DSpark Cuts DeepSeek Inference Costs — And That's Driving China Internet's Outperformance
SCMP Business reports that DSpark, a system built on DeepSeek's architecture, is materially easing inference bottlenecks and reducing chip strain — translating directly into lower AI operational costs for Chinese internet platforms. This is the kind of deflationary AI cost signal that re-rates the margin profile for NTES, PDD, and Tencent's cloud businesses. NetEase +7.74% today is the clearest expression of the market pricing in lower AI compute expenses improving unit economics. For investors tracking China's AI stack vs the US model, DSpark's efficiency gains represent the domestically-derived alternative to Nvidia-dependent infrastructure — technically significant given US chip export controls that block China's access to the H100/A100 generation. The trend is lower inference cost, higher AI-platform margin, and KWEB as the primary beneficiary.
Read at SCMP Business ↗