DeepSeek reportedly raises $15 billion

Set Trending Topics as a preferred source on Google.

DeepSeek is currently closing its latest funding round, according to Reuters. The Chinese AI startup is said to have raised around 100 billion yuan ($14.98 billion) from existing and new investors, as four people familiar with the matter said on Thursday.

Major Chinese names involved in the round

Battery manufacturer Contemporary Amperex Technology (CATL) and WeChat operator Tencent were among the largest investors in this round, said two of the sources, all of whom wished to remain anonymous as the information is confidential. New investors include home appliance manufacturer Midea, food delivery giant Meituan, automaker Geely, and semiconductor manufacturer GigaDevice.

Further investors include Boyu Capital, CPE, and Legend Capital, as well as existing shareholder IDG Capital, according to three of the sources. The total number of investors in this funding round exceeded 15, one of the sources said.

The fundraising is expected to be completed in the coming days, but the financial details and investor composition are not yet finalized and could still change, a second person said.

DeepSeek could double its funding target

DeepSeek, CATL, and Tencent were not immediately available for comment. The other companies mentioned in this article did not immediately respond to requests for comment.

With the current round, DeepSeek may have doubled its original funding target of 50 billion yuan (approximately $7.5 billion). This underscores the strong interest from investors, many of whom receive support from the Chinese state and major corporations. China is eager to get ahead of the US in the race for AI leadership.

IPO could be imminent

The fundraising also signals that DeepSeek is moving closer to tapping public markets. In September, Reuters reported that the young company had hired CITIC Securities to prepare for a possible IPO on the STAR Market.

Last month, the startup unveiled its latest model, DeepSeek-V4.1-Flash, which it says is designed for greater capability, faster inference, higher throughput, and scaling to larger models.