CGN Tech Blog: DeepSeek’s Permanent Price Cut Turns AI Infrastructure Into a Margin Fight
The Chinese AI startup’s V4-Pro price cut links model competition to chip supply, inference costs and cloud economics.
PALO ALTO | DeepSeek’s decision to make a 75 percent price cut permanent on its flagship V4-Pro model turns the AI race into a sharper contest over infrastructure cost, chip supply and developer adoption.
Reuters reported that the company cut V4-Pro API costs to between 0.025 and 6 yuan per million tokens depending on usage type, down from 0.1 to 24 yuan previously.
The move lands inside China’s broader AI-chip push, where Huawei’s Ascend chip sales have benefited from U.S. export controls that restrict Nvidia’s most advanced chip sales in China.
The technology question is not only model performance. It is whether lower inference pricing changes cloud procurement, developer habits and the economics of running AI products at scale.
Additional Reporting By: Reuters
What this means
The signals to watch are developer adoption, cloud pricing responses, Huawei chip capacity and whether Western AI firms answer with lower prices or premium-performance positioning.