HONG KONG — Chinese artificial intelligence startup DeepSeek has raised about 100 billion yuan, or roughly $15 billion, in a funding round backed by Tencent and the battery maker CATL, according to Bloomberg, which cited people familiar with the matter. Reuters, citing four people with knowledge of the matter, reported this week that the round is expected to close in the coming days.

The money would give the Hangzhou-based company a much larger war chest as it prepares to list in Shanghai, and as it tries to close the resource gap with US rivals.

Bloomberg reported that the fundraising target was raised at least twice, from 50 billion yuan to 80 billion yuan and then to the final figure. Reuters said the round has doubled the original goal. DeepSeek could not be reached for comment, Reuters said, and a second source told the agency that the financial details and the investor lineup could still change.

Tencent, the operator of WeChat, and CATL were among the largest investors, according to Bloomberg and Reuters. Reuters described both as existing shareholders. It said more than 15 investors took part, including the existing backers NetEase and IDG Capital, and the home appliance maker Midea, Boyu Capital and CPE.

New investors named by Reuters include the chipmaker GigaDevice, the food delivery company Meituan and the carmaker Geely. Geely declined to comment, and the other firms did not respond, Reuters said. The Wall Street Journal, as summarised by PYMNTS, also listed Legend Capital, which is backed by Lenovo, and Stony Creek Capital among the participants.

Reuters reported that DeepSeek was valued at about 500 billion yuan before the new capital. The Journal put the figure at about $74 billion. In June, the company raised $7.4 billion in its first external round, Reuters said. Filings by two Chinese investors in July had put the valuation in that June round at 350.88 billion yuan, about $52 billion.

DeepSeek drew global attention in 2025 when it showed that competitive AI models could be built at relatively low cost, a result that triggered a sell-off in technology stocks. Bloomberg said the new money could help it narrow the resource gap with OpenAI and Anthropic, each of which has raised more than $100 billion. DeepSeek still has considerably less financial firepower than either.

The Journal reported, according to PYMNTS, that founder Liang Wenfeng told investors in September that annualised revenue had reached $1 billion, more than double the $400 million to $500 million reported in July. Most of that comes from API access to its models. The Journal said the company has kept gross margins of 70 to 80 percent on its V4 flagship model while charging less than US competitors. Annualised revenue is not a measure of profit, and the company carries heavy infrastructure costs.

The company still relies on Nvidia processors for its most demanding training runs, the Journal reported. It is working with Huawei to move more of that work onto Chinese chips. Last month DeepSeek released DeepSeek-V4.1-Flash, which the company says offers greater capability, faster inference and higher throughput, Reuters said.

The Journal reported that DeepSeek has accelerated plans for a Shanghai listing in early 2027, possibly ahead of OpenAI. It has hired the state-owned Citic Securities as adviser, and recruited the former venture capitalist Yan Wentao as chief financial officer in September. Reuters said it reported last month that the company wanted to begin the IPO process this year and was preparing for a listing on Shanghai's STAR Market. Bloomberg said DeepSeek has stepped up spending on data centres and open-source models ahead of a possible offering.

The deal arrives at an awkward moment for AI share sales. Reuters said investors have grown uneasy about heavy AI spending, which set off a new global technology sell-off this month and led to delayed or withdrawn data centre listings. It added that AI companies have made weak debuts in South Korea and Hong Kong, and that shares of Z.AI and MiniMax, which listed in Hong Kong in January, have fallen sharply from their peaks.

Lian Jye Su, chief analyst at the research firm Omdia, told Reuters it was "encouraging to see such a high valuation for DeepSeek", which he said showed investor confidence in its open-source strategy. He added that because the backers are Chinese companies and critical suppliers in domestic supply chains, DeepSeek could become a key part of China's domestic technology ecosystem.

The next milestone is the formal close of the round, which Reuters said could come within days. DeepSeek's IPO filing would follow, with a Shanghai listing targeted for early 2027, according to the Journal.