China’s AI company DeepSeek has announced a significant increase in API service pricing. Additionally, there are reports that the second round of financing, which was temporarily suspended due to leaked recordings of founder Liang Wenfeng, has been restarted.
On August 6, DeepSeek released a statement saying, “We plan to adjust the pricing of DeepSeek API services overall in the near future, with a significant increase expected. Please plan your usage accordingly. Specific details will be subject to formal notification.”
DeepSeek API has introduced a peak-valley time-based billing mechanism, with double billing during workdays’ peak hours and maintaining a lower price during nights, weekends, and holidays to ease the pressure on computing resources.
Industry analysts believe that the imminent substantial price hike by DeepSeek is a response to the high cost pressure of computing power. Large-scale model inference consumes significant resources of high-end GPUs, and maintaining low prices for an extended period would continue to drain enterprise cash flow.
According to Bloomberg’s report on August 6, DeepSeek plans to significantly raise the prices of its artificial intelligence services, marking an unusual shift for the Chinese company that could be a turning point in the Chinese AI market. Previously, leaders of other AI companies in China had followed DeepSeek’s model by providing low-cost open-source services.
The report indicates that DeepSeek is undergoing a large-scale financing and has already begun preparations for its earliest initial public offering (IPO) scheduled for this year. Founder Liang Wenfeng will face the triple challenge of balancing investor expectations, rapid market expansion, and the substantial funds needed to establish high-cost computing infrastructure.
As reported by China’s “Finance” magazine on August 5, DeepSeek has restarted its second round of financing, with plans to raise 50 billion yuan in this round, scheduled to be finalized in late August.
Prior to this, DeepSeek launched its first round of financing in April this year and completed negotiations in June, raising 50 billion yuan with a valuation exceeding 350 billion yuan, setting a record for the largest funding scale in the first round of financing in China’s AI large model history. The pre-investment valuation for the second round of financing has increased by about 43% compared to the first round, reaching 500 billion yuan. If successful, the cumulative fundraising from the two rounds will exceed 100 billion yuan.
According to insiders, the intended fund size for the first round of financing had exceeded 100 billion yuan, but ultimately only 50 billion yuan was raised, indicating that there is still a significant amount of capital interested in participating and competing for shares in this round. Currently, some investment institutions that had previously been in contact have not received news about the restart of financing.
Previously, DeepSeek abruptly halted the signing of the second round of financing at the end of July, partly due to leaked recordings of meetings with investors circulating online, which had angered Liang Wenfeng.
According to meeting summaries compiled by several Chinese media outlets, some of Liang Wenfeng’s statements may have displeased the Chinese government. Political observer Xia Yan believes that the Chinese government has long emphasized the advantages of China’s massive data and low labor costs. However, Liang Wenfeng bluntly pointed out the deficiencies in “high-quality AI training data” in China, even admitting that DeepSeek can only “tag low-cost data” initially. Furthermore, his characterization of the current chaotic AI industry as “resource-wasting and scattered” goes against the propaganda line of the Chinese government, contradicting the policies of various local governments that are actively subsidizing and supporting the large model industry.