
Supercomputing is splitting into two tracks: publicly funded exascale systems ranked by TOP500/HPL FLOPS and hyperscaler-built AI campuses measured by megawatt–gigawatt capacity.
Supercomputing has split into two distinct competitive tracks measured by different metrics. Publicly funded systems are ranked by the TOP500 list using a calculation speed metric called HPL FLOPS, while privately built AI training campuses are measured by their power capacity in megawatts and gigawatts. This split means two systems can both legitimately claim to be the world's most powerful supercomputer while serving entirely different purposes and using different workloads. The distinction matters because national governments now treat compute capacity as strategic infrastructure, leading them to fund their own systems rather than rely solely on private hyperscaler facilities.

Local opposition groups surged to 430 from 76 since 2025, while recent Virginia project failures suggest community acceptance is emerging as a new site-selection variable for AI infrastructure.

Meta’s 1 GW Alberta campus reveals how hyperscalers now secure power and transmission years before announcing AI campuses.

The company is taking a modular approach to designing these chips, anticipating that their needs will change as AI evolves rapidly by the time the chips are in production.
Want to go deeper than the news? Explore live, cohort-based AI courses taught by practitioners.
Browse AI courses on Maven