Whether AI token demand is translating into real economic output has emerged as a key precondition for forecasting DRAM and high-bandwidth memory supply and demand. Executives at Computex 2026 projected token consumption to surge through 2030, while memory makers outlined capacity expansion plans. Analysts and industry sources noted, however, that AI outputs are often not captured in statistics, making it difficult to verify whether demand is structural or temporary and limiting order visibility.

China\'s AI company DeepSeek is in talks to raise about 70 billion yuan while maintaining free model releases and a low-cost API strategy, with efficiency in computing and memory seen as its core edge. Analysts cited sharply lower memory use versus rivals and optimised cache design that cuts costs and improves inference speed even on lower-performance chips. The approach is also linked to constraints on access to Nvidia\'s newest AI chips.