AI's energy demand: Why low-carbon electricity and grid upgrades are crucial for tech growth

A new opinion piece argues that AI development is not just a software story but a physical infrastructure challenge. Training and running large AI models require huge amounts of stable electricity, and as tech firms push for carbon neutrality, the source of that power matters. The article highlights that current renewable subsidies focus too much on generation volume and not enough on system integration, storage, and flexibility, which are critical for AI data centers that cannot tolerate grid instability. The authors point to research comparing Hong Kong, mainland China, and Australia, showing that subsidy frameworks often miss the value of technologies like hydrogen for long-duration storage and dispatchable power. They call for a shift toward performance-based incentives that reward flexibility and peak reduction. For cities like Hong Kong aiming to be innovation hubs, energy policy must be treated as industrial policy, with investment in grid modernization and storage alongside renewable capacity. This piece is relevant for anyone tracking how AI growth will drive electricity demand and reshape clean energy policy. It connects the dots between data center expansion, renewable integration, and the need for smarter subsidy design that supports a resilient, low-carbon grid.