01/09/2026
For the past few years, one of the loudest arguments against the AI boom has been its enormous and rapidly growing demand for electricity.
That concern is legitimate.
But look deeper at what that demand is unleashing: a massive accelerant for the renewable energy buildout.
Microsoft has agreements covering up to 40 GW of new renewable energy, including 10.5 GW through Brookfield. Amazon has backed 700+ carbon-free energy projects totalling more than 40 GW, dominated by solar and wind, alongside batteries, geothermal and selected nuclear investments.
Google spent US$4.75 billion acquiring Intersect Power and is developing gigawatt-scale clean-energy parks alongside AI data centres. It's partnering on the world's largest planned 100-hour battery, backed by new wind and solar, and has contracted 1 GW of demand response, allowing data centres themselves to respond to grid conditions.
China requires new data centres across its eight national computing hubs to source at least 80% renewable electricity, alongside enormous investment in batteries and ultra-high-voltage transmission. And now Australia is developing national rules requiring hyperscale data centres to bring forward new, additional renewable generation to meet their enormous new electricity demand, backed by the firming required to make it reliable.
Something much bigger is happening here. AI isn't bringing back the baseload grid. It may actually be accelerating its replacement.
Every hyperscale data centre creates an enormous new load, requiring enormous amounts of new generation, storage and grid infrastructure. Increasingly, the technologies capable of being deployed at the required speed, scale and cost are:
➡️ Solar ➡️ Wind ➡️ Batteries ➡️ Transmission ➡️ Flexible demand
Yes, nuclear, geothermal and other firm clean generation can contribute where their economics and deployment timelines make sense. Gas will continue playing peaking and firming roles in some grids. Hyperscalers need reliable electricity, not ideological purity.
But that's precisely the point. Microsoft, Amazon, Google and other hyperscalers aren't investing billions in SWB because they've suddenly become environmental charities. They're doing it because they need staggering quantities of new electricity, quickly and competitively.
And scale has consequences:
➡️ More deployment strengthens supply chains.
➡️ More batteries firm renewables.
➡️ More transmission unlocks renewables.
➡️ Flexible loads help balance the grid.
➡️ More SWB demand drives costs lower.
That's why I increasingly think AI could become one of the largest accelerants of the energy transition we've ever seen.
The very industry accused of becoming one of the world's great new energy problems may instead become a Trojan Horse for rebuilding the electricity system itself.
And inside this one isn't an army of Greeks. It's an army of SWB.
I'm currently working on a six-part video and blog series, THE GREAT DATA CENTRE DEBATE, digging much deeper into all of this. The public concerns surrounding AI data centres are real and deserve serious answers: electricity and household bills, water, land, jobs, grid infrastructure, emissions and ultimately who pays.
The purpose isn't to dismiss those concerns. It's to put each one through the same test:
➡️ Follow the evidence. Follow the economics. See where it leads.
The irony of the AI boom may ultimately be extraordinary. The technology accused of becoming one of the world's great new energy problems may instead become one of the most powerful forces accelerating the transformation of the electricity system.
The horse is already inside the gates.
The future isn't being built around yesterday's grid. It's being built around speed, flexibility and economics.
