DeepSeek Leads Multi‑Tens‑Million Dollar Seed Round for Winverse AI, Boosts Embodied AI Strategy

DeepSeek, the Chinese artificial‑intelligence powerhouse best known for its large‑language models, is accelerating its foray into embodied AI. On 12 October 2026 the firm announced that it will act as lead investor in a seed‑stage financing round for Winverse AI, a robotics‑focused startup founded earlier this year by former Tencent senior scientist Hu Han.

The financing is described as “several tens of millions of dollars,” a sum that lifts Winverse AI’s post‑money valuation to roughly $400 million, according to the report published by finance.biggo.com.

Background on Winverse AI and Its Founder

Winverse AI was created in 2026 after Hu Han departed Tencent in July 2026 to devote his efforts to embodied intelligence. Hu’s deep experience in large‑scale AI research is regarded as a cornerstone for the nascent company, which aims to fuse cutting‑edge perception models with highly agile robotic platforms.

DeepSeek’s involvement goes well beyond the provision of capital. Liang Wenfeng, a senior executive at DeepSeek, has publicly outlined two strategic pathways for the firm’s embodied‑AI ambitions: either building an internal team from scratch or allocating resources to external startups such as Winverse. The seed round signals a clear preference for the latter route, at least in the near term.

DeepSeek’s Existing Stake in Unitree Robotics

DeepSeek already occupies a strategic niche within China’s robotics ecosystem through its investment in Unitree Robotics. The company contributed approximately 141 million CNY (about $21.1 million) to Unitree’s initial public offering in Shanghai in August 2026, thereby acquiring close to a 2 % equity stake.

Unitree’s IPO priced shares at 150.80 yuan each, raising roughly 6.1 billion CNY. A procurement pact between the two firms grants Unitree priority access to DeepSeek’s model‑training services, while DeepSeek receives preferential treatment when purchasing Unitree’s robots.

Financial Context and Fundraising Momentum

Bloomberg estimates that DeepSeek is on the cusp of raising at least 80 billion CNY (approximately $11.9 million) in a new funding round, surpassing its original target of 50 billion CNY. This influx of capital is expected to fuel both internal research and development as well as external investments such as the Winverse seed round.

The company’s aggressive hiring strategy underpins this financial expansion. DeepSeek announced a doubling of staff across all divisions and the formation of a new “Harness” team, led by former Jane Street quant Cui Tianyi, to accelerate the integration of language models with physical systems.

• Seed round of several tens of millions of dollars for Winverse AI
• Post‑money valuation of Winverse AI at ~ $400 million
• ~141 million CNY (~$21.1 M) invested in Unitree’s IPO for ~2 % stake
• Procurement pact giving Unitree priority for DeepSeek’s model‑training services

Performance benchmarks illustrate the competitive landscape. OpenAI’s GPT‑6 achieved a 95 % success rate on a standard robotic control task, while DeepSeek’s reference model Fable 5.1 managed only 40 % on the same benchmark, according to the same Bloomberg analysis.

Hangzhou, often dubbed the “Silicon Valley of China” for AI and robotics, hosts DeepSeek, Unitree, and a host of other innovators. The city’s vibrant ecosystem provides fertile ground for collaborations that blend cutting‑edge AI research with hardware development.

For English‑speaking organizations, the combined effect of DeepSeek’s funding and partnership strategy translates into faster access to high‑performance embodied AI solutions. Companies looking to deploy robots for logistics, manufacturing, or service applications can now tap into a supply chain where model training, robot hardware, and integration expertise are tightly aligned under a single strategic umbrella.

Implications for the Global Robotics Market

The infusion of tens of millions of dollars into Winverse AI positions the startup to accelerate prototype development, attract top talent, and secure early pilot projects with Chinese manufacturers. This could narrow the gap between Chinese and Western embodied‑AI players, especially as DeepSeek leverages its existing relationships with Unitree.

Analysts note that DeepSeek’s dual‑track approach—building internal capabilities while simultaneously nurturing external ventures—mirrors strategies employed by leading AI conglomerates in the United States and Europe. The model allows the firm to experiment with diverse architectures without bearing the full risk of in‑house development.

Moreover, the procurement pact with Unitree ensures that any breakthroughs in model training can be rapidly transferred to physical robots, shortening the time‑to‑market for new applications such as warehouse automation and autonomous delivery.

In the short term, Winverse AI is expected to focus on integrating DeepSeek’s perception models with Unitree’s quadruped platforms, a combination that could showcase the practical benefits of embodied AI to potential enterprise customers.

Localisation : Cette annonce a été publiée à Hangzhou le 13 octobre 2026, soulignant l’importance croissante de la région dans le paysage mondial de la robotique et de l’intelligence artificielle.