigawatt-hours of networked home battery storage as of 30 September. Both executives linked better use of existing energy resources to the pressure data centres are putting on power demand and public concern about electricity costs.
Al-Shakarchi also described Overview Energy’s plan to collect sunlight with satellites and beam it as near-infrared light to existing utility-scale solar farms, allowing them to generate power at night. He argued that, if no new utility-scale solar farms were built in the US after 2030, switching existing farms on at night could supply a quarter of the country’s load growth over the following decade. That is the company’s projection, not a demonstrated outcome. Meta has agreed to early access to up to one gigawatt of Overview’s planned capacity; the article says an orbital demonstration is planned for 2028, with commercial delivery expected by 2030.
Why it matters
The AI build-out is also an energy and infrastructure story. More efficient use of existing plants, grid connections and stored power could ease some pressure to build new capacity, though it does not make the additional demand disappear. Sunrun says some markets still do not allow home batteries to send power back to the grid, making regulation part of the practical equation.
The space-solar proposal is the more eye-catching pitch, but the near-term argument is about flexibility: coordinating energy supply, storage and demand so existing equipment does more work. That could matter to data-centre operators and households alike, particularly as electricity costs and local opposition to new facilities become part of the AI debate.
Our read
The strongest point here is less science fiction than good housekeeping: before building more, see whether the grid’s existing assets can be used better. That is not a magic answer to AI’s power appetite, and Overview’s orbital timetable still has to survive contact with engineering and economics. But it gives the energy debate a useful question beyond “where can we build another power plant?”
What to watch
- Whether Overview’s planned 2028 orbital demonstration happens and supports its projections.
- Whether more regions let household batteries send electricity back to the grid.
- Whether data-centre growth leads to practical grid reforms, new capacity, or both.
Discussion spark: Should governments and utilities prioritise getting more from existing grids and batteries before approving new power infrastructure for data centres, or does that risk slowing capacity the AI build-out needs?
Sources and evidence
- Source update (19 November 2026, 08:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.