Korean neocloud providers, including Vessl AI (which raised its H100 hourly rate 24.7% to $2.98) and Elice, have lifted GPU rental prices, mirroring increases by US neoclouds CoreWeave and Nebius. H100 cloud rates climbed 21.9% in a month to $3.28, about 70% of H200 pricing, reversing the historical pattern of older chips depreciating as new models arrive, because supply of the newest accelerators cannot keep up with demand. Some Korean companies abandoned bids for government AI transformation projects and universities scaled back research due to inability to secure GPUs, so compute access is now cited as a determinant of AI competitiveness. For financial institutions, this is a direct market-price data point on the scarcity and pricing of AI compute at the retail-rental level, with knock-on effects for Korean AI vendors' cost structures and public-sector project pipelines.
262 other tracked developments touching the same claims.