TMTPOST — In the global race to dominate artificial intelligence, the timeline for raising capital has contracted from years to weeks. DeepSeek, the Chinese artificial intelligence startup that recently startled the technology industry with its highly efficient models, is already preparing for its next financial phase.

The company has begun preliminary talks with new investors about a fresh funding round. This move comes mere weeks after DeepSeek closed its first-ever round of external financing in June 2026. That initial round raised over $7.4 billion, valuing the company at over $50 billion.

The proposed new round targets a pre-money valuation of at least 480 billion yuan, or roughly $71 billion. This represents a significant valuation jump in just over a month, though the final details of the deal are not yet settled.

At the same time, DeepSeek is looking beyond private venture capital. The startup has begun early preparations for an initial public offering on the Chinese mainland. The company has started talks with accounting and banking advisors, targeting an IPO filing near the end of 2026 or in early 2027. This timeline could set the stage for a public debut in 2027.

The Concrete Demands of Infrastructure

The unusually swift pace of DeepSeek’s fundraising highlights the heavy capital requirements of modern AI development. Building frontier models is no longer just a software challenge. It is an expensive hardware and infrastructure challenge.

DeepSeek plans to use the new capital to significantly expand its physical infrastructure. The company needs to build out its own data centers and purchase more advanced AI chips. In addition to purchasing hardware, DeepSeek is reportedly developing its own proprietary AI chip. This custom silicon initiative aims to reduce the startup's reliance on processors from Nvidia and Huawei.

These expenditures are critical as DeepSeek shifts focus toward autonomous AI agents. These software systems are designed to perform complex tasks without human intervention, but they require massive amounts of continuous computing power to run effectively.

Beyond hardware, the company is facing an intense talent war within China. DeepSeek plans to use its capital to double the size of many of its core teams. The startup recently launched a broad recruitment drive to expand every department, attempting to hire top-tier researchers away from both domestic rivals and global competitors.

An Iron Grip on Governance

DeepSeek's rapid ascent is backed by a unique mix of private wealth, corporate giants, and state resources. Founded in 2023, the startup is owned by the quantitative hedge fund Zhejiang High-Flyer Asset Management.

The firm’s founder, Liang Wenfeng, has structured the company in a highly unusual way. Liang put $3 billion of his own money into the previous $7.4 billion fundraising round, making him the startup's largest investor. Following that transaction, Liang’s net worth more than doubled to $36 billion, according to the Bloomberg Billionaires Index. This positions him as the world's richest creator of AI models, placing him ahead of Western peer founders like Anthropic’s Dario Amodei and OpenAI’s Greg Brockman.

Liang has maintained absolute control over the startup's direction. Most external backers did not invest directly in DeepSeek. Instead, they put their capital into a limited partnership managed by Liang. These investors face a strict five-year lock-up period and receive zero voting rights.

Despite these restrictive terms, prominent Chinese corporations and venture capital firms eagerly joined the syndicate. Backers in the previous round included Tencent, JD.com, NetEase, and battery manufacturer CATL, alongside venture capital firms like IDG. Significantly, the state-backed National Artificial Intelligence Industry Investment Fund also became a minority shareholder. This state backing underscores DeepSeek's position as a central player in China's national effort to compete globally on AI.

Open-Source Research Versus Commercial Pressure

DeepSeek emerged as a prominent player in China’s AI landscape after releasing its open-source R1 reasoning model. The system demonstrated performance comparable to leading Western models but was trained using much more cost-efficient methods. This breakthrough proved that Chinese labs could compete globally despite strict U.S. export restrictions on high-end hardware.

However, the path forward involves balancing research ambitions with commercial realities. DeepSeek’s senior management has informed potential investors that the startup intends to prioritize groundbreaking research over short-term monetization. In investor meetings, Liang Wenfeng pledged to keep developing open-source models while focusing on the long-term goal of achieving artificial general intelligence.

This non-commercial focus may create an interesting tension as the company prepares for the public markets. Working with accounting firms, DeepSeek is aiming to complete its financial reports by the end of December 2026 to prepare for its mainland IPO filing. Ultimately, the timing of the IPO and the success of the new private funding round remain in flux. The final execution of these financial plans will depend heavily on broader market conditions, regulatory approvals, and the startup's ability to maintain its technological edge in a crowded field.