Electrification Cuts Global Energy Demand 40%, AI Data Centres Test Grids

The numbers come from Nick Eyre, an Oxford professor. Hannah Ritchie reworked his study in a Substack post dated 29 September. Electricity demand rises from 110 to 189 exajoules, but total energy demand drops from 416 to 247. "Electrification is efficiency," Ritchie writes, quoting the International Energy Agency.

The same day, CBC News reported that Newfoundland and Labrador's energy minister, Lloyd Parrott, told the provincial legislature in mid-September that the province's "door is open for business" to AI data centre developers. A department spokesperson, Brodie Thomas, later confirmed in an email that companies have approached the government, but declined to name proponents or projects. According to CBC, tech journalist Paris Marx questioned whether selling excess Churchill Falls power to data centres makes sense. "Does it make sense to be giving that electricity over to data centres when there might be better ways that it can be used?" Marx said. TechNL CEO Andrea King said Labrador has low-carbon hydro, cold weather and land. She added: "It's not just 'does this make sense?' by itself. You have to look at economic costs and opportunity costs of what else you could do with that electricity."

The tension is not confined to Canada. On 29 September, Efficient Computer, a US chip startup, announced a $97M Series B round led by TQ Ventures. That brings its total raise to $173 million at a $650 million valuation. In a blog post, CEO Brandon Lucia claimed the company's processors deliver "10-100x energy-efficiency improvement compared to traditional CPU architectures." The Register did not cover the round, and the post offers no independent verification.

Efficient's pitch rests on Amdahl's Law, the idea that unaccelerated parts of a workload cap total gains. Lucia argues that fixed-function AI accelerators risk obsolescence and leave general-purpose computation inefficient. That argument matters for European industry, where the recent headlines list high energy prices and power market changes as pressure points. Those headlines are context, though, and not part of this report's sources.

Oil remains the wild card. In a Conversation article published on 29 September, two energy economists traced what happened after the US and Israel attacked Iran. Federal data showed crude at the primary US hub at $66 a barrel in late February 2026, before the attacks, and $101 a barrel on 13 April. That spike sends most of the extra money to oil companies, the authors write. Benefits flow to shareholders, governments or sovereign wealth funds depending on ownership.

The electrification case is not a forecast of lower bills. Ritchie's model assumes no efficiency gains beyond electrification and a switch to hydrogen, and it excludes growth in energy use as countries develop. The study also assumes 90% of space heating is replaced by heat pumps, with 10% hydrogen, and that only half of heavy-goods vehicle distance is electrified. High-temperature industrial processes stay largely un-electrified.

For Europe, the arithmetic cuts both ways. Less final energy means less exposure to oil and gas price shocks, but more electricity demand means more grid, more generation and more competition for the same power that data centres want. The Canadian debate, the chip funding and the oil price episode are three versions of the same question: who gets the electrons, and at what price.