DeepSeek is pairing a round of about RMB100 billion with a plan to at least double every department, turning a lab known for its small team into a large employer.
DeepSeek is expanding its workforce quickly and has laid out a new growth plan for investors, according to a person close to the company quoted by Ifeng Tech on October 27. The report, carried in full by 21st Century Business Herald, said the plan DeepSeek is presenting to investors includes a major expansion of its team. The Hangzhou-based developer, whose offices are mainly in Beijing and Hangzhou, has not commented publicly on the report.
The report arrives as DeepSeek finishes its second funding round in five months. Reuters reported on October 22 that DeepSeek had raised about RMB100 billion, roughly US$15 billion, at a pre-money valuation of RMB500 billion, with CATL and Tencent as the largest investors.
The Ifeng report gives no headcount, now or planned, and no detail on products or revenue targets in the growth plan. What it does show is direction. A lab that became famous in early 2025 for building strong models with a small team is now hiring across research, computing infrastructure, products and corporate functions at the same time.
The hiring push has been building since June
DeepSeek set the pace in a recruitment notice on June 25, when it said it would at least double the size of every department and opened 33 positions. The roles covered technical research, computing infrastructure, putting products into use, frontier algorithm research and general functions, and included data-center construction jobs in Ulanqab in Inner Mongolia, Ifeng Tech reported. DeepSeek had begun hiring to build its own data-center infrastructure by August.
The June notice also framed the company’s ambitions in stark terms. "Humanity is on the eve of AGI," DeepSeek said, according to Yicai Global’s June report. Full-stack development and algorithm roles had the most openings, eight in total, including positions on the Agent Harness team, which builds coding agents that DeepSeek benchmarks against Anthropic’s Claude Code.
The scale grew in September. Cui Tianyi, who leads DeepSeek’s Harness team, posted about 150 openings in September and called the hiring need unprecedented. Cui joined DeepSeek in March, and the team aims to turn the company’s models into AI agents, Yicai Global said in a September report. The roles were mainly for backend developers and engineers working on elastic computing for agents, with two to ten years of experience, based mostly in Beijing.
Funding and product traction give the expansion its budget
The money to pay for the hiring is arriving in two large rounds. DeepSeek raised US$7.4 billion in its first external round in June, Reuters reported, and the new round of about RMB100 billion doubles its original target of RMB50 billion. Reuters said more than 15 investors were taking part, including GigaDevice, Meituan, Geely, Midea, NetEase, Boyu Capital, CPE, Legend Capital and IDG Capital, and that terms could still change.
Chinese media coverage put the planned size at RMB50 billion at first, then RMB80 billion, then about RMB100 billion, which would take DeepSeek’s total funding this year to roughly RMB150 billion, according to Ifeng Tech. The company has not disclosed the final amount or the investor list.
The products are giving the hiring a commercial rationale. The Ifeng report pointed to DeepSeek’s decision to open-source DeepInfer, its internal inference engine, which runs on both Nvidia and Huawei Ascend chips, and to OpenRouter data showing DeepSeek models at the top of the platform’s weekly usage ranking for three weeks running. TechCrunch noted that the fundraising came two days after the US ordered federal civilian agencies to remove DeepSeek’s technology.
A bigger DeepSeek will face the problems of a bigger company
Rapid growth tests a lab whose culture was built around a few hundred researchers. When DeepSeek released its R1 model, founder Liang Wenfeng said the company had a team of about 160 people, Yicai Global reported. Cui said in his September post that DeepSeek’s existing backend systems might not meet future demand as the company scales, a sign that the hiring is partly about rebuilding infrastructure rather than adding researchers.
The Ifeng report also described Liang as living in simply furnished rented housing in Beijing, a detail that fits the company’s low-profile image but says little about how it will manage several times its current staff.
The next signals are the formal close of the funding round, with an investor list and final amount, and any headcount figure DeepSeek discloses. Hiring announcements from the Ulanqab data-center team would show how much of the new money goes into owned computing capacity rather than models and agents.