China’s DeepSeek to make permanent 75% price cut on flagship V4‑Pro AI model: How Nvidia may be the reason

Chinese AI startup DeepSeek has announced that it is permanently slashing the price of its flagship V4-Pro model by 75%. The massive price cut keeps the cost of using DeepSeek's most powerful software tool at just a quarter of its original rate. While DeepSeek did not explicitly say what allowed them to drop prices so drastically, it is assumed that the easy availability of Huawei chips may have been the reason for this drop.Notably, when DeepSeek ⁠launched V4 last month, it said the Pro version would cost up to 12 times more than the less powerful Flash version ⁠due to “constraints in high-end compute capacity,” limiting availability, according to a report by news agency Reuters. It added Pro pricing was expected to fall sharply once ⁠Huawei Ascend 950 supernodes are launched in large quantities in the second half of the year.According to an official statement from the company, DeepSeek has reduced its Application Programming Interface (API) costs for the V4-Pro model to between 0.025 and 6 yuan per million tokens (approximately $0.0035 to $0.83), depending on the specific type of deployment. Previously, those exact same workloads cost developers between 0.1 and 24 yuan.In the AI industry, a “token” is a fraction of a word or a unit of text processed by a large language model. By pushing token costs down to a fraction of a cent, DeepSeek is aiming to undercut both domestic rivals and Western AI firms, making advanced AI integration incredibly cheap for software developers.The economics behind DeepSeek's 75% discount may be tied to the ongoing geopolitical trade war over advanced semiconductors. Strict US export controls currently bar Nvidia from selling its top-of-the-line AI graphics processing units (GPUs) to Chinese firms. While these American restrictions were designed to slow down China's AI progress, they have had an opposite effect: creating a massive, exclusive market boom for Nvidia's biggest domestic rival, Huawei.DeepSeek relies heavily on Huawei's homegrown Ascend 950 chips to power and maximise the performance of its V4 models. By optimizing its software to run on local Chinese silicon rather than relying on expensive, restricted or black-market Nvidia hardware, DeepSeek has been able to significantly lower its operational infrastructure costs.