Grid constraints and rising electricity prices are prompting data center operators to explore new ways to power their facilities.
issued Wednesday, July 22, 2026 · 01:29 PM
Advances in artificial intelligence could dramatically change energy demand in sectors that are putting pressure on the world’s power grids, according to Singapore investment firm Temasek.
Russell Tam, head of emerging technologies at Temasek Global Investments, said the adoption of more efficient AI architectures, greater innovation in materials discovery and improvements in chip manufacturing could significantly reduce the impact on the industry.
“We may see the energy equation for generating AI tokens change quite dramatically,” Tam said at the Bloomberg Sustainable Business Summit in Singapore on Tuesday, July 21. The existing generation of AI, he said, “is a fundamentally inefficient AI architecture, despite its amazing and rapid progress.”
Temasek said in early 2026 that it was unlikely to meet its goal of halving carbon emissions from its portfolio by 2030 from 2010 levels, due in part to rising energy demand driven by AI.
Grid constraints and rising electricity costs are forcing data center operators to look for new ways to power their facilities. At the current pace, U.S. data centers will account for about 20% of U.S. electricity consumption by 2035, up from 5.9% today, according to BloombergNEF.
Companies like UK-based CuspAI, backed by investors such as Temasek and Bezos Expeditions, aim to improve semiconductor production by reducing or eliminating the use of some rare metals.
“We have invested in very novel materials and chip architectures that are significantly more energy efficient,” Tam said. “We have invested in new, unproven AI model architectures that are significantly more energy efficient.”Bloomberg
