An October announcement that Nvidia would supply 260,000 GPUs to South Korea was seen as a diplomatic gain, but most supply is concentrated among large companies. About 20 percent is available for academia, research institutes and startups, with the government leasing capacity from cloud firms. The author argues AI demand is shifting toward low-latency inference, where domestic NPUs could be deployed, but cheap large-scale GPU supply could crowd them out. He calls for regional distributed inference hubs using idle public assets and local demand aggregation.

South Korea’s Budget Office said it held a meeting on the direction of fiscal investment in the AI semiconductor industry after its budget chief visited FuriosaAI’s Seoul office. Officials, AI chip company executives and finance experts attended. The office is running a project to visit more than 100 sites to reflect field views in next year’s budget proposal. It plans to reflect views from the meeting in the 2027 budget plan and fiscal management plan.